33

Rainfall Redistribution by the Canopy of a Juvenile Lodgepole Pine
Stand
by
Chad Edward Lishman
B.A. Thompson Rivers University, 2009
P.Ag., British Columbia Institute of Agrology, 2013

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE (ENVIRONMENTAL SCIENCE)
Thompson Rivers University
Kamloops, British Columbia, Canada
September, 2015

Thesis examining committee:
Darryl Carlyle-Moses (PhD), Thesis Supervisor and Associate Professor and Chair,
Department of Geography & Environmental Studies, Thompson Rivers University
David Hill (PhD), Associate Professor and Committee Member, Department of Geography &
Environmental Studies, Thompson Rivers University
Thomas G Pypker (PhD), Assistant Professor and Committee Member, Department of
Natural Resource Science, Thompson Rivers University
Rita Winkler (PhD, R.P.F.), Research Hydrologist, British Columbia Ministry of Forests,
Lands and Natural Resource Operations
Delphis Levia (PhD), External Reader, Professor of Ecohydrology and Director,
Environmental Science and Environmental Studies, Department of Geography,
University of Delaware

© Chad Edward Lishman, 2015

ii
Thesis Supervisor: Dr. Darryl Carlyle-Moses
ABSTRACT
From May to October 2010, throughfall associated with 38 rainfall events in three belowcanopy zones (inner-canopy, mid-canopy, and canopy-periphery) of nine juvenile lodgepole
pine trees and in open areas between canopies was measured along north, south, east and
west transects radiating from each tree bole. Median cumulative throughfall (%) significantly
differed among the canopy zones under differing rain depth classes and was negatively
correlated with tree size metrics, but only for certain combinations of canopy zones and rain
depth classes. Although median cumulative throughfall (%) was negatively correlated with
canopy cover fraction when all below-canopy gauges were considered, this relationship only
held true for the inner- and mid-canopy zones under relatively small rain depth classes.
Additionally, cumulative throughfall (%) was often dependent on transect direction, with
event throughfall showing a dependence on the direction of storm origin. Temporal
persistence of throughfall was assessed using time stability plots and Spearman rank
coefficients of rain event and depth class pairings for all gauges and by canopy zone. The
influence of temporal lag and meteorological variables on persistence was also assessed. The
implications of our findings on throughfall sampling and for future ecohydrological studies in
juvenile lodgepole pine and similar stands are discussed. Additionally, stemflow was
measured so that a full canopy water balance could be derived for each event and over the
study period. Canopy interception loss, throughfall and stemflow accounted for 87.7, 10.5,
and 1.8% of the study-period rainfall, respectfully. Both the reformulated Gash and
reformulated Liu models were found to have satisfactory simulated study-period interception
loss; however, estimates at the rain-event scale were often poor. Interception loss from the
juvenile stand is largely a factor of canopy cover fraction and relatively high during-rainfall
evaporation rates.

iii
TABLE OF CONTENTS
ABSTRACT…………………………………………………………………………………..ii
TABLE OF CONTENTS…………………………………………………………………….iii
LIST OF TABLES……………………………………………………………………………v
ACKNOWLEDGEMENTS………………………………………………………………..…ix
CHAPTER 1: INTRODUCTION…………..………...……………………………………….1
REFERENCES .................................................................................................................... 6
CHAPTER 2: THROUGHFALL SPATIAL VARIABILITY AND TEMPORAL
PERSISTANCE BELOW AND BETWEEN THE CANOPIES OF JUVENILE
LODGEPOLE PINE (Pinus contorta)…………………………………………………….....10
INTRODUCTION…...………………………...…………………………………………10
Study area ....................................................................................................................... 13
Rainfall and throughfall measurement ........................................................................... 14
Canopy cover .................................................................................................................. 15
Temporal persistence of throughfall ............................................................................... 15
Data analysis .................................................................................................................. 17
RESULTS .......................................................................................................................... 17
Incident precipitation ...................................................................................................... 17
Absolute and relative throughfall depth ......................................................................... 17
Throughfall spatial variability ........................................................................................ 19
Throughfall, tree size and canopy cover ......................................................................... 22
Throughfall and transect direction ................................................................................. 22
Temporal stability of throughfall .................................................................................... 24
DISCUSSION…… ............................................................................................................ 30
CONCLUSION .................................................................................................................. 36
REFERENCES .................................................................................................................. 37
CHAPTER 3: ANALYTICAL MODELLING OF CANOPY INTERCEPTION LOSS
FROM A JUVENILE LODGEPOLE PINE (Pinus contorta var. latifolia ) STAND……….44
INTRODUCTION ............................................................................................................. 44
MATERIALS AND METHODS ....................................................................................... 46
Research Site ................................................................................................................... 46
Rainfall, Throughfall and Stemflow Measurement……………………………………..49

iv
Modeling Procedure ....................................................................................................... 52
Statistical Analysis and Model Evaluation…………………………………………………...55
RESULTS............................................................................................................................ 56
Plot-Scale Canopy Water Balance.................................................................................. 56
Modeling Results ............................................................................................................. 58
DISCUSSION AND CONCLUSION ............................................................................... 60
REFERENCES .................................................................................................................. 63
CHAPTER 4: CONCLUSION………………………………………………………………67
LIMITATIONS .................................................................................................................. 68
APPLICATION OF RESEARCH ..................................................................................... 69
FUTURE STUDY DIRECTIONS ..................................................................................... 69
REFERENCES .................................................................................................................. 71

v
LIST OF TABLES
Table 2.1: Characteristics of the individual lodgepole pine trees used in this study.
Correlations among the tree size metrics were all significant (p ≤ 0.05): Height – Basal Area
(r = 0.908; rs = 0.933); Height – Projected Crown Area (r = 0.909; rs = 0.933), Basal Area –
Projected Crown Area (r = 0.999, rs = 1.0)…………………………………………………15
Table 2.2: Event number, date(s) of occurrence, rainfall depth (P, mm), event duration (DE,
h), rainfall duration (DR, h), mean rainfall intensity(IMean, mm h-1), maximum 30 minute
rainfall intensity, IMax-30, mm h-1), number of intra-storm breaks ≥ 1 h, (NB), total duration of
intra-storm breaks ≥ 1 h (DB, h) and the principal cardinal direction of storm origin
(DoSO)………………………………………………………………………………………18
Table 2.3: Spearman rank correlation coefficients (n = 38) between the CV of event TF%
and the storm characteristics of rainfall depth, P (mm), event duration, DE (h), mean rainfall
intensity IMean (mm h-1), maximum 30-minute rainfall intensity, IMax-30 (mm h-1), number of
storm breaks ≥ 30 minutes, NB, and total duration of those breaks, DB (h). Level of
significance: *p ≤ 0.05; ** p ≤ 0.01………………………………………………………….22
Table 2.4: Spearman rank correlation coefficients (n = 9) between tree size characteristics
and median cumulative TF% in each of the four sampling zones for four rain depth classes
and for the entire season. Level of significance: *p ≤ 0.05, **p ≤ 0.01. Note: Basal area and
PCA are listed together since the ranks of each of the tree basal areas of the nine trees match
the ranks of each of the PCA values (i.e., rs = 1.0)…………………………………………23
Table 2.5: Median cumulative TF% as a function of canopy position and transect direction
(n = 9 for each case). Values in bold indicate medians that significantly differ (median test:
*
p ≤ 0.10, **p ≤ 0.05, ***p ≤ 0.01) from one another with the transect direction of the differing
value superscripted (N = north, S = South, E = east, and W = west)……………………….24
Table 2.6: Spearman (right) and Pearson (left) correlation coefficients between cumulative
TF% of four different rain-depth classes ( < 2, 2 to < 5, 5 to < 10, and ≥ 10 mm) for the four
sampling zones (inner-canopy, mid-canopy, canopy-periphery, and outside-of-canopy) (n =
36, *p ≤ 0.05, **p ≤ 0.01; n.s. = not significant)…………………………………………….26
Table 2.7: Coefficient values associated with meteorological factors influencing Spearman
rank coefficient values of paired events and the intercept, and R2 for each model for all
gauges pooled and under each of the four canopy zones. (n = 703, *p ≤ 0.05, **p ≤ 0.01; ***p
≤ 0.001)……………………………………………………………………………………....32
Table 3.1: Components of canopy interception loss, throughfall and stemflow in the sparse
Gash model. Adapted from Valente et al. (1997)…………………………………………...54
Table 3.2: Percentage of rainfall partitioned into throughfall, stemflow and canopy
interception loss for four rainfall depth classes……………………………………………..58

vi
Table 3.3: Gash and Liu model parameter values derived for the study stand……………..58
LIST OF FIGURES
Figure 2.1: Geographic Location of the study plot. Adapted from Winkler et al. (2010)…..13
Figure 2.2: Box-plots of cumulative TF% for different rain depth classes and for the season
for the different sampling zones: a) inner-canopy, b) mid-canopy, c) canopy-periphery, and
d) outside-of-canopy…………………………………………………………………………20
Figure 2.3: Accumulated (a) and event (b) TF CV% as a function of rain depth for the innercanopy (closed circles), mid-canopy (open circles), canopy-periphery (closed squares) and
outside-of-canopy (open squares)……………………………………………………………21
Figure 2.4: Ratio of event storm-ward to lee ward TF throughout the study period for the a)
inner-canopy, b) mid-canopy, c) canopy-periphery, and d) outside-of-canopy sampling
zones…………………………………………………………………………………………25
Figure 2.5: Temporal stability plot of normalized throughfall (TFSi) for all 144 gauges.
Individual dots represent event TFSi values, while the solid line passes through the mean
TFSi value for all gauges ranked from lowest to highest mean TFSi………………………..28
Figure 2.6: Temporal stability plots for the 36 gauges in each of the a) inner-canopy, b) midcanopy, c) canopy-periphery, and d) outside-of-canopy sampling zones with median set to
zero and unit variance (median absolute difference, MAD)…………………………………29
Figure 3.1: Geographic location of study plot………………………………………………47
Figure 3.2: Aerial view of the study plot. Photo Credit: R.D. Winkler…………………….48
Figure 3.3: Climograph of study area based on ClimateBC ver 4.72 results using study plot
latitude, longitude and elevation as inputs. Note: White bars = snow, black bars = rain…..48
Figure 3.4: The Onset® tipping bucket rain gauge (left) and Meteorological Service of
Canada Type-B rain gauge (right) used in this study………………………………………..49
Figure 3.5: Throughfall gauge layout schematic. Shaded green = tree canopy, brown circle =
tree bole, and blue squares = throughfall gauges……………………………………………50
Figure 3.6: Stemflow and throughfall collection system for one of the selected study
trees…………………………………………………………………………………………..52
Figure 3.7: a) Event throughfall and b) event stemflow (mm) ─●─ and interception loss
(mm) ---○--- versus rainfall (mm)……………………………………………………………57

vii
Figure 3.8: Liu modeled interception loss (mm) versus Gash modeled interception loss
(mm)…………………………………………………………………………………………59
Figure 3.9: Modeled accumulative interception loss (mm) ---- and observed interception loss
(mm) ____ as a function of accumulative rainfall (mm)………………………………………59
Figure 3.10: Sensitivity of the reformulated Gash (a) and Liu (b) models to changes in model
parameter values……………………………………………………………………………..61
Figure 4.1: Conceptual model comparing a juvenile and mature lodgepole pine tree rainfall
partitioning…………………………………………………………………………………..70

viii
LIST OF SYMBOLS
A – Total Area beneath the tree canopy
a – Slope
az – Area of the below canopy zone
BA – Basal area
BC – British Columbia
CPA – Crown projection area
DOSO – Direction of Storm Orgin
F – Funnelling ratio
Ha – Hectare
Ic – Interception Loss
MPB – Mountain Pine Beetle
PA – Annual rainfall
Pd – Drainage partitioning coefficient
PCA – Projected Crown Area
– Incident precipitation falling on the canopy
– Canopy storage capacity
– Trunk Storage Capacity
R – Incident precipitation
S – Canopy Storage Capacity
SF – Stemflow
St – Trunk Storage Capacity
TF – Throughfall
TFT – Total below canopy throughfall
– Mean throughfall catch within the below canopy zone

ix
ACKNOWLEDGEMENTS

I owe my deepest gratitude to my supervisor, Dr. Darryl Carlyle-Moses, for
encouraging me to join the Masters of Science program at TRU and for his continuous
support, participation, expertise and guidance that made the completion of this thesis
conceivable. Thank you for my thesis committee, Dr. Rita Winkler, Dr. David Hill, and Dr.
Delphis Levia, for their patience, time and direction. Thank you for my research crew,
Pearce Sanders, Andrew Pillar, Jenn Golden, and Adam McKee, for their extraordinary
effort, countless hours and support.
This thesis would not have been possible without the funding support awarded to Dr.
Darryl Carlyle-Moses, provided by the NSERC (Natural Sciences and Engineering Research
Council of Canada) Discovery Grant and the FIA-FSP (Forest Investment Account – Forest
Science Program).
I would like to thank my family and friends for their support over the past few years
that have helped push me to complete this thesis, and especially to Jenna Lishman who
continues to push me forward.

33

CHAPTER 1
INTRODUCTION
Lodgepole pine (Pinus contorta var. latifolia Dougl) forests of the central interior
region of British Columbia are habitats which are under constant threat from extensive forest
fires, mountain pine beetle (MPB) outbreaks, logging, grazing and range land encroachment
(B.C. Ministry of Forests, Mines and Lands. 2010). This area plays a crucial role in being
the headwaters for many of the regions creeks and rivers. With the recent MPB outbreak,
10.1 million ha of lodgepole pine forests were affected, which resulted in increased
harvesting, salvaging and planting in the region (Axelson et al., 2009). As a result, juvenile
lodgepole pine stands are becoming increasingly dominant in the region, replacing the
harvested, or deceased mature stands. Numerous studies have been conducted in British
Columbia and around the world regarding rainfall interception; however they have been
primarily focused on mature forests (Spittlehouse, 1998). Thus, for a better understanding of
the hydrological effects of the transition from a mature forest to a juvenile forest dominated
landscape, rainfall interception and the interaction between near-surface soil moisture and the
atmospheric boundary layer must be studied.
Spatial delivery of precipitation from the tree canopy to the soil occurs through two
processes, throughfall and stemflow. Throughfall and stemflow are usually calculated as the
difference between incident gross precipitation and interception. In temperate forests,
stemflow and throughfall typically account for 70-90% of the incident gross precipitation
(Levia and Frost, 2003). The bulk of understory precipitation, however, reaches the forest
floor in the form of throughfall. At the stand scale, throughfall commonly represents > 90%
of the understory precipitation input (Brabender, 2005). The spatial variability of throughfall
is not inconsequential. Price et al. (1997), for example, found that the spatial variability of
throughfall, expressed as a coefficient of variation (CV), in a black spruce (Picea mariana)
stand decreased asymptotically with rainstorm size, averaging 60% of rainfall for ‘small’
events and 24% for ‘large’ rainfalls (> 10 mm). Although the spatial variability of
throughfall is a consequence of rainfall interacting with the canopy, the interactions that
occur and the resulting throughfall spatial pattern, or lack thereof, appear to be ecosystem
dependent (e.g., Loustau et al. 1992; Carlyle-Moses et al., 2004; Nadkarni and Sumera,
2004). Determining if throughfall input is random or if it is systematically related to stand
1

2
characteristics (e.g., distance from the tree bole) is an important consideration for the proper
estimation of throughfall input to the forest floor and for soil moisture variability.
Some researchers have found that there is no spatial correlation in forest throughfall
(Laostau et al. 1992; Bellot and Escarre, 1998), while others have found evidence of spatial
autocorrelation (Loescher et al. 2002; Gomez et al. 2002). Keim et al. (2005) concluded that
throughfall spatial correlation was highly variable and differed depending on type of stand
and season, which yielded results of no correlation in some stands to correlation lengths from
3-10 m in others.
In a mixed deciduous forest in Belgium, Staelens et al. (2006) conducted a spatial and
temporal throughfall variability study to determine if there were any spatial patterns. They
concluded that during the leafed periods, throughfall was spatially correlated up to a distance
of 3-4 m. In addition, they determined that there was considerable temporal stability of the
spatial throughfall pattern in the growing season and the dormant period. Furthermore,
throughfall water during the growing season drastically decreased with increasing canopy
cover, and was determined to be closely correlated with branch cover; however, the spatial
pattern of throughfall in the dormant period was not related to the branch cover.
The distance over which throughfall is correlated is determined by stand
characteristics such as tree species, stand density, and canopy structure (Keim et al., 2005).
However, it is difficult to compare the spatial correlation lengths and patterns (or lack
thereof) of throughfall between one stand and another, due, in large part, to the multitude of
factors that affect throughfall, such as tree geometry, basal area, tree density, age, understory
abundance, crown cover, season, and numerous other variables (Huber and Iroumé, 2001).
When compared to throughfall, stemflow inputs to the forest floor have a greater
spatial variability and are more concentrated (Levia, and Frost, 2003). With the high spatial
variability of stemflow, Tanka et al. (1996) proposed that the actual amount of stemflow
infiltrating in a circular pattern around the tree trunk may represent as much as 10-20% of the
incident gross precipitation, which results in soil moisture and ground water recharge. As
mentioned, the spatial variability of stemflow is large, controlled in part, by wind, rain
intensity and duration, crown projection area, tree density, age, branch inclination and
species type (Levia and Frost 2003). Another factor influencing stemflow is angled rainfall
(Van Stan et al., 2011). Due to wind, most rain events do not fall vertically but on an angle.

3
This creates uneven interception due to taller trees causing a partial or full rainshadow on
neighbouring trees, which causes uneven stemflow production, with the taller trees producing
a greater volume than the rain shadowed trees (Levia and Frost, 2003). In general the spatial
variability of stemflow generation decreases as the magnitude of a precipitation event
increases (Levia and Frost, 2003).
Stemflow typically accounts for 5 to 10% of the gross precipitation in most forest
environments. (Levia and Frost, 2003; Li et al., 2008; Tanaka et al.,1996; Taniguchi et al.,
1996). However, in mature coniferous forests stemflow is typically a negligible (< 1% of
precipitation) component of forest water balance (Price et al., 1997; Spittlehouse, 1998).
Since stemflow accounts for only a few percent of the gross precipitation, many studies have
ignored its hydrological significance. However, the volume of water delivered to the base of
trees may be very large, far exceeding rainfall and throughfall inputs on a per unit area basis
(Murakami, 2009). Herwitz (1986) introduced a metric for quantitatively describing the
importance of stemflow as a point input. This funneling ratio is found as:

where F is the funneling ratio (dimensionless), SFvol. is the stemflow volume (litres), Pg is
gross precipitation and BA represents the tree basal area (m2).
The stemflow funneling ratio represents the amount of water delivered to the base of
the tree compared to that which would be captured by an unobstructed rain gauge having an
orifice area equal to the tree BA. Thus, F values > 1 indicate that portions of the tree canopy
outside of the tree bole are contributing to stemflow.
Most published F values are from either semi-arid or tropical rainforest
environments. Funneling ratios from these studies have been found to be highly variable,
ranging from 2 for an individual mesquite (Prosopis laevigata) tree in northeastern Mexico
(Navar and Bryan, 1990) to 276 for an individual Sierra palm (Prestoea montana) in Puerto
Rico (Holwerda et al., 2006, as calculated by McKee, 2010). The only study to be conducted
in a temperate deciduous forest during growing season conditions (Carlyle-Moses and Price,
2006) found that geometric mean F values were 7.3, 20.6 and 26.3 for red oak (Quercus

4
rubra), sugar maple (Acer saccharum) and American beech (Fagus grandifolia),
respectively. No known studies have reported F values for mature coniferous forests;
however, McKee and Carlyle-Moses (2010) found that funneling ratios for juvenile
lodgepole pine during a single event had a maximum F of 111.7 during a 17.4 mm rainfall.
Tanaka et al. (1996) observed that in a red pine (Pinus densiflora) forest in
Japan,stemflow inputs were concentrated over localized circular areas at the tree base, and
showed strong vertical movement in the soil profile. They found that all stemflow infiltrates
within a 50 cm radius around large trees (circumferences > 40 cm) and approximately 30 cm
radius around smaller trees (circumferences < 20 cm). The amount of water concentrated was
so large that it represented an important groundwater recharge mechanism in this pine forest.
Vincke and Thiry (2008) have shown that water tables near the soil surface not only provided
a key source of water for trees and other vegetation, but often show diurnal fluctuations. In
the day when evapotranspiration is occurring, the water table declines, whereas at night the
water table rebounds due to the absence of evapotranspiration. Furthermore, stemflow
production by beech (Fagus sylvatica) was linked to the fast response portion of a storm
hydrograph from a hillslope in eastern England (Crabtree and Trudgill, 1985).This has been
shown to contribute vast amounts of concentrated nutrients at the base of trees creating
‘fertile islands’ (Whiteford et al., 1997). Thus, although a minor component of the forest
water balance, stemflow, because of its concentrated demeanor, may be an important
hydrologic mechanism for other components of the forest water cycle.
Macropores are one explanation of how stemflow precipitation can reach the water
table. With macropore flow, soil water can flow faster and deeper by channeling infiltrated
water along roots through preferential flow pathways that are formed by localized
compaction of soil by roots. In the active rhizosphere along living roots and old root
channels, these pathways influence water and nutrient distribution (Johnson and Lehmann,
2006). On forested hill slopes, localized concentrated stemflow inputs at the tree base
bypassed the soil matrix through macropores leading to rapid inputs to a nearby stream (Levi
and Frost, 2003). Li et al., (2009) have found that arid and semi-arid shrubs divert
precipitation as stemflow where it infiltrates the soil and is stored in the deeper soil layers
through stemflow-root channelization, for later when plant uptake is needed. This is thought
to be an adaption to survive drought periods.

5
Canopy interception loss is when a proportion of rain falls is intercepted by the
canopy and evaporates back into the atmosphere. Canopy interception loss , which often
constitutes a ample portion of the total evaporation flux to the atmosphere from forests
(Carlyle-Moses and Gash, 2011), is almost always derived by taking the difference between
incident rainfall and the sum of TF and SF (e.g., Carlyle-Moses et al., 2010; Návar, 2013;
Saito et al., 2013).
Throughout the world interception studies have been conducted in various climates
and ecosystems producing interception results ranging from 10 to 50 % of annual or seasonlong rainfall (Roth et al. 2007). Canopy interception loss, from mature lodgepole pine forests
or stands that comprise lodgepole pine with other species is appreciable, accounting for 24 –
29% of season-long rainfall (Spittlehouse, 1998; Moore et al., 2008). Interception is affected
by numerous biological factors, including tree density, number of branches, species, height,
basal area, crown projection area, and branch angel, as well as climatic factors such as
intensity, amount and duration of rainfall. In addition wind speeds, direction, air
temperature and humidity all effects interception pre and post event (Crockford and
Richardson, 2000).
There have been several key models that have been developed for estimating Ic,
ranging from analytical models such as the Gash model (Gash, 1979) to linear regression (
Helvy and Patric, 1965), physical-based numerical (Rutter et al, 1971, 1975) and stochastic
models (Calder, 1986).
Chapter 2 examines a juvenile lodgepole pine (Pinus contorta Douglas ex Louden
var. latifolia Engelm. ex S. Watson) stand in south-central British Columbia, in order to
determine the magnitude of point and structural throughfall as well as stemflow at the rainfall
event and growing-season temporal scales. In addition, this chapter evaluates if throughfall
(both point and structural) and stemflow spatial variability exhibits temporal persistence and
if so attempt to determine what the influence, if any, tree and storm characteristics may have
on this stability variability. Finally Chapter 2 provides recommendations for sampling
strategies for future research in similar forested environments.
Chapter 3 deals with interception modelling based on original field observations
conducted in the same previously mentioned lodgepole pine stand on the Bonaparte Plateau,
north of Kamloops, British Columbia, Canada. The goal of this chapter was to evaluate the

6
performance of the reformulated Gash (Valente et al., 1997) and reformulated Liu (CarlyleMoses and Price, 2007) Ic models at both event and season-long time scales, as well as to
determine the quantitative importance of throughfall, stemflow and Ic. Furthermore, this
chapter assesses if the data collected supports a ‘water-box’ or an ‘exponential wetting
approach’ of the canopy saturation process.

REFERENCES
Axelson, J., Alfaro, R., Hawkes, B. 2009. Influence of fire and mountain pine beetle on the
dynamics of lodgepole pine stands in British Columbia, Canada. Forest Ecology and
Management 257: 1874–1882.
B.C. Ministry of Forests, Mines and Lands. 2010. The State of British Columbia’s Forests,
3rd ed. Forest Practices and Investment Branch, Victoria, B.C.
www.for.gov.bc.ca/hfp/sof/index.htm#2010_report
Bellot, J., Escarre, A. 1998. Stemflow and throughfall determination in a resprouted
Mediterranean holm-oak forest, Annales des Sciences Forestieres 55: 847–865.
Brabender, B. 2005. Scaling leaf area index and Rainfall interception in Lodgepole pine.
MSc Thesis, University of Alberta.
Calder, I.R. 1986. A stochastic model of rainfall interception. Journal of Hydrology: 89: 6571.
Carlyle-Moses DE, Park AD, Cameron JL. 2010. Modelling rainfall interception loss in
forest restoration trials in Panama. Ecohydrology 3: 272-283
Carlyle-Moses DE, Price AG. 2006. Growing-season stemflow production within a
deciduous forest of southern Ontario. Hydrological Processes 20: 3651-3663.
Carlyle-Moses DE, Price AG. 2007. Modelling canopy interception loss from a Madrean
pine-oak stand, Northeastern Mexico. Hydrological Processes 21: 2572-2580
Carlyle-Moses DE, Gash JHC. 2011. Rainfall Interception Loss by Forest Canopies. In:
Forest Hydrology and Biogeochemistry: Synthesis of Past Research and Future
Directions, Levia DF, Carlyle-Moses DE, Tanaka T (eds.). Ecological Series 216,
Springer-Verlag: Heidelberg, Germany; 407-423.
Carlyle-Moses DE, Flores Laureano J.S., Price A.G. 2004. Throughfall and throughfall
spatial variability in Madrean oak forest communities of northeastern Mexico. Journal
of Hydrology 297:124-135.

7
Crabtree, A.F., Trudgill, S.T., 1985. Hillslope hydrochemistry and stream response on a
wooded, permeable bedrock: the role of stemflow. Journal of Hydrology 80: 161–178.
Crockford, R. H., Richardson, D. P.2000, Partitioning of rainfall into throughfall, stemflow
and interception: effect of forest type, ground cover and climate. Hydrological
Processes 14: 2903–2920.
Gash JHC. 1979. An analytical model of rainfall interception in forests. Quarterly Journal of
the Royal Meteorological Society 105: 43-55
Gomez, J.A., Vanderlinden, K., Gira´ldez, J.V., Fereres, E., 2002. Rainfall concentration
under olive trees. Agricultural Water Management 55: 53–70.
Helvey, J. D., Patric J. H.1965. Canopy and litter interception of rainfall by hardwoods of
eastern United States, Water Resour. Res., 1(2)
Herwitz SR. 1986. Infiltration-excess caused by Stemflow in a cyclone-prone tropical
rainforest. Earth Surface Processes and Landforms 11:401-412.
Holwerda F., Scatena F.N, Bruijnzeel L.A. 2006. Throughfall in a Puerto Rican lower
montane rain forest: A comparison of sampling strategies. Journal of Hydrology 327:
592-602.
Huber A., Iroumé A .2001. Variability of annual rainfall partitioning for different sites and
forest covers in Chile. Journal of Hydrology 248:78–92
Hupet F., Vanclooster, M., 2001. Effect of the sampling frequency of meteorological
variables on the estimation of the reference evapotranspiration. Journal of Hydrology
243: 192–204
Johnson MS., Lehmann J. 2006. Double-funnelling of trees: Stemflow and root-induced
preferential flow. Ecoscience 13: 324-333.
Keim, R.F., Skaugset, A.E., Weiler, M., 2005. Temporal persistence of spatial patterns in
throughfall. Journal of Hydrology 314: 263–274.
Levia DF Jr., Frost EE. 2003. A review and evaluation of stemflow literature in the
hydrologic and biogeochemical cycles of forest and agricultural ecosystems. Journal of
Hydrology 274: 1-29.
Li, X, Liu, L., Gao, S., Ma., Y., Yan, Z. 2008. Stemflow in three shrubs and its effect on soil
water enhancement in semiarid loess region of China. Agricultural and Forest
Meteorology 148: 1501-1507.

8
Li X., Yang, Z., Lim Y, Lin H. 2009. Connecting ecohydrology and hydropedology in desert
shrubs: stemflow as a source of preferential flow in soils. Hydrology and Earth System
Science 13: 113-1144.
Loescher, H.W., Powers, J.S., Oberbauer, S.F., 2002. Spatial variation of throughfall volume
in an old-growth tropical wet forest, Costa Rica. Journal of Tropical Ecology 18: 397–
407
Lousteau, D., Berbigier, P., Granier, A. 1992. Interception loss, throughfall and stemflow in a
maritime pine stand, II. An application of Gash’s analytical model of interception.
Journal of Hydrology 138: 469–485.
McKee AJ 2010. The quantitative importance of stemflow: An evaluation of past research
and
results from a study in lodgepole pine (Pinus contorata var. latifolia) stands in southern
British Columbia. M.Sc. in Environmental Science thesis, Thompson Rivers
University.
McKee AJ., Carlyle-Moses DE. 2010. Stemflow: A potentially important point source of
water for growth. Linking Innovations and Networking Knowledge 11:11-12.
Moore, RD, Winkler, RD, Carlyle-Moses, DE (2008) Watershed response to the McLure
forest fire: presentation summaries from the Fishtrap Creek workshop, March 2008.
Streamline Watershed Manage Bulletin 12: 1-11
Nadkarni N., and Sumera M. 2004. Old-Growth Forest Canopy Structure and Its Relationship
to Throughfall Interception. Forest Science 50: 290-298
Návar J. 2013. The performance of the reformulated Gash’s interception loss model in
Mexico’s northeastern temperate forests. Hydrological Processes 27: 1623-1633
Navar, J., Bryan, R.B., 1990. Interception loss and rainfall redistribution by three semi-arid
growing shrubs in northeastern Mexico. Journal of Hydrology 115: 51–63
Price A.G., Dunham K, Carleton T. 1997. Variability of water fluxes through the black
spruce (Picea mariana) canopy and feather moss (Pleurozium schreberi) carpet in the
boreal forest of northern Manitoba. Journal of Hydrology 196:310–323
Roth, BE, Slatton, KC, Cohen, MJ. 2007. On the potential for high-resolution lidar to
improve rainfall interception estimates in forest ecosystems. Frontiers in Ecology and
the Environment 5: 421-428
Rutter, AJ, Kershaw, KA, Robins, PC. 1971. A predictive model of rainfall interception in
forests I: derivation of the model from observations in a plantation of Corsican pine.
Agricultural Meteorology 9: 367-384

9
Rutter, A., Morton, A., Robin, P., 1975. A Predictive Model of Rainfall Interception in
Forests. II. Generalization of the Model and Comparison with Observations in Some
Coniferous and Hardwood Stands. Journal of Applied Ecology 12: 367–380
Saito T, Matsuda H, Komatsu M, Xiang Y, Takahashi A, Shinohara Y, Otsuki K. 2013.
Forest canopy interception loss exceeds wet canopy evaporation in Japanese cypress
(Hinoki) and Japanese cedar (Sugi) plantations. Journal of Hydrology 507: 287-298
Spittlehouse, DL 1998. "Rainfall Interception in Young and Mature Coastal Conifer Forests."
In: Mountains to Sea: Human Interaction with the Hydrological Cycle,
Y. A1i1a (ed.), Canadian Water Resources Association, Cambridge.
Staelens J, Schrijver AD, Verheyen K, Verhoest N. 2006. Spatial variability and temporal
stability of throughfall water under a dominant beech (Fagus sylvatica L.) tree in
relationship to canopy cover. Journal of Hydrology 330: 651-662
Taniguchi M, Tsujimura M, Tanaka T. 1996. Significance of stemflow in groundwater
recharge. 1: Evaluation of this stemflow contribution to recharge using a mass balance
approach. Hydrological Processes 10: 71-80
Tanaka T, Taniguchi M, Tsujimura M. 1996. Significance of stemflow in groundwater
recharge. 2: A cylindrical infiltration model for evaluating the stemflow contribution to
groundwater recharge. Hydrological Processes 10:81-88
Valente, F., David, J.S., Gash, J.H.C., 1997. Modelling interception loss for two sparse
eucalypt and pine forests in central Portugal using reformulated Rutter and Gash
analytical models. Journal of Hydrology 190: 141–162
Van Stan II JT, Stiegert CM, Levia DF, Scheick CE. 2011. Effects of wind-driven rainfall on
stemflow generation between codominant tree species with differing crown
characteristics. Agricultural and Forest Meteorology 151:1277-1286
Vincke C, Thiry Y. 2008. Water table is a relevant source for water uptake by a Scots pine
(Pinus sylvestris L.) stand: Evidences from continuous evapotranspiration and water
table monitoring. Agricultural and Forest Meteorology 148: 1419–1432
Whitford WH, Anderson J, Rice PM. 1997. Stemflow contribution to the ‘fertile island’
effect in creosote bush, Larrea tridentate. Journal of Arid Environments 35: 451-457

10
CHAPTER 2
THROUGHFALL HETEROGENEITY AND TEMPORAL PERSISTENCE BELOW
AND BETWEEN THE CANOPIES OF JUVENILE LODGEPOLE PINE (Pinus
contorta)
INTRODUCTION
Rain falling on a forest is either stored on the canopy and subsequently evaporates (canopy
interception loss) or it is routed by the canopy into understory precipitation. Understory
precipitation takes two forms - throughfall, TF, which reaches the forest floor by passing
through canopy gaps or by dripping from vegetation components, and stemflow, SF, which
reaches the ground by flowing down tree boles. At the plot-scale and beyond, TF represents
the largest volumetric component of understory precipitation, often accounting for > 70% of
gross precipitation at season-long or annual time scales (Carlyle-Moses and Price, 2007;
Levia et al., 2011; Macinnis-Ng et al., 2012), while SF is usually volumetrically minor at this
spatial scale, commonly representing < 5% of seasonal or gross precipitation (e.g., Helvey
and Patric, 1965; Marin et al., 2000; Reid and Lewis, 2009). Thus, TF represents a relatively
large component of understory precipitation and the quantity of water available for the
terrestrial portion of the hydrologic cycle. Throughfall is also a key transfer mechanism in
the biogeochemical cycles of wooded ecosystems (Parker, 1983; Levia et al., 2011; Bhat et
al., 2011).
Canopy interception loss - which often constitutes a sizable portion of the total
evaporation flux to the atmosphere from forests (Carlyle-Moses and Gash, 2011) - is almost
always derived by taking the difference between incident rainfall and the sum of TF and SF
(e.g., Carlyle-Moses et al., 2010; Návar, 2013; Saito et al., 2013). Given the quantitative
importance of TF compared to SF at the spatial scales in which canopy interception loss
estimates are made, TF spatial variability is an important consideration since any errors
accompanying estimates of the mean areal TF will be propagated in the canopy interception
loss estimate. Thus, accurate estimates of TF magnitudes are a requisite for hydrological and
biogeochemical studies conducted in forests; however, obtaining this accuracy poses a
methodological challenge (Lloyd and Marques, 1988; Holwerda et al., 2006; Zimmermann
and Zimmermann, 2014), largely due to the spatial heterogeneity of this hydrologic input
(e.g., Loustau et al., 1992; Price et al., 1997; Marin et al., 2000; Price and Carlyle-Moses,
2003).

11
Although much is yet to be learned (see Pypker et al., 2011), spatially varying TF
inputs have been attributed to the complexity of the forest cover including overlapping tree
canopies, varying tree heights, canopy gaps, and the arrangement and variation in leaf and
branch area (Zirlewagen and von Wilpert, 2001; Carlyle-Moses et al., 2004; Konishi et al.,
2006; Staelens et al., 2006; Poppenborg and Hölscher, 2009; Nanko et al., 2011).
Additionally, TF heterogeneity may be dependent on meteorological factors such as rainfall
depth and intensity, as well as wind speed and direction (e.g., Herwitz and Slye, 1995;
Carlyle-Moses 2004; David et al., 2006; Zimmermann et al., 2008). In some forests the
interaction of rainfall with the canopy produces systematic TF spatial patterns, such as where
TF varies as a function of the distance from the tree bole (e.g., Stout and McMahon, 1961;
Ford and Deans, 1978; Pedersen, 1992). In other forests, presumably due to the intricacy in
canopy composition, patterns of TF may be considered spatially random (Loustau et al.,
1992; Carlyle-Moses et al., 2004). Much focus has been paid to the spatial heterogeneity of
TF, especially as it influences TF sampling in terms of the number of individual gauges used
to meet certain statistical objectives related to the TF estimate (e.g., Kimmins, 1973; Puckett,
1991; Price and Carlyle-Moses, 2003), the periodic reassignment of gauge locations (e.g.,
Lloyd and Marques, 1988; Ziegler et al. 2009; Ritter and Regalado, 2014), and the type of
gauge used (Zimmermann et al., 2010; Carlyle-Moses et al., 2014).
Although an important consideration for TF sampling, TF heterogeneity may also
represent a connection between canopy structure and ecosystem functioning at the understory
level and within the soil profile (Ford and Deans, 1987; Carleton and Kavanagh, 1990;
Bouten et al., 1992; Manderscheid and Matzner, 1995). If such linkages exist in a particular
forest, the potential importance of TF spatial patterns on these various ecohydrologic
processes would either necessitate or at least be more pronounced if the spatial delivery of
TF was stable over time (Keim et al., 2005). The temporal persistence of TF spatial input has
been studied in tropical (Zimmermann et al., 2007; Zimmermann et al., 2008; Wullaert et al.,
2009), temperate deciduous (e.g., Keim et al., 2005; Gerrits et al., 2010) and temperate
coniferous forests (e.g., Raat et al., 2002; Keim et al., 2005; Carlyle-Moses et al., 2014), as
well as within a eucalyptus plantation (Sato et al. 2011). Additionally, the temporal stability
of TF spatial variability has been examined at the tree-scale for two deciduous species (see
Staelens et al., 2006; Fathizadeh et al., 2014). Temporally stable patterns in TF delivery

12
observed in these studies ranged from relatively ‘weak’ (e.g., Gerrits et al., 2010) to
‘significant’ (e.g., Staelens et al., 2006), suggesting that linkages, if any do exist, between the
spatial variability of TF and ecohydrological processes at the understory or within the soil
may be forest or tree-type dependent.
Mountain pine beetle (Dendroctonus ponderosae Hopkins) (MPB) has severely
affected the majority of pine species in the province of British Columbia, Canada, especially
lodgepole pine (Pinus contorta Douglas ex Louden var. latifolia Engelm. ex S. Watson). By
2004, more than seven million hectares of lodgepole pine forests were infested by MPB in
the province (Rice et al., 2007). The loss of mature lodgepole pine stands, including those
lost to salvage logging, has resulted in larger portions of the province being covered by
juvenile pine stands through planting and natural regrowth. With this change from mature to
juvenile cover at such a large scale, questions regarding the hydrology and ecosystem
functioning of impacted areas have been raised (Uunilla et al., 2006). Studies evaluating
canopy water balance dynamics in mature and juvenile lodgepole pine stands have found that
young stands produce volumetrically greater TF and SF than their mature counterparts (e.g.,
Spittlehouse, 1998); however, no studies to date have examined of the spatiotemporal
patterns associated with this increased TF in juvenile lodgepole pine environments. Such
studies are a requisite to improving our understanding of the hydro-ecological impacts of
disturbance regimes, such as MPB, in these and similar environments.
The objectives of this study, are to: i) determine the magnitude and spatial variability,
expressed by the coefficient of variation CV, of TF under four structurally distinct areas of
canopy coverage (inner-canopy, mid-canopy, canopy-periphery, and areas free of cover
between individual canopies [outside-of-canopy]), ii) evaluate if TF spatial variability
exhibits temporal persistence and if so to determine what the influence, if any, certain tree
and storm characteristics have on this stability, and iii) based on meeting objectives i and ii,
to provide recommendations for sampling strategies and for future research in these and
similar forests.

MATERIALS AND METHODS

13
Study area
Measurements of rainfall and TF were conducted at the Mayson Lake hydrological research
area, which is located on the Thompson-Bonaparte plateau approximately 60 km NNW of
Kamloops, British Columbia, Canada ( 51°12'49” N 120°23'43" W) at an elevation of 1290
m a.m.s.l. (Figure 2.1).

Figure 2.1: Geographic Location of the study plot. Adapted from Winkler et al.
(2010).

The study area has a long-term (1981-2010) mean annual precipitation depth of ~ 640
mm with ~280 mm falling during the growing-season (mid-May through early October)
(Wang et al., 2006; 2012). Precipitation in the form of rainfall dominates the growing-season
period, while precipitation during the dormant season is largely comprised of snow. Mean
annual temperature is 3.3 oC, with mean monthly temperatures ranging from -7.1 oC in
December to 13.9 oC in July. The Köppen climate classification for the region is Dfc (subarctic) while the provincial Biogeoclimatic Ecosystem Classification designation (BC
Ministry of Forests and Range, 2008) is Montane Spruce dry mild variant (B.C. Ministry of
Forests and Range, 2008).

14
The plot selected for this study was harvested in 1999 and planted with lodgepole
pine seedlings within the following 2 years. Point-centered quarter analysis (Cottam and
Curtis, 1956) revealed that in 2010 the plot (2600 m2) was dominated by lodgepole pine with
5640 stems ha-1, and a basal area of 9.6 m 2 ha-1, while subalpine fir represented 980 stems ha1

, and a basal area of 0.6 m 2 ha-1. No other tree species occurred in the plot. Average

lodgepole pine tree height was ~ 2.2 m (median ~1.7 m), with a maximum height of 5.2 m.
Average projected crown area (PCA) of the pines was ~ 0.9 m2 (median ~ 0.5 m2), with a
maximum of 3.2 m2, while the total PCA of the plot was ~ 4940 m2 ha-1. Although trees
sometimes occurred in groups with overlapping crowns, the majority of crowns, including
those of the trees in which TF was sampled, occurred in isolation from one another.

Rainfall and throughfall measurement
Rainfall depth and intensity were measured on an event basis from May 16 – October 4, 2010
using an Onset® tipping bucket rain gauge (model # S-RGB-M002) with a diameter of 15.4
cm, resolution = 0.2 mm tip-1. Additionally, rainfall depth was measured with a
Meteorological Service of Canada Type-B rain gauge (diameter = 11.3 cm). The tipping
bucket and the Type-B gauge openings were situated 1 m above the ground surface, located ~
40 m south-west of the site. A rain event was defined as a period of rainfall bounded by 8
hours with no measurable rainfall as indicted by the tipping bucket rain gauge record in
conjunction with Canada’s National Climate Data and Information Archive 10-minute time
step radar imagery (Silver Star Mountain Doppler radar station). An 8 hour bound for
individual events was selected as this was the time required for canopies in this and a similar
plot in the area to completely dry (McKee, 2010). Weather radar was also used to determine
the number and duration of intra-storm breaks ≥ 10 minutes and the direction of storm origin.
Wind speed and direction data were collected using an Onset® weather station (Hobo® Micro
Station Data Logger – H21-002 and associated sensors). However, the data files associated
with this station were corrupted and as such the data were not available for analysis.
Throughfall was measured on an event basis under the isolated crowns of nine trees,
the characteristics of which are provided in Table 2.1. For each tree 16 manually-read TruChek® wedge gauges (catch area = 36.3 cm2 each) were used, with four of these gauges
placed along four transects (N, W, S, E). Along each transect one gauge was placed adjacent

15
Table 2.1: Characteristics of the individual lodgepole pine trees used in this study.
Correlations among the tree size metrics were all significant (p ≤ 0.05): Height – Basal Area
(r = 0.908; rs = 0.933); Height – Projected Crown Area (r = 0.909; rs = 0.933), Basal Area –
Projected Crown Area (r = 0.999, rs = 1.0).
Tree ID
Height (m)
Basal Area (cm2)
Projected Crown Area (m2)
1
1.39
6.5
0.48
2
1.88
14.5
0.78
3
1.20
3.6
0.38
4
2.68
18.4
0.93
5
3.00
37.1
1.62
6
2.14
13.2
0.74
7
3.69
28.4
1.30
8
3.22
40.6
1.75
9
4.22
66.5
2.71
to the tree bole (inner-canopy), one at the mid-canopy point, one at the canopy-periphery,
and the fourth gauge situated at a random distance in the open area between tree canopies,
but within 45˚ of the top of the selected tree.
Canopy cover
Canopy cover fraction was determined above each TF gauge under the three below-canopy
zones using methods similar to those Staelens et al. (2006). Digital non-hemispherical colour
photographs (Canon Powershot A40) were taken vertically (using a bull’s-eye spirit level) at
the height of the top of each gauge (Llorens and Gallart, 2000; Staelens et al., 2006) under
overcast conditions in early October 2010. Photographs were scaled to an equivalent size of
the TF gauge opening (approximately 36 cm2) centered on the first branching level from the
ground. Canopy cover fraction was then derived using Geomatica 10.3.2. (PCI Geomatics
Inc., Richmond Hill, Ontario) using supervised pixel classification of the scaled photograph.
Because of the overlap between needle and wood components in the canopy, no attempt was
made to determine the fraction of the canopy that was foliage and that which was wood.

Temporal persistence of throughfall
Temporal persistence has been analyzed using differing techniques, many of which assume
that the distribution of the variable of interest follows a normal distribution (e.g., Vachaud et
al., 1985; Raat et al., 2002; Keim et al., 2005). However, at the rainfall event scale TF often
does not exhibit a normal distribution (e.g., Lloyd and Marques, 1988; Loustau et al., 1992)

16
and thus the method of standardizing TF used by Zimmermann et al. (2008), in which each
stationary TF gauge is corrected to zero median and unit variance, was used in this study:

(2.1)

where, TFSi is the standardized value of the throughfall depth measured for an individual
gauge, TFi (mm), TFmedian is the median TF depth (mm) of all gauges, and MAD represents
the median absolute deviation of TF(mm).
A plot of TFSi for each event and for each TF gauge, ranked from minimum to
maximum mean TFSi, shows the deviation of TFSi for all TF sampling points (Zimmerman et
al., 2008). These temporal stability plots show two types of persistence – extreme
persistence, referring to mean TFSi values that deviate from the median and are within either
the upper or lower quartiles of the mean TFSi ranked gauges, and general persistence, which
refers to the TFSi that deviate from the median TFSi and are within the interquartile range
(Keim et al., 2005; Zimmermann et al., 2008). Although others have used ± 1 standard
deviation as a criterion for persistence (e.g., Raat et al., 2002; Wullaert et al., 2009),
Zimmermann et al. (2008) suggest the stricter criterion of using the 95% confidence limits of
the mean TFSi. For comparative purposes with previous studies both the ± one standard
deviation and the 95% confidence limits of the mean TFSi are used in this study.
Additionally, mean TFSi values were examined to determine if they were significantly
different than zero at the 95% confidence level (Carlyle-Moses et al., 2014), which would
indicate non-random spatial patterns over time even if they did not meet the established
criteria for persistence.
Following Zimmermann et al. (2008), Spearman rank correlation coefficients, rs, for
all possible event pairings (n = 703 ) for TF were derived and examined to determine whether
rs could be modeled as a function of the temporal lag (days) between rainfall events, Δday. If
rs is a function of Δday then a change to the spatial pattern of TF through time can be
identified. Similar to Zimmermann (2008), differences in storm characteristics (rainfall depth
ΔP, event duration ΔDE, rain duration ΔDR, mean rainfall intensity ΔIMean, maximum 30
minute rainfall intensity ΔIMax-30, the number of intra-storm breaks ≥ 1 h each ΔNB and their
duration ΔDB, and the direction of storm origin ΔDoSO were included as additional variables

17
to determine what influence, if any, they may have on variations in rs. The best model was
then selected using the Akaike information criterion (Sakamoto et al., 1986) in keeping with
Zimmermann et al. (2008). The above procedure was conducted on all TF data from all four
sampling zones pooled and for each of the four sampling zones separately.

Data analysis
Statistical analysis was conducted using IBM® SPSS® Statistics Version 22 and Microsoft ®
Excel 2010. Where means could not be compared – determined using the Shapiro-Wilk test
(Shapiro and Wilk, 1965) for normality – analysis of differences were carried out using a
Tukey-type median test (Mood, 1950; Levy, 1979; see Zar 1984) for multiple comparisons
among medians or the Mann-Whitney test (Mann and Whitney, 1947) when two medians
were compared.

RESULTS
Incident precipitation
A total of 41 precipitation events ≥ 0.4 mm occurred during the course of this study. Two of
these events fell as a mix of rain and snow, while one rainfall event was not collected as
instrumentation was temporally removed due to a wildfire in close proximity to the site. For
the remaining 38 events, total rainfall depth was 252.9 mm, or approximately 90% of the
mean growing-season rainfall for the site. Individual rain events ranged from 0.5 mm (Event
7) to 30.8 mm (Event 4). Meteorological characteristics of each of these events along with
the date of event occurrence are provided in Table 2.2. The direction of storm origin was
predominately westerly with 30 of the events (79%) having a storm origin ranging from SSW
to NNW, with 19 of these events (50% of all events) having a south-westerly (SSW – WSW)
origin. Five events (13%) had a south-easterly origin (SSE – ESE), while only 2 events (5%)
came from a north-easterly direction. One event (Event 1) came directly from the south.
Absolute and relative throughfall depth
Data were available for 5459 (99.8%) of the possible 5472 point TF measurements (144 TF
gauges x 38 rain events), with the remainder discarded due to the gauges being disturbed
before

18
Table 2.2: Event number, date(s) of occurrence, rainfall depth (P, mm), event duration (DE,
h), rainfall duration (DR, h), mean rainfall intensity(IMean, mm h-1), maximum 30 minute
rainfall intensity, IMax-30, mm h-1), number of intra-storm breaks ≥ 1 h, (NB), total duration of
intra-storm breaks ≥ 1 h (DB, h) and the principal cardinal direction of storm origin (DoSO).
Event #
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38

Event Date
(2010)
05 / 17
05 / 18
05 / 26
05 / 27 - 30
05 / 31 - 06 / 03
06 / 04 - 05
06 / 05
06 / 07
06 / 07 - 08
06 / 08 - 11
06 / 15
06 / 21
06 / 22
06 / 25
06 / 30
07 / 04
07 / 12
07 / 13
07 / 20
07 / 21 - 22
08 / 03
08 / 07 - 08
08 / 08 - 09
08 / 21 - 22
08 / 26 - 27
08 / 29
08 / 31 - 09 / 01
09 / 04
09 / 06 - 07
09 / 07 - 08
09 / 08
09 / 10 - 11
09 / 15 - 16
09 / 23 - 24
09 / 24
09 / 26
09 / 28
10 / 10

P
(mm)
5.9
1.5
0.7
30.8
16.8
7.5
0.5
0.8
9.1
20.5
20.0
7.0
2.5
6.7
1.3
12.8
6.2
5.0
2.4
2.1
1.4
8.0
5.9
6.0
16.4
1.4
6.4
5.7
15.1
4.9
1.7
6.8
2.4
1.9
3.2
0.7
2.3
2.6

DE
(h)
4.8
5.5
4.2
72.0
63.0
19.7
3.3
3.2
24.3
65.0
8.2
2.8
7.7
26.2
3.3
17.2
11.5
3.3
6.7
25.2
0.7
29.0
30.8
23.3
21.8
0.5
28.0
1.0
23.7
1.7
3.5
6.0
10.2
2.2
2.2
2.3
3.7
2.2

DR
(h)
4.8
4.8
3.3
29.3
19.0
14.0
2.7
1.0
8.2
33.3
6.3
2.8
3.2
6.7
1.7
8.7
4.2
3.3
2.2
4.5
0.7
3.3
11.0
4.2
6.5
0.5
13.3
1.0
15.2
1.7
2.2
6.0
5.0
1.8
2.2
1.5
1.8
1.0

IMean
(mm h-1)
1.2
0.3
0.2
1.1
0.9
0.5
0.2
0.8
1.1
0.6
3.2
2.5
0.8
1.0
0.8
1.5
1.5
1.5
1.1
0.5
2.1
2.2
0.5
1.4
2.5
2.8
0.5
5.7
1.0
2.9
0.8
1.1
0.5
1.0
1.5
0.5
1.3
2.6

IMax-30
(mm h-1)
3.1
0.4
0.2
5.0
9.6
0.8
0.5
0.3
11.5
1.8
8.8
9.5
4.1
5.7
0.8
3.8
5.4
2.7
3.2
2.3
2.1
5.4
1.8
3.6
9.9
2.8
3.6
7.7
1.1
6.6
1.4
2.7
1.4
1.4
2.7
0.6
2.7
3.0

NB
0
0
0
13
11
2
0
1
4
11
1
0
1
3
1
3
4
0
3
5
0
5
7
3
7
0
4
0
3
0
1
0
3
0
0
0
1
0

DB
(h)
0.0
0.0
0.0
32.5
36.3
2.7
0.0
1.2
15.0
26.2
1.5
0.0
4.0
14.7
1.2
7.0
5.2
0.0
4.0
19.5
0.0
24.3
18.7
19.2
15.3
0.0
14.7
0.0
6.5
0.0
1.3
0.0
4.0
0.0
0.0
0.0
1.5
0.0

DoSO
S
SW
SSE
SE
SW
SSW
SSW
WSW
WSW
SSE
NE
SW
SW
SW
WSW
NW
WNW
NNE
NNW
WNW
NW
W
WSW
SW
SW
WNW
WSW
SW
WNW
SE
ESE
W
NW
SW
SW
SW
WNW
SW

a measurement could take place. Throughfall data were collected for all 38 events from 132
of the 144 (91.7%) sample points. When all gauges are considered, cumulative point TF
depths ranged from 69.1 mm (27.3% of rainfall) for the gauge situated in the inner-canopy
zone of Tree 7 along the north transect to 327.9 mm (129.7% of rainfall) for the gauge
located at the canopy-periphery of the southern transect of Tree 8. Thus, over the study

19
period cumulative point TF was found to range by 258.8 mm, slightly greater than the
cumulative rainfall depth, and had a maximum cumulative TF to minimum cumulative TF
ratio of 4.7. At the rainfall event scale the relative TF ranges were often much greater. For
example, during Event 2 with a rainfall depth of 1.5 mm, TF was found to range from 0.0
mm for two gauges (along the east transect of Tree 8 in the inner-canopy and mid-canopy
zones) to 6.4 mm (426.7% of rainfall) in the canopy-periphery zone along the south transect
of Tree 4.
For comparative purposes with other studies (e.g., Staelens et al., 2006; Fathizadeh et
al., 2014) and given the range and frequency of rain event depths, cumulative TF was derived
for four rain depth classes: < 2 (n = 10), 2 to < 5 (n = 8), 5 to < 10 (n = 13), and ≥ 10 mm (n
= 7). Additionally, for meaningful comparisons among the different rain depth classes TF
was expressed as a percentage of cumulative rainfall, TF% (Sato et al., 2011; Fathizadeh et
al., 2014). There was a tendency for TF% to increase with both increasing distance from the
tree bole and with increasing size of the rainfall depth class (Figure 2.2). The TF% frequency
distributions of the inner-canopy zone for the 5 to < 10 mm, and ≥ 10 mm rain depth classes
as well as for the entire season significantly differed from the normal distribution, as did the
cumulative TF% distributions associated with < 2, 5 to < 10 mm, and ≥ 10 mm rain depth
classes for the canopy-periphery zone (Shapiro –Wilk test, p ≤ 0.05). The median cumulative
TF% values for the four rain depth classes and for the season significantly differed (p ≤ 0.05)
among the sampling zones. The exceptions were i) cumulative median TF% for the innercanopy and mid-canopy zones did not significantly differ for the < 2 and 2 to < 5 mm rain
depth classes, and ii) mid-canopy zone and canopy-periphery zone cumulative TF% were not
significantly different (p ≤ 0.05) under the ≥ 10 mm rain depth class and for the cumulative
season rainfall scenario.

Throughfall spatial variability
Figure 2.3a shows that the spatial variability of accumulated TF depth (mm), expressed by
the coefficient of variation (TF CV%), decreased in an asymptotic fashion with accumulated
rainfall depth for all four sampling zones. Once the first five rain events of the study (totaling
55.7 mm) had fallen, TF CV% remained stable at approximately 30, 26, and 18% for the

20
a

160
120
80
40
0
160

b

120

Throughfall (%)

80
40
0
160

c

120
80
40
0

d
160
120
80
40
0
<2

2 to < 5 5 to < 10

≥ 10

Season

Rainfall Depth Class (mm)

Figure 2.2: Box-plots of cumulative TF% for different rain depth classes and for the season
for the different sampling zones: a) inner-canopy, b) mid-canopy, c) canopy-periphery, and
d) outside-of-canopy.

21
inner-canopy, mid-canopy, and canopy-periphery zones, respectively. Accumulated TF CV%
for gauges situated in the outside-of-canopy zone reached a stable value of approximately 4%
once four events with an accumulated rainfall depth of 38.9 mm had fallen. At the rainfall
event scale, TF CV% decreased asymptotically with increasing rainfall depth for all sampling
zones, including the outside-of-canopy zone (Figure 2.3b).

a

40
30
20

100
75
50

10

25

0

0
0

50

100

150

200

250

Accumulated Rainfall Depth (mm)

b

125
CV of Event TF (%)

CV of Accumulated TF (%)

50

0

5

10

15

20

25

30

35

Event Rainfall Depth (mm)

Figure 2.3: Accumulated (a) and event (b) TF CV% as a function of rain depth for the innercanopy (closed circles), mid-canopy (open circles), canopy-periphery (closed squares) and
outside-of-canopy (open squares).
Using the Shapiro-Wilk test the frequency distributions of TF CV% for all sampling
zones were found to be significantly different (p ≤ 0.05) than the normal distribution. As
with cumulative TF%, differences among and between median TF CV% values of the
different sampling zones were assessed using nonparametric tests. The median value of TF
CV% for gauges situated in the outside-of-canopy zone (10.3%) was significantly lower
(Mann-Whitney test, p < 0.001) than that of gauges under the canopy, i.e., all three belowcanopy zones combined, (50.2%). When the three below-canopy zones are compared
significant differences (p ≤ 0.05) were found between median TF CV % values between the
inner-canopy gauges (56.4%) and those situated at the canopy-periphery (34.6%) and
between gauges located within the mid-canopy zone (48.4%) and those at the canopyperiphery. No significant difference (p ≥ 0.05) was found between the median TF CV% of
the inner-canopy zone and the mid-canopy zone. Significant rs values (p ≤ 0.05) between

22
event TF CV% for each sampling zone and various rain event characteristics are provided in
Table 2.3.

Table 2.3: Spearman rank correlation coefficients (n = 38) between the CV of event TF%
and the storm characteristics of rainfall depth, P (mm), event duration, DE (h), mean rainfall
intensity IMean (mm h-1), maximum 30-minute rainfall intensity, IMax-30 (mm h-1), number of
storm breaks ≥ 30 minutes, NB, and total duration of those breaks, DB (h). Level of
significance: *p ≤ 0.05; ** p ≤ 0.01.
TF CV%
P
DE
IMean
IMax-30
NB
DB
-1
(mm)
(h)
(mm h
(mm h )
(h)
1
)
Inner-Canopy
-0.488**
-0.372*
0.059
-0.335*
-0.299
-0.300
Mid-Canopy
-0.649**
-0.467** -0.019
-0.369*
-0.305
-0.333*
**
**
*
Canopy-Periphery -0.491
-0.465
0.212
-0.239
-0.393
-0.404*
Outside-of-0.871**
-0.661** -0.114
-0.515**
-0.524**
-0.571**
Canopy
Throughfall, tree size and canopy cover
No significant Spearman or Pearson correlations (p ≤ 0.05) were found among tree size
metrics (tree height, basal area, and projected canopy area) and the median of cumulative
TF% for any rain depth class or for cumulative rain over the season for both the inner-canopy
and outside-of-canopy zones. However, tree size characteristics were found to be correlated
with median cumulative TF% in the mid-canopy zone for smaller rain depth classes and with
canopy-periphery median TF% for the largest rain depth class (≥ 10 mm) and cumulative
season rainfall (Table 2.4).
Cumulative season-long TF% was found to be significantly correlated (p ≤ 0.01) with
canopy cover fraction (%) above the gauge when all three below-canopy sampling zones are
grouped together (rs = -0.517, r = -.459, n = 99). However, when cumulative TF% was
compared with canopy cover fraction within each of the below-canopy zones separately,
cumulative TF% was found to be only significantly correlated with canopy cover fraction for
the < 2 mm (p ≤ 0.01, rs = -0.430, r = -0.465, n = 36) and 2 to < 5 mm rain depth classes (p ≤
0.05, rs = -0.386, r = -0.360, n = 35) for the mid-canopy zone.
Throughfall and transect direction
Significant differences among median cumulative TF% values for different transect

23
Table 2.4: Spearman rank correlation coefficients (n = 9) between tree size characteristics
and median cumulative TF% in each of the four sampling zones for four rain depth classes
and for the entire season. Level of significance: *p ≤ 0.05, **p ≤ 0.01. Note: Basal area and
PCA are listed together since the ranks of each of the tree basal areas of the nine trees match
the ranks of each of the PCA values (i.e., rs = 1.0).
Median Cumulative TF%
Rain Depth Class
< 2 mm 2 to < 5 mm 5 to < 10
≥ 10 mm
Season
mm
Inner-Canopy
Tree Height
0.185
-0.150
-0.033
-0.367
-0.300
Basal Area or PCA
0.277
-0.167
0.033
-0.400
-0.300
Mid-Canopy
Tree Height
Basal Area or PCA

-0.695**
-0.628

-0.800**
-0.750*

-0.117
-0.050

-0.500
-0.533

-0.550
-0.567

Canopy-Periphery
Tree Height
Basal Area or PCA

0.267
0.400

0.167
0.050

-0.517
-0.417

-0.833**
-0.867**

-0.767*
-0.717*

Outside-of-Canopy
Tree Height
Basal Area or PCA

0.467
0.400

0.008
-0.042

0.000
0.232

-0.400
-0.100

0.317
-0.233

directions under differing rain-depth classes and canopy zones were found (Table 2.5). These
differences were largely limited to the inner-canopy and mid-canopy zones and for rain-depth
classes at or below 5 to < 10 mm. However, the cumulative median TF% values of the south
transect canopy-periphery and the east transect canopy-periphery also differed significantly,
albeit at the p ≤ 0.10 level. A trend is evident, and significant differences exist, in cumulative
median TF% for east transects being lower than south and or west transects and, although
less prevalent, north transect values being lower than those associated with west transects.
The influence of the direction of storm origin DoSO, on differences in TF between
transects was evaluated by taking the ratio of the storm-ward TF to lee-ward TF. Storm-ward
TF was equated with the median TF of a given canopy zone along the transect of the same
direction when the DoSO was a cardinal direction (e.g., S), or, in the case where the DoSO
was an intercardinal (e.g., SW) or a secondary-intercardinal direction (e.g., SSW) the median
TF value was derived, respectively, as the average or the weighted average of the median TF
of the two corresponding transects. The lee-ward TF was derived as above, but for the
opposite cardinal, intercardinal, or secondary-intercardinal direction as the DoSO. Storm-

24
Table 2.5: Median cumulative TF% as a function of canopy position and transect direction
(n = 9 for each case). Values in bold indicate medians that significantly differ (median test:
*
p ≤ 0.10, **p ≤ 0.05, ***p ≤ 0.01) from one another with the transect direction of the differing
value superscripted (N = north, S = South, E = east, and W = west).
Median Cumulative
Rain Depth Class
TF%
Position
< 2 mm
2 to < 5 mm 5 to < 10 mm ≥ 10 mm Season
Inner-Canopy
North
28.6
33.3
49.2
53.2
47.5
South
29.4
37.5E*
56.0E*
58.5
44.8
W*
S*
W**,S*
East
25.0
22.9
43.8
55.1
37.5
West
40.0E*
49.4
63.3E**
49.0
53.8
Mid-Canopy
North
South
East
West

38.5W*
42.9
28.6W***
64.3E***, N*

46.4W*
60.3E*
44.8W***, S*
77.3E***,N*

65.7W**
76.3
59.3W**
82.0E**, N**

85.1
72.5
70.1
91.0

62.7
66.7
55.0
80.0

Canopy-Periphery
North
South
East
West

69.2
69.2
57.1
68.4

83.8
82.5
63.1
75.0

83.3
93.2
71.6
85.7

90.2
90.9E*
78.6S*
83.3

82.7
89.0
66.1
83.3

Outside-of-Canopy
North
South
East
West

71.4
71.4
71.4
78.9

81.5
82.8
83.7
88.6

90.0
90.0
92.3
92.3

91.6
92.3
92.5
94.6

88.2
89.0
88.0
90.7

ward to lee-ward TF ratios for each storm and for each sampling zone are shown in Figure
2.4, where it is evident that DoSO does have an influence on TF heterogeneity within each
zone. Mean event ratios (n = 38) for all four zones significantly differed from unity: innercanopy (1.55 ± 1.11, p = 0.0022); mid-canopy (1.49 ± 0.79, p = 0.0001); canopy-periphery
(1.22 ± 0.38, p = 0.0004), and outside-of-canopy (1.15 ± 0.47, p = 0.0435).
Temporal stability of throughfall
Table 2.6 list the Spearman and Pearson correlation coefficients between cumulative TF% of
the four different rain depth classes for the four sampling zones. In general, correlation
coefficients were highest between cumulative TF% with similar rainfall depths. For example,

25
7

a

6
5
4
3
2
1
0
7

b

Ratio of Storm-Ward to Lee-Ward TF

6
5
4
3
2
1
0
7
6
5
4
3
2
1
0
7

d

6
5
4
3
2
1
0

Event #
0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

34

36

38

Figure 2.4: Ratio of event storm-ward to lee ward TF throughout the study period for the a)
inner-canopy, b) mid-canopy, c) canopy-periphery, and d) outside-of-canopy sampling zones.

26
Table 2.6: Spearman (right) and Pearson (left) correlation coefficients between cumulative
TF% of four different rain-depth classes ( < 2, 2 to < 5, 5 to < 10, and ≥ 10 mm) for the four
sampling zones (inner-canopy, mid-canopy, canopy-periphery, and outside-of-canopy) (n =
36, *p ≤ 0.05, **p ≤ 0.01; n.s. = not significant).
Rain Depth Class
2 to < 5 mm
5 to < 10 mm
≥ 10 mm
Inner-Canopy
< 2 mm
0.576**, 0.654**
0.633**, 0.656**
0.391* , 0.347*
2 to < 5 mm
0.806**, 0.891**
0.717**, 0.695**
5 to < 10 mm
0.776**, 0.799**
Mid-Canopy
< 2 mm
2 to < 5 mm
5 to < 10 mm
Canopy-Periphery
< 2 mm
2 to < 5 mm
5 to < 10 mm
Outside-of-Canopy
< 2 mm
2 to < 5 mm
5 to < 10 mm

0.869**, 0.868**

0.827**, 0.791**
0.889**, 0.853**

0.524**, 0.584**
0.554**, 0.630**
0.616**, 0.771**

0.719**, 0.733**

0.658**, 0.597**
0.813**, 0.793**

0.484**, 0.507**
0.742**, 0.810**
0.827**, 0.848**

0.544**, 0.550**

0.387* , 0.423*
0.830**, 0.831**

0.149n.s., 0.173n.s.
0.573**, 0.598**
0.788**, 0.797**

the strongest correlations were found between the 2 to < 5 and 5 to < 10 mm rain-depth
classes for the inner-canopy, mid-canopy, and outside-of-canopy zones (rs ranging from
0.806 to 0.881, p ≤ 0.05 , r from 0.831 to 0.891, p ≤ 0.05), while for the canopy-periphery
zone the strongest correlations were found between the 5 to < 10 and ≥ 10 mm rain-depth
classes (rs = 0.827, p ≤ 0.05; r = 0.848, p ≤ 0.05). The weakest correlations were found
between the < 2 mm and the ≥ 10 mm rain-depth classes under all four sampling zones with
the correlations not being significant for the outside-of-canopy-zone.
Spearman rank coefficient values for TF for all event pairings were derived for each
of the canopy zones. Significant (p < 0.05) positive rs values occurred for 277 (39.4%), 309
(44.0%), 370 (52.6%), and 295 (42.0%) of the 703 event pairings for the inner-canopy, midcanopy, canopy-periphery and outside-the-canopy zones, respectively. Significant negative
pairing were also found for the inner-canopy, mid-canopy and outside-the-canopy zones, but
comprised only 16 (2.3%), 9 (1.3%), and 4 (0.6%) of the pairings, respectively. No

27
significant negative rs pairings were found for the canopy-periphery zone. Maximum
positive rs values for individual event pairings ranged from 0.839 for the canopy-periphery
between events 12 and 19 to 0.891 for the inner-canopy between events 23 and 28, while
maximum negative rs values ranged from -0.232 (not significant) for the canopy-periphery
(between events 17 and 36) to -0.620 for the inner-canopy (between events 12 and 20).
Figure 2.5 shows the temporal stability plot with all gauges from all zones pooled. Twentynine (81%) and 4 (11%) of the 36 gauges comprising the lower quartile were found to have
extreme temporal persistence following the ± 1 SD and 95% confidence limit criteria,
respectively. Extreme temporally persistence of TF input was also observed in the upper
quartile region with 16 of the 36 gauges (44%) meeting the ± 1 SD criterion; albeit no gauge
met the 95% confidence limit criterion. No gauges meet either persistence criterion within
the inter-quartile region and thus, no general persistence was identified. Gauges situated in
the inner-canopy dominated the extreme temporally persistent low TFSi values (69%), with
the mid-canopy (24%) and canopy periphery (7%) comprising the rest (i.e., no outside-ofcanopy gauges). Seventy-five percent of the gauges identified as exhibiting extreme temporal
persistence in the upper quartile were located outside-the-canopy, with gauges situated in the
canopy periphery (19%) and mid-canopy (6%) accounting for the remainder. One-hundred
and five of the 144 gauges (73%) were found to have TFSi values that differed from zero at
the 95% confidence level; 52 (36%) below zero and 53 (37%) above.
Figure 2.6 shows the temporal stability plots when each canopy zone is considered
separately. At the canopy zone scale the frequency of temporally persistent TFSi decreases
considerably compared to when all zones are pooled. This is especially the case for the
outside-of-canopy, inner-canopy, and mid-canopy zones, with only 1 (3%), 1 (3%), and 2
(6%) of the gauges exhibiting extreme persistence (i.e., in the region of the upper and lower
quartiles of the ranked sample points) using the ± 1 SD criterion, respectively. No gauges
were found to have extreme persistence using the 95% confidence limit criterion in the three
aforementioned zones. The presence of temporally persistent TF was greater in the canopyperiphery zone than the other canopy zones, but was not as prevalent as when all zones are
pooled. Nine of the 36 gauges (25%) in the canopy-periphery zone were found to exhibit
extreme persistence using the ± 1 SD criterion and none using the 95% confidence limit
criterion. When each canopy zone is considered separately, no gauges with general

28

20

Normalized Throughfall (TFSii)

15

10

5

0

-5
0

12

24

36

48

60

72

84

96

108

120

132

144

Gauge Order

Figure 2.5: Temporal stability plot of normalized throughfall (TFSi) for all 144 gauges.
Individual dots represent event TFSi values, while the solid line passes through the mean
TFSi value for all gauges ranked from lowest to highest mean TFSi.
persistence (i.e., in the interquartile region of the ranked sample points) were found using
either criterion. At the canopy-zone scale 20 (56%), 19 (53%), 21 (58%), and 22 (61%) of the
36 gauges in each the inner-canopy, mid-canopy, canopy-periphery, and outside-of-canopy
zones, respectively, had individual mean TFSi values that differed significantly (p ≤ 0.05)
from zero. Exceptionally large single-event deviations of TFSi from the median (i.e., zero)
were found. When all gauges are pooled, individual TFSi values were as high as + 18.0
(Event 8, Tree 2, mid-canopy, north), while at the canopy sampling zone scale the maximum
TFSi value was + 25.3 (Event 26, Tree 2, inner-canopy, north). It is important to note that
these two extreme values correspond to the same tree and transect.
When all gauges are pooled and when each of the canopy zones are analyzed
separately the rs of event pairings were found to be significantly (p < 0.001) negatively
correlated to the temporal lag, Δday. However, the amount of variability in the rs values of

29

a

20
10
0
-10

b

20
10
0
-10

c

20
10
0
-10

20

d
10

0

-10
1

6

11

16

21

26

31

36

Gauge Order

Figure 2.6: Temporal stability plots for the 36 gauges in each of the a) inner-canopy, b) midcanopy, c) canopy-periphery, and d) outside-of-canopy sampling zones with median set to
zero and unit variance (median absolute difference, MAD).

30

the event pairings explained was very low in each case ranging from R2 = 0.027 for the
outside-the-canopy zone to R2 = 0.078 for the mid-canopy zone. For each canopy zone and
when all gauges were pooled the addition of meteorological variables did increase the
amount of variability in rs explained (Table 2.7); however, a sizable portion of the variability
remained unaccounted for with R2 values only reaching 0.050 (inner canopy) to 0.130 (all
gauges pooled).

DISCUSSION
Under all four lodgepole pine canopy zones TF % increased with increasing rain-depth class,
in keeping with the findings of studies conducted for other trees or forest types (Gómez et al.,
2002; Price and Carlyle-Moses, 2003; Ziegler et al., 2009; Fathizadeh et al., 2014). For the
three below-canopy zones the increase in TF % with increasing rain depth likely reflects the
gradually saturation of the canopy and the resultant fraction of rainfall reaching the ground as
TF (Carlyle-Moses et al., 2004). For relatively small events, TF is comprised solely of, or is
at least dominated by, free thoughfall – that is, TF that passes directly through gaps in the
canopy (see Bringfelt and Härsmar, 1974; Gash 1979). With increasing rain input the water
storage capacity of the canopy decreases and thus the contribution to TF in the form of
canopy drip increases (Jackson, 1975; Carlyle-Moses et al., 2014). The lower median
cumulative TF % values for the < 2 and 2 to < 5 mm rain depth classes for the inner- and
mid-canopy zone compared with the canopy-periphery suggests that these sheltered areas
closer to the tree bole have limited free thoughfall input and or require greater rain depths to
generate sizable TF drip. Cumulative median TF % for the canopy-periphery was relatively
high for small rain depth classes only increasing slightly under greater rain depth scenarios
compared to the inner and mid-canopy. Additionally, no significant difference in cumulative
TF % between the canopy-periphery or the mid-canopy was found for the ≥ 10 mm rain
depth class. These findings suggests that, in general, the sparse canopy periphery is
associated with relatively large free throughfall contributions and that the role of canopy drip
TF at this zone is, for the most part, relatively minor compared with the inner- and midcanopy. Additionally, because of the excurrent nature of the branching of these juvenile
lodegpole pine trees, we speculate that drainage of intercepted rainfall from the canopy

31
periphery may move along branches inwards towards the tree bole (Ford and Deans, 1978). It
should be noted that these observations are at the canopy-zone scale and not at the gauge
scale as TF generated from canopy drip may result in individual gauges at the periphery and
the other below canopy zones receiving relatively large and sometimes temporally persistent
TF inputs.
The general pattern of TF increasing with increasing distance from the boles of
juvenile lodgepole pine observed in this study is in agreement with observations reported for
other trees (e.g., Aussenac, 1970; Johnson, 1990; Beier et al., 1993; Ito et al., 2008).
However, this pattern is not universal with some researchers (e.g., Ford and Deans, 1978;
Robson et al., 1994; Sato et al., 2011) finding the opposite to be true, that TF generally
deceases with increasing distance from the bole. Some researchers found that the position of
greatest TF input is located mid-way between the tree bole and the canopy periphery
(Carleton and Kavanagh, 1990; Nanko et al., 2011), while others researchers found that no
relationship between TF quantity and the position of the gauge relative to the tree bole exists
(Loustau et al., 1992; Gómez et al., 2002; Loescher et al., 2002). Keim et al. (2005) found
that the relationship between TF depth and distance to the nearest tree bole varied depending
on species and age. We suggest that the pattern observed in this present study may be the
result of a combination of factors, namely i) the isolated nature of the study trees with no
overlap from other canopies, ii) the increase in canopy cover and, from casual observations in
the field, the increase in the proportion of wood (which typically has a higher water storage
capacity than foliage – see Herwitz, 1985; Llorens and Gallart, 2000) comprising that cover
and the associated interception efficiency in the inner portions of the canopy compared with
the periphery, and iii) although the excurrent branches may divert intercepted rainfall from
the canopy periphery, the smoothness of these branches promotes flowall the way to the bole
resulting in stemflow generation rather than canopy drip TF. The relatively large stemflow
funneling ratios for juvenile lodgepole pines reported by McKee (2010) supports this
assertion.
The increase in median cumulative TF% in the outside-of-canopy zone with
increasing rain depth suggests that trees influence TF input beyond their projected canopy
areas, confirming the results of others (e.g., King and Harrison, 1998; David et al., 2006).

32

Table 2.7: Coefficient values associated with meteorological factors influencing Spearman rank coefficient values of paired events
and the intercept, and R2 for each model for all gauges pooled and under each of the four canopy zones. (n = 703, *p ≤ 0.05, **p ≤
0.01; ***p ≤ 0.001).
Zone
Δday
ΔP
ΔD
ΔDP
ΔIMean
ΔIMax30
ΔNB
ΔDB
ΔDoSO Intercept
R2
E

All Pooled

-0.001***

-

-

-

-0.025***

-0.012***

-

+0.004***

-0.0003*

+0.523***

0.130

Innercanopy

-0.002***

-

-

-

-

-

-

+0.003**

-

+0.268***

0.050

MidCanopy

-0.002***

-0.004***

-

-0.006***

-

-0.012***

-

+0.004**

-

+0.384***

0.105

CanopyPeriphery

-0.001***

-0.004***

-

-

-

-

-

+0.002**

-

+0.422***

0.073

OutsideofCanopy

-0.002***

-

-

-

-

-0.009**

+0.005*

-

-0.0007***

+0.421***

0.112

33
The interaction of the canopy with wind-driven rainfall can create rain-shadows in the leeward areas beyond the tree canopy (Herwitz and Slye, 1992; 1995; David et al., 2006). At a
given wind speed, wind-driven rainfall will tend to be more inclined for events with lower
rainfall depths since these events tend to be of lower intensity (in our study rs between P and
IMean = 0.332, p = 0.0418). The angle of inclination of rain falling at higher intensities is less
affected by wind given the larger drop size and associated kinetic energy (Laws and Parsons,
1943). However, the lack of wind velocity data for our study limits any further conclusions
regarding the role of the interaction of wind speed, rain depth and rain intensity. As stated by
David et al. (2006), estimates of canopy interception loss from isolated trees using only
below canopy gauges will violate the assumptions of mass balance and will result in an
underestimate of the true interception efficiency of the tree(s) under wind-driven rainfall
conditions.
As has been reported by others (e.g., Lin et al., 1997; Rodrigo and Àvila, 2001; Vrugt
et al., 2003; Staelens et al., 2006), the TF CV% in the study stand as a whole and under all
four canopy sampling zones decreased with increased rain depth up to a certain rain depth
threshold. In this present study the variability of TF was significantly greater below tree
canopies than in the areas between the canopies. Additionally, TF heterogeneity was
significantly greater in the inner-and mid-canopies compared to the canopy-periphery. These
findings suggest that canopy structure influences TF variability (see also Staelens et al.,
2006; Nanko et al., 2011; Pypker et al., 2011; Sato et al., 2011); however, our results also
suggest that the way in which canopy structure influences TF variability is complex.
Although canopy cover fraction was significantly negatively correlated with TF% when all
gauges were pooled, when each of the three below- canopy zones were considered separately
only canopy cover fraction was negatively correlated with TF% in the mid-canopy zone
under the two smallest rain depth classes. Additionally, tree size metrics, including tree
height and basal area were not correlated to TF% for the inner-canopy zone, while they were
for the mid-canopy zone for rain depth classes at or below 2 to < 5 mm and for the canopyperiphery for the ≥ 10 mm depth class. Thus, different canopy traits influence the partitioning
of water into TF differently depending on location under the canopy (e.g., inner-canopy
versus canopy-periphery) and the magnitude of the rain event, in agreement with the findings
of Carlyle-Moses et al. (2004) in a mature pine-oak forest in northeastern Mexico.

34
Our results suggest that tree traits work in concert with meteorological factors in
determining the spatial pattern of TF for a given storm. For example, TF CV% decreased
with increasing 30-minute maximum rain intensity, the increasing duration of the event, and
or the number and or duration of intra-storm breaks depending on the canopy sampling zone
and or the rain depth class. We speculate that these storm variables, namely the 30-minute
maximum rain intensity and intra-storm breaks, may influence TF CV% by reducing the
development and or connectivity of flow paths (i.e., translocation) in the canopy, with these
flow paths resulting in large drip points that increase TF CV% (see Herwitz, 1987; CarlyleMoses and Price, 2006; Nanko et al., 2011; Sato et al., 2011). Others (e.g., Bouten et al.,
1992; Raat et al., 2002) have also found that TF CV% decreases with some aspect of rainfall
intensity, however, Hansen (1995) concluded, based on results from a Norway spruce forest
in Denmark, that the amount of TF reaching different layers of the canopy remained
approximately equal regardless of rainfall intensity, while Loescher et al. (2002) found, in an
old-growth tropical forest, that TF CV% increased with increasing rain intensity. Thus, the
influence of rainfall intensity on TF CV% seems to be dependent on some aspect of canopy
structure and is not likely to be universal across all forest or tree types.
Direction of storm origin was found to have an important influence on TF quantity at the
gauge scale, with TF typically being larger on the storm-ward side of the canopy than the leeward side for gauges in all four sampling zones. We suggest that this is the result of the
probable relationship between the direction of storm origin and wind direction; however, the
lack of wind speed and direction data for our study limits our interpretation of the processes
involved. However, as aforementioned, wind-driven rainfall has been found to create rainshadows on the lee-ward sides of trees in tropical forests (Herwitz and Slye, 1992; 1995) and
Van Stan et al. (2011) found, in a temperate deciduous forest of the eastern United States,
that preferential stemflow generation occurred during periods of inclined rainfall by wind
from the east to north-northeast. These results, coupled with those of the present study,
suggest that the spatial variability of understory precipitation is controlled, at least to some
degree, by wind speed and direction and these variables should be evaluated in future studies
on the controls on TF variability.
Temporal persistence of TF spatial patterns was found in this forest type as it has in
others (e.g., Keim et al., 2005; Zimmermann et al., 2009; Sato et al., 2011; Carlyle-Moses et

35
al., 2014). A greater degree of persistence was found when all gauges were pooled compared
to when canopy zones were evaluated separately. This finding is probably a consequence of
the large differences in canopy trait characteristics between the zones. Persistence of median
cumulative TF was greatest between rain depth classes of more similar magnitudes. At the
event and gauge scale, however, persistence was much more variable reflecting the complex
interaction of varying storm conditions with varying sub-canopy zone characteristics. In our
study the maximum 30-minute rain intensity, duration of rain, direction of storm origin and
the number and duration of storm breaks were found to be correlated with temporal stability
for at least one of the zones or when all zones were pooled. Although spatial patterns were
more similar with more alike maximum 30-minute intensity (in agreement with Zimmermann
et al., 2008), rain duration and storm directions, the relationship between the number and
duration of storm breaks was found to be opposite. That is, with greater dissimilarities in
break periods and frequencies among events, the more similar the event spatial TF patterns.
This finding suggests that, for our study at least, an increase in the number and or duration of
storm breaks may dampen the degradation in temporal persistence resulting from differences
in the magnitudes of other variables such as rain depth and the maximum 30-minute
intensity. Nonetheless, much of the variability in the persistence of TF spatial patterns from
one event to the next in this study was not explained. Again, wind speed and direction may
be important factors that require future study.
The temporal persistence of TF variability in our study was found to marginally
degrade with time across the study period, probably as a result of the slight changes to the
canopy during growth across the time span of the study (five months). Zimmerman et al
(2008) found that temporal persistence was much more sensitive to temporal lag in an open
tropical rainforest in Brazil, especially during the early wet season when there was rapid
vegetative growth. Zimmermann et al. (2008) also found that as the wet season progressed
the degradation in temporal persistence was markedly reduced due to the more stable cover
characteristics. Thus, our results coupled with the findings of Zimmermann et al. (2008) and
others (e.g., Staelens et al., 2006) suggest that the temporal persistence of TF spatial patterns
is strongly linked with the temporal dynamics of the canopy. The relatively minor influence
of temporal lag on TF spatial patterns over the course of our study and the finding that a large
number of gauges had associated median TFSi values that significantly differed from zero

36
suggests that statistical independence of TF samples cannot be assumed in this stand if a
stationary gauge approach is used (see Loescher et al., 2002; Carlyle-Moses et al., 2014).

CONCLUSION
The present study has found that TF is modified by the canopies of juvenile lodgepole pine in
a systematic fashion, with this understory precipitation input typically increasing and its
associated spatial heterogeneity decreasing with increasing distance from the tree bole. Tree
and storm characteristics, such as tree size and direction of storm origin, were found to
explain some of the variability in the spatial delivery of TF, however, the role of other
variables that are likely drivers of this heterogeneity in this forest type, including wind speed
and direction as well as wood cover, still need to be assessed.
Our results suggest that TF sampling in these juvenile forests be done with relatively
denser networks of TF gauges in the inner- and mid-canopy zones as compared with the
canopy-periphery and areas outside of the canopy due to the increased TF variability
associated with the former zones. Areas outside of the canopy, however, are influenced by
the canopies of juvenile pine, probably as a result of the effective catch of inclined rainfall,
and thus need to be sampled. Sampling strategies should also take the prevailing direction of
storm origin (or wind) into consideration. Given the degree of temporal persistence observed,
it is recommended that TF gauges be periodically moved to different areas of the canopy so
that assumptions of statistical independence are met.
The systematic nature of the spatial heterogeneity of TF and its temporal persistence
suggests that directly or indirectly TF influenced hydrological, biogeochemical, and
ecological aspects of these young forests may also have associated spatiotemporal trends.
Further study to establish if such linkages exist is required. It is important to note that this
study was conducted during the growing-season when rainfall dominates. Hydrological,
biogeochemical and ecological processes may, however, be greatly influenced by the
distribution of snow during the dormant season and the subsequent spatial patterns associated
with snow melt. Thus, future studies should take the initial soil moisture pattern and
subsequent changes to that pattern as a consequence of drainage and evapotranspiration into

37
account when assessing the influence of growing-season TF spatial patterns on hydroecological processes Additionally, spatiotemporal patterns in TF and associated hydroecological processes, if they exist, should be assessed at different stages of growth to
determine the longevity of TF spatial patterns (i.e., to establish if the same TF spatial patterns
persist over several growing seasons) and so that the impact of forest disturbance regimes,
such as mountain pine beetle, and subsequent re-growth may be better understood.

REFERENCES
Aussenac G. 1970. Effect of forest cover on the ground distribution of rainfall. Annales des
Sciences Forestières 27: 383-399.
BC Ministry of Forests and Range. 2008. Biogeoclimatic Ecosystem Classification
Subzone/Variant Map for the Kamloops Forest District, Southern Interior Forest
Region [map]. 1:250,000. Ministry of Forests and Range, Research Branch. Victoria,
BC.
Bhat S, Jacobs J, Bryant M. 2011. The chemical composition of rainfall and throughfall in
five forest communities: A case study in Fort Benning, Georgia. Water, Air, & Soil
Pollution 218: 323-332.
Bringfelt B, Härsmar P-O. 1974. Rainfall interception in a forest in the Velen hydrological
representative basin: Treatment of data from the summer and autumn of 1973. Nordic
Hydrology 5: 146-165
Beier C, Hansen K, Gundersen P.1993. Spatial variability of throughfall fluxes in a spruce
forest. Environmental Pollution 81: 257-267
Bouten W, Heimovaara TJ, Tiktak A. 1992. Spatial patterns of throughfall and soil water
dynamics in a Douglas fir stand. Water Resources Research 28: 3227-33
Carleton TJ, Kavanagh T. 1990. Influence of stand age and spatial location on throughfall
chemistry beneath black spruce. Canadian Journal of Forest Research 20: 1917-1925
Carlyle-Moses DE. 2004. A reply to R. Keim’s comment on “Measurement and modelling of
growing-season canopy water fluxes in a mature mixed deciduous forest stand,
southern Ontario, Canada”. Agricultural and Forest Meteorology 124: 281-284
Carlyle-Moses DE, Flores Laureano JS, Price AG. 2004. Throughfall and throughfall spatial
variability in Madrean oak forest communites of northeastern Mexico. Journal of
Hydrology 297: 124-135

38
Carlyle-Moses DE, Lishman CE, McKee AJ. 2014. A preliminary evaluation of throughfall
sampling techniques in a mature coniferous forest. Journal of Forestry Research 25:
407-413
Carlyle-Moses DE, Park AD, Cameron JL. 2010. Modelling rainfall interception loss in
forest restoration trials in Panama. Ecohydrology 3: 272-283
Carlyle-Moses DE, Price AG. 2007. Modelling canopy interception loss from a Madrean
pine-oak stand, Northeastern Mexico. Hydrological Processes 21: 2572-2580
Carlyle-Moses DE, Price AG. 2006. Growing-season stemflow production within a
deciduous forest of southern Ontario. Hydrological Processes 20: 3651-3663
Carlyle-Moses DE, Gash JHC. 2011. Rainfall Interception Loss by Forest Canopies. In:
Forest Hydrology and Biogeochemistry: Synthesis of Past Research and Future
Directions, Levia DF, Carlyle-Moses DE, Tanaka T (eds.). Ecological Series 216,
Springer-Verlag: Heidelberg, Germany; 407-423.
Cottam G, Curtis JT. 1956. The use of distance measures in phytosociological sampling.
Ecology 37: 451-460
David TS, Gash JHC, Valente F, Pereira JS, Ferreira MI, David JS. 2006. Rainfall
interception by an isolated evergreen oak tree in a Mediterranean savannah.
Hydrological Processes 20: 2713-2726
Fathizadeh O, Attarod P, Keim RF, Stein A, Amiri GZ, Darvishsefat AA. 2014. Spatial
heterogeneity and temporal stability of throughfall under individual Quercus brantii
trees. Hydrological Processes 28: 1124-1136
Ford ED, Deans JD. 1978. The effects of canopy structure on stemflow, throughfall and
interception loss in a young sitka spruce plantation. Journal of Applied Ecology 15:
905-917.
Gash JHC. 1979. An analytical model of rainfall interception in forests. Quarterly Journal of
the Royal Meteorological Society 105: 43-55
Gerrits AMJ, Pfister L, Savenije HHG. 2010. Spatial and temporal variability of canopy and
forest floor interception in a beech forest. Hydrological Processes 24: 3011-3025
Gómez JA, Vanderlinden K, Giráldez JV, Fereres E. 2002. Rainfall concentration under olive
trees. Agricultural Water Management 55: 53-70
Hansen K. 1995. In-canopy throughfall measurements in Norway spruce: Water flow and
consequences for ion fluxes. Water, Air and Soil Pollution 85: 2259-2264

39
Helvey JD, Patric JH. 1965. Canopy and litter interception of rainfall by hardwoods of
eastern United States. Water Resources Research 1: 193-206
Herwitz SR. 1987. Raindrop impact and water flow on the vegetative surfaces of trees and
the effects on stemflow and throughfall generation. Earth Surface Processes and
Landforms 12: 424-432
Herwitz SR. 1985. Interception storage capacities of tropical rainforest canopy trees. Journal
of Hydrology 77: 237-252
Herwitz SR, Slye RE. 1995. Three-dimensional modeling of canopy tree interception of
wind-driven rainfall. Journal of Hydrology 168: 205-226
Herwitz SR, Slye RE. 1992. Spatial variability in the interception of inclined rainfall by a
tropical rainforest canopy. Selbyana 13: 62-71
Holwerda F, Scatena FN, Bruijnzeel LA. 2006. Throughfall in a Puerto Rican lower montane
rain forest: A comparison of sampling strategies. Journal of Hydrology 327: 592-602
Ito A, Onda Y, Nanko K, Fukuyama T, Moriwaki H. 2008. Experimental study on spatial
distribution of throughfall under a Japanese cypress tree. Journal of Japan Society of
Hydrology and Water Resources 21: 273-284
Johnson RC. 1990. The interception, throughfall and stemflow in a forest in Highland
Scotland and the comparison with other upland forests in the U.K. Journal of
Hydrology 118: 281-287
Keim, RF, Skaugset, AE, Weiler, M. 2005. Temporal persistence of spatial patterns in
throughfall. Journal of Hydrology 314: 263–274
Kimmins, JP. 1973. Some statistical aspects of sampling throughfall precipitation in nutrient
cycling studies in British Columbian coastal forests. Ecology 54: 1008-1019.
King BP, Harrison SJ. 1998. Throughfall patterns under an isolated oak tree. Weather 53:
111–121
Konishi S, Tani M, Kosugi Y, Takanashi S, Sahat MM, Nik AR, Niiyama K, Okuda T. 2006.
Characteristics of spatial distribution of throughfall in a lowland tropical rainforest,
Peninsular, Malaysia. Forest Ecology and Management 224: 19-25
Laws JO, Parsons DA. 1943. The relation of raindrop-size to intensity. Transactions of the
American Geophysical Union 24: 452-460
Levia DF, Keim RF, Carlyle-Moses DE, Frost EE. 2011. Throughfall and stemflow in
wooded ecosystems. In: Forest Hydrology and Biogeochemistry: Synthesis of Past

40
Research and Future Directions, Levia DF, Carlyle-Moses DE, Tanaka T (eds.).
Ecological Series 216, Springer-Verlag: Heidelberg, Germany; 425-444.
Levy KJ. 1979. Pairwise comparisons associated with the K independent sample median test.
The American Statistician 33: 138-139
Lin T-C, Hamburg SP, King H-B, Hsia Y-J. 1997. Spatial variability of throughfall in a
subtropical rainforest in Taiwan. Journal of Environmental Quality 26: 172-180
Llorens P, Gallart F. 2000. A simplified method for forest water storage capacity
measurement. Journal of Hydrology 240: 131-144
Lloyd CR, Marques FA de O. 1988. Spatial variability of throughfall and stemflow
measurements in Amazonian rainforest. Agricultural and Forest Meteorology 42: 6373
Loescher HW, Powers JS, Oberbauer SF. 2002. Spatial variation of throughfall volume in an
old-growth tropical wet forest, Costa Rica. Journal of Tropical Ecology 18: 397-407
Loustau, D, Berbigier, P, Granier, A, El Hadj Moussa, F. 1992. Interception loss, throughfall
and stemflow in a maritime pine stand, II. An application of Gash’s analytical model
of interception. Journal of Hydrology 138: 469–485
Macinnis-Ng CMO, Flores EE Muller H, Schwendenmann L. 2012. Rainfall portioning into
throughfall and stemflow and associated nutrient fluxes: land use impacts in a lower
montane tropical region of Panama. Biogeochemistry 111: 661-67
Manderscheid B, Matzner E. 1995. Spatial heterogeneity of soil solution chemistry in a
mature Norway spruce (Picea abies (L.) Karst.) stand. Water, Air and Soil Pollution 85:
1185-1190
Mann HB, Whitney DR. 1947. On a test of whether one or two random variables is
stochastically larger than the other. Annals of Mathematical Statistics 18: 50-60
Marin CT, Bouten W. Savink J. 2000. Gross rainfall and its partitioning into throughfall,
stemflow and evaporation of intercepted water in four forest ecosystems in western
Amazonia. Journal of Hydrology 237: 40-57
McKee AJ. 2010. The Quantitative Importance of Stemflow: An Evaluation of Past Research
and Results from a Study in Lodgepole Pine (Pinus contorta var. latifolia) Stands in
Southern British Columbia. M.Sc.Env.Sc. Thesis, Graduate Program in Environmental
Science. 113.
Mood AM. 1950. Introduction to the Theory of Statistics. McGraw Hill: New York; 433.

41
Nanko K, Onda Y, Ito A, Moriwaki H. 2011. Spatial variability of throughfall under a single
tree: Experimental study of rainfall amount, raindrops, and kinetic energy.
Agricultural and Forest Meteorology 151: 1173-1182
Návar J. 2013. The performance of the reformulated Gash’s interception loss model in
Mexico’s northeastern temperate forests. Hydrological Processes 27: 1623-1633
Parker, G.1983. Throughfall and stemflow in forest nutrient cycle. Advances in
Ecological Research 13, 57-103
Pedersen LB, 1992. Throughfall chemistry of Sitka spruce stands as influenced by tree
spacing. Scandinavian Journal of Forest Research 7: 433–444
Poppenborg P, Hölscher D. 2009. The influence of emergent trees on rainfall distribution in a
cacao agroforest (Sulawesi, Indonesia). Flora – Morphology, Distribution, Functional
Ecology of Plants 204: 730-736
Price AG, Dunham K, Carleton T, Band L. 1997. Variability of water fluxes through the
black spruce (Picea mariana) canopy and feather moss (Pleurozium schreberi) carpet in
the boreal forest of northern Manitoba. Journal of Hydrology 196: 310–323
Price AG, Carlyle-Moses DE. 2003. Measurement and modelling of growing-season canopy
water fluxes in mature mixed deciduous forest stand, southern Ontario, Canada.
Agricultural and Forest Meteorology 119: 69-85
Puckett LJ. 1991. Spatial variability and collector requirements for sampling throughfall
volume and chemistry under a mixed-hardwood canopy. Canadian Journal of Forest
Research 21: 1581-1588
Pypker TG, Levia DF, Staelens J, Van Stan II JT. 2011. Canopy Structure in Relation to
Hydrological and Biogeochemical Fluxes. In: Forest Hydrology and Biogeochemistry:
Synthesis of Past Research and Future Directions, Levia DF, Carlyle-Moses DE,
Tanaka T (eds.). Ecological Series 216, Springer-Verlag: Heidelberg, Germany; 371388.
Raat KJ, Draaijers GPJ, Schaap MG, Tietema A, Verstraten JM, 2002. Spatial variability of
throughfall water and chemistry and forest floor water content in a Douglas fir forest
stand. Hydrology and Earth System Science 6: 363-374
Reid LM, Lewis J. 2009. Rates, timing, and mechanisms of rainfall interception loss in a
coastal redwood forest. Journal of Hydrology 375: 459-470
Rice AV, Thormann MN, Langor DW. 2007. Mountain pine beetle associated blue-stain
fungi cause lesions on jack pine, lodgepole pine, and lodgepole x jack pine hybrids in
Alberta. Canadian Journal of Botany 85: 307-315

42
Ritter A, Regalado CM. 2014. Roving revisited, towards an optimum throughfall sampling
design. Hydrological Processes 28: 123-133
Robson AJ, Neal C, Ryland GP, Harrow M. 1994. Spatial variations in throughfall chemistry
at the small plot scale. Journal of Hydrology 158:107– 22
Rodrigo A, Àvila A. 2001. Influence of sampling size in the estimation of mean throughfall
in two Mediterranean holm oak forests. Journal of Hydrology 243: 216-227
Sakamoto Y, Ishiguro M, Kitagawa G. 1986. Akaike Information Criterion Statistics. Reidel:
Dordrecht, Netherlands; 290
Saito T, Matsuda H, Komatsu M, Xiang Y, Takahashi A, Shinohara Y, Otsuki K. 2013.
Forest canopy interception loss exceeds wet canopy evaporation in Japanese cypress
(Hinoki) and Japanese cedar (Sugi) plantations. Journal of Hydrology 507: 287-298
Sato AM, de Souza Velar A, Netto, ALC.2011. Spatial variability and temporal stability of
throughfall in a eucalyptus plantation in the hilly lowlands of southern Brazil.
Hydrological Processes 25: 1910-1923
Shapiro SS, Wilk MB. 1965. An analysis of variance test for normality (complete samples).
Biometrika 52: 591-611
Spittlehouse DL. 1998. Rainfall interception in young and mature conifer forests in British
Columbia. Proceedings of the 23rd Conference on Agricultural and Forest
Meteorology, 2-6 November 1998, Albuquerque, New Mexico: American
Meteorological Society
Staelens J, Schrijver AD, Verheyen K, Verhoest N. 2006. Spatial variability and temporal
stability of throughfall water under a dominant beech (Fagus sylvatica L.) tree in
relationship to canopy cover. Journal of Hydrology 330: 651-662
Stout B, McMahon J. 1961. Throughfall variation under tree crowns. Journal of Geophysical
Research 66: 1839-1843
Uunila L, Guy B, Pike R. 2006. Hydrologic effects of mountain pine beetle in the interior
pine forests of British Columbia: Key questions and current knowledge. Journal of
Ecosystems and Management 7: 37-40
Vachaud G, Passerat De Silans P, Balabanis P, Vauclin M. 1985. Temporal stability of
spatially measured soil water probability function. Soil Science Society of America
Journal 49: 822-828
Van Stan II JT, Stiegert CM, Levia DF, Scheick CE. 2011. Effects of wind-driven rainfall on
stemflow generation between codominant tree species with differing crown
characteristics. Agricultural and Forest Meteorology 151: 1277-1286

43

Vrugt JA, Dekker SC, Bouten W. 2003. Identification of rainfall interception model
parameters from measurements of throughfall and forest canopy storage. Water
Resources Research 39: 1259
Wang, T, Hamann A, Spittlehouse DL, Aitken SN. 2006. Development of scale-free climate
data for Western Canada for use in resource management. International Journal of
Climatology 26: 383-397.
Wang T, Hamann A, Spittlehouse DL, Murdock QT. 2012. ClimateWNA—High-resolution
spatial climate data for Western North America. Journal of Applied Meteorology and
Climatology 51: 16-29
Wullaert H, Pohlert T, Boy J, Valarezo C, Wilcke W. 2009. Spatial throughfall heterogenity
in a montane rain forest in Ecuador: Extent, temporal stability and drivers. Journal of
Hydrology 377: 71-79
Zar JH. 1984. Biostatistical Analysis (2nd Edition). Prentice hall: Englewood Cliffs, New
Jersey; 718
Ziegler AD, Giambelluca TW, Nullet MA, Sutherland RA, Tanatasarin C, Vogler JB,
Negishi JN. 2009. Throughfall in an ever-green dominated forest stand in northern
Thailand: Comparison of mobile and stationary methods. Agricultural and Forest
Meteorology 149: 373-384
Zimmermann A, Germer S, Neill C, Krusche AV, Elsenbeer H. 2008. Spatio-temporal
patterns of throughfall and solute deposition in an open tropical rainforest. Journal of
Hydrology 360: 87-102
Zimmermann A, Wilcke W, Elsenbeer H. 2007. Spatial and temporal patterns of throughfall
quantity and quality in a tropical montane forest in Ecuador. Journal of Hydrology 343:
80-96
Zimmermann A, Zimmermann B. 2014. Requirements for throughfall monitoring: The roles
of temporal scale and canopy complexity. Agricultural and Forest Meteorology 189190: 125-139
Zimmermann B, Zimmermann A., Elsenbeer H. 2009. Rainfall redistribution in a tropical
forest: Spatial and temporal patterns. Water Resources Research 45: W011413
Zimmermann B, Zimmermann A., Lark R.M, Elsenbeer H. 2010. Sampling procedures for
throughfall monitoring: A simulation study. Water Resources Research 46: W01503.
DOI: 10.1029/2009WR007776.
Zirlewagen D, von Wilpert K. 2001. Modeling water and ion fluxes in a highly structured,
mixed-species stand. Forest Ecology and Management 143: 27-37

44
CHAPTER 3
ANALYTICAL MODELLING OF CANOPY INTERCEPTION LOSS FROM A
JUVENILE LODGEPOLE PINE (Pinus contorta var. latifolia ) STAND
INTRODUCTION
In the central interior of British Columbia (BC), Canada, the mountain pine beetle
(Dendroctonus ponderosae Hopkins) (MPB) has severely affected the majority of pine
species in the region, but none more than the lodgepole pine (Pinus contorta Douglas ex
Louden var. latifolia Engelm. ex S. Watson). By 2004, more than seven million hectares of
lodgepole pine forests were infested by MPB in BC (Rice et al., 2007). The loss of mature
lodgepole pine stands, including those lost to salvage logging, has resulted in an increase in
the number of juvenile pine stands in the interior of BC through planting and natural
regrowth. With this change from mature forests to juvenile forests at such a large spatial
scale, the water balance of impacted areas may be altered, although the magnitude of such
change is uncertain. For example, juvenile stands produce more stemflow - intercepted
rainfall that is diverted to the base of trees by flowing along branches and down the boles of
trees - than do mature stands (Spittlehouse, 1998; McKee and Carlyle-Moses, 2010). In
addition, with the decrease in canopy cover, throughfall - rainfall that either passes directly
through canopy gaps or as canopy drip -will likely be greater in juvenile stands compared to
their mature counterparts. Given the scale of the MPB infestation, there is a need to quantify
the hydrology of these juvenile stands and to simulate various components of the hydrologic
cycle, including throughfall, stemflow, and canopy interception loss (Ic) - the proportion of
rainfall that is intercepted, stored and subsequently evaporated from the forest canopy.
Canopy interception loss, Ic, from mature lodgepole pine forests or stands that
comprise lodgepole pine with other species is appreciable, accounting for 24 – 29% of
season-long rainfall (Spittlehouse, 1998; Moore et al., 2008), falling in the mid to upper
range of the span of 10 – 50% for forest canopies across the globe reported by Roth et al.
(2007). Given the relatively sparse canopy and low stand stature, and thus decreased
aerodynamic conductance of juvenile lodgepole pine stands compared to their mature
counterparts, Ic likely represents a smaller fraction of seasonal rainfall than from mature
forests. This difference of Ic is further supported by a large difference in leaf area index
(LAI). Coops et al., (2009) found in a 90-100 year old lodgepole stands in southeastern BC,

45
the LAI ranged from 2.83 to 8.0, whereas, in juvenile lodgepole pine (12 years old) near
Prince George resulted in a LAI of 1.5 (Dhar et al., 2015). However, in the absence of an Ic
study in a juvenile lodgepole pine environment the quantitative impact of the transition from
mature to juvenile stands on Ic cannot be expressed with any degree of certainty.
Since Ic has often found to be of a quantitative importance a number of simulation
models have been developed to model this component of landscape evaporation. The Gash
analytical model (Gash, 1979) and its subsequent revisions (Gash et al. 1995; Valente et al.
1997) is the most commonly applied Ic model (Muzylo et al. 2009; Carlyle-Moses and Gash
2011). The Gash model represents a compromise between the overly simplistic linear
regression approach (e.g., Helvey and Patric 1965) that is only representative of the forest
community from which measurements were made, and the data-demanding, physically-based
numerical model of Rutter et al. (1971; 1975). A review of the Ic literature suggests that 16
studies have been conducted to date that have evaluated the Gash model or its reformulations
in coniferous forests. However, the Gash model has yet to be applied to lodgepole pine
forests and none of the 16 studies assessed the model under juvenile coniferous conditions.
The Gash analytical model follows the ‘water box approach’ in which drainage is
assumed to occur only when the canopy storage capacity, Sc has been satisfied. Once Sc has
been exceeded additional rainfall is partitioned into during-event evaporation, canopy drip
throughfall, or is directed to the boles of the trees where it may become stored, evaporated
during rainfall, or reach the forest floor as stemflow. Canopy storage is assumed to remain
constant once a certain threshold rainfall depth has been reached and does not increase with
additional rainfall input. However, the physical basis of the ‘water box approach’ has been
questioned (e.g., Klaasen et al. 1998), with many interception loss studies (e.g., Aston 1979;
Calder 1990) suggesting that Sc increases in an exponential fashion with increasing rainfall.
The gradual wetting of the canopy, through processes such as the relatively slow saturation
of the underside of leaves compared with the upper-sides and raindrops splashing onto
already wetted components of the canopy, will be referred to as the ‘exponential wetting
approach’ for the remainder of this thesis. Interception loss models that follow the
‘exponential wetting approach’ permit drainage (throughfall and stemflow) from unsaturated
canopies, unlike the ‘water box approach’. Liu (1997) derived an analytical exponential
wetting model which, like the Gash model, has been found to generate Ic values that are in

46
good agreement with observed values in a number of vegetation communities (see Liu, 2001;
Carlyle-Moses and Price 2007). The Liu model was reformulated by Carlyle-Moses and
Price (2007) to simulate sparse-canopy Ic in much the same way Gash et al. (1995) and
Valente et al. (1997) reformulated the Gash model for the same purpose.
The objectives of this study, were to i) determine the quantitative importance of
throughfall, stemflow and Ic, ii) evaluate the performance of the reformulated Gash (Valente
et al. 1997) and reformulated Liu (Carlyle-Moses and Price, 2007) Ic models at both event
and season-long time scales, and iii) assess the relative importance of during rainfall event
evaporation and storage as components of canopy interception loss from these young forests.

MATERIALS AND METHODS
Research Site
Research was conducted at the Mayson Lake hydrological research area, which is located on
the Thompson-Bonaparte plateau approximately 60 km NNW of Kamloops, British
Columbia, Canada ( 51°12'49” N 120°23'43" W, 1290 m a.m.s.l.) (Figures 3.1 and 3. 2). The
plot selected for this study was harvested in 1999 and planted with lodgepole pine seedlings
within the following 2 years. Point quarter analysis revealed that in 2010 the plot (2600 m2)
was dominated by lodgepole pine with 5640 stems ha-1, and a basal area of 9.6 m 2 ha-1, while
subalpine fir represented 980 stems ha-1, and a basal area of 0.6 m 2 ha-1. No other tree species
occurred in the plot. Average lodgepole pine tree height was ~ 2.2 m (median ~1.7 m), with a
maximum height of 5.2 m, while the average projected crown area (PCA) of these pines was
~ 0.9 m2 (median ~ 0.5 m2), with a maximum of 3.2 m2. The total PCA of the plot was ~
4940 m2 ha-1.
The study area has a mean annual precipitation depth of ~ 559 mm and temperature
of 2.7 oC, with mean monthly temperatures ranging from -7.2 to 14.8 oC (Winkler et al.,
2010). Rain is the dominate form of precipitation from mid-April through late September,
with snow dominating the rest of the year (Figure 3.3). The Köppen climate classification for
the region is Dfc (sub- arctic) while the provincial Biogeoclimatic Ecosystem Classification
designation is Montane Spruce (dm3: North Thompson, dry mild variant) (B.C. Ministry of
Forests and Range, 2008).

47

Figure 3.1: Geographic location of study plot.

48

Study Plot

100

20

80

5

60

-10

40

-25

20

-40

0

Temperature (0C)

Precipitation (mm)

Figure 3.2: Aerial view of the study plot. Photo Credit: R.D. Winkler.

-55
J

F

M

A

M

J

J

A

S

O

N

D

Figure 3.3: Climograph of study area based on ClimateBC ver 4.72 results using study plot
latitude, longitude and elevation as inputs. Note: White bars = snow, black bars = rain.

49
Rainfall, Throughfall and Stemflow Measurement
Rainfall duration and intensity data were measured on an event basis from May 16 –
September 30, 2010 using an Onset® tipping bucket rain gauge (model # S-RGB-M002) with
a diameter of 15.4 cm and a resolution of 0.2 mm tip-1. Additionally, rainfall depth was
measured with a Meteorological Service of Canada Type-B rain gauge (diameter = 11.3 cm).
The tipping bucket and the Type-B gauge openings were situated at a height of 1 m above the
ground surface, located ~ 40 m south-west of the site (Figure 3.4). For this study a rain event
was defined as a period of rainfall bounded by periods of 8 hours with no measurable rainfall
as indicted by the tipping bucket rain gauge or evidence of rainfall from radar imagery
compiled by Environment Canada’s National Climate Data and Information Achieve1, as this
was the observed maximum time required for the boles and canopies to dry.

Figure 3.4: The Onset® tipping bucket rain gauge (left) and Meteorological Service of
Canada Type-B rain gauge (right) used in this study

1

See historical radar imagery at:
http://www.climate.weatheroffice.gc.ca/radar/index_e.html?RadarSite=XSS&sYear=2013&sMonth=4&sDay=1
&sHour=11&sMin=00&Duration=2&ImageType=Default

50
Throughfall was measured on an event basis under the canopies of nine trees whose
characteristics are provided in Table 2.1. For each tree 16 manually-read Tru-Chek® wedge
gauges (catch area = 36.3 cm2 each) were used, with four of these gauges placed along four
transects (N, W, S, E). Along each transect one gauge was placed adjacent to the tree base,
one at the mid-canopy point, one at the canopy edge and the fourth gauge situated at a
random distance in the open area between tree canopies, but within 45˚ of the top of the
selected tree (Figure 3.5).

Figure 3.5: Throughfall gauge layout schematic. Shaded green = tree canopy, brown circle =
tree bole, and blue squares = throughfall gauges.
Throughfall below each tree was estimated by taking the area-weighted average of each
of the three below canopy zones (adjacent to tree, mid-canopy and canopy edge):
∑

̅̅̅̅̅

(3.1)

51
where, TFT is total below canopy throughfall (mm), A is the total area (m2) beneath the tree
canopy, az is the area of the below canopy zone (m2), and ̅̅̅̅̅ is the mean throughfall catch
(mm) within the below canopy zone.
Plot-scale below canopy throughfall at the rainfall event scale was estimated using one
of two methods depending if there was a significant (p < 0.10) relationship between TFT and
tree size (basal area). For those events in which a relationship between TFT and basal area
was found, plot-scale below canopy throughfall was estimated by applying this relationship
to the tree basal area frequency distribution collected as part of the point quarter exercise. For
rainfall events where no significant relationship between TFT and basal area could be found,
mean below canopy throughfall for the plot was simply estimated as the average below
canopy throughfall of the nine study trees. Finally, plot-scale throughfall was found by taking
the sum of the area weighted plot-scale below canopy throughfall and the area weighted open
area throughfall.
Stemflow was measured using the bottom of the two liter pop bottle to create a plastic
collar, which we then sealed to the base of the tree with silicone. A small ~1cm diameter
tube was inserted between the stemflow collar and the base of the tree to transfer the
stemflow to either a tipping bucket rain gauge or a four liter storage device. Collected
stemflow was measured after each event using a graduated cylinder. All stemflow collars
were tested on a weekly basis and repaired if needed. Plot-scale stemflow at the rainfall
event scale was estimated using the approach used to scale throughfall; with plot-scale
stemflow found by applying a linear regression of tree-scale stemflow versus tree basal area
for events in which a significant (p < 0.10) relationship between stemflow and tree basal area
was found, and by taking the average stemflow of the sampled trees for events in which no
significant relationship between stemflow and tree size was found. Figure 3.6 shows a
stemflow collar + reservoir alongside the throughfall gauges for one of the study trees.
Plot-scale Ic was derived from the measured rainfall and calculated plot-scale
throughfall and stemflow values using:

(3.2)

52

Figure 3.6: Stemflow and throughfall collection system for one of the selected study trees.
where,

is incident precipitation (mm) falling on the canopy and TF and SF are the plot-

scale throughfall and stemflow inputs (mm) below the canopy.

Modeling Procedure
Rainfall interception loss, throughfall and stemflow were modeled using the reformulated
Gash model (Valente et al. 1997) and reformulated Liu model (Carlyle-Moses and Price
2007). The Gash model is an analytical simulation tool that combines the conceptual
framework of the sparse Rutter model (see Valente et al. 1997), including the division of a
plot area into covered and uncovered sub-areas, with the ease of empirical equations. The
sparse Gash model, like the original, considers that interception loss follows three phases for

53
discrete rainfall events: 1) the wetting-up of the canopy phase, 2) the saturated canopy phase,
and 3) the canopy drying phase.
The various components of Ic as well as throughfall and stemflow were derived using
the sparse Gash model and their formulations are provided in Table 3.1. The parameter
values for the Gash model were derived following the methods outlined in Valente et al.
(1997). The mean rainfall rate falling on a saturated canopy [ R(mm h-1)] was determined
from 36 event rainfalls during the study period. The mean evaporation rate from a saturated
canopy was determined by using Gash’s (1979) method:
E=aR

(3.3)

where, a is the slope of the canopy interception loss (mm), versus incident precipitation (>
0.5mm) linear regression.
Valente et al. (1997) use a modified envelope curve approach for determining the
canopy storage capacity [S (mm)]. In addition, Valente et al. (1997) determined the drainage
partitioning coefficient (Pd), as well as the trunk storage capacity [St (mm)] which are used
the reformulated Gash model. Parameters E, S, and St are scaled to the fractional cover c: Ec,
Sc, and Stc. The parameter constant ε, which relates the evaporation rate of trunks to the
saturated canopies, was estimated based on Valente et al. (1997) and Price and CarlyleMoses (2003).
The depth of rainfall required to satisfy the canopy storage capacity [(
trunk storage capacity [(

[

(mm)] and

(mm)] were determined using the following formulas:

]

(3.4)

(3.5)

Once all of the climatic and stand parameters have been determined the Valente et al.
(1997) modified sparse Gash model formula can be used to calculate the interception loss:

54
Table 3.1: Components of canopy interception loss, throughfall and stemflow in the sparse
Gash model. Adapted from Valente et al. (1997).
Component of Interception Loss

Formulation

Amount of incident

required to saturate the canopy (

Amount of incident

required to saturate the trunks (

̅

)

̅

]

̅

∑

)

and from the canopy for n storms large enough to saturate the
canopy (

̅

̅

)

Interception loss from the canopy for m small storms
insufficient to saturate the canopy (

̅

[

̅

̅

[

)

̅

∑(

)]

] ∑(

)

Interception loss from the trunks for q storms that saturate the
trunks (

)

and from n-q storms that do not saturate the trunks (

)

Stemflow

̅
̅

[

Throughfall

̅

[

̅

] ∑(

∑

)
̅

[
∑(

̅

]

)

Note: ̅ is the mean rainfall intensity (mm h-1) falling on a saturated canopy, ̅ is the average evaporation rate
of intercepted rainfall (mm h-1) from a saturated canopy on a per unit area of canopy basis, S is the canopy
storage capacity (mm), c is the canopy cover fraction, is the ratio of the evaporation rate from the trunks to
that of the canopy, St is the storage capacity of the trunks (mm), and
represents the trunk partitioning
coefficient.

∑

∑

[ {
{

}

}]

∑

(3.6)

where, m, n and q represent the number of storms that do not saturate the canopy, that do
saturate the canopy and that saturate the trunks, respectively.
The version of the Liu (1997) model used in this study was the analytical single-storm
form since the canopy was assumed to be completely dry before each rainfall event. The

55
cumulative interception loss estimate is given as the sum of the single-storm estimates. The
single-storm form of the Liu (1997) model is:

[

(

)

]

[

(

)]

(3.7)

where Cm is the stand storage capacity (mm), including the storage capacities of the canopy
and trunks (i.e. Sc + St), p is the free throughfall coefficient, which represents the proportion
of throughfall that reaches the forest floor by passing directly through the canopy and is
regularly represented as p = 1- c (Valente et al. 1997) .
Assuming p = 1- c, Eq. 6 can be rewritten for simulating Ic from sparse forests using
the nomenclature of the reformulated Gash model (Carlyle-Moses and Price 2007):

[

[

(

)

]

[

( )]

]

(3.8)

where, Cmc is the storage capacity per unit area of the canopy + trunks (i.e. Sc + St).
Statistical Analysis and Model Evaluation
Statistical analysis and modeling was performed and analyzed using Microsoft® Office Excel
2007 (Microsoft Corporation, Redmond, WA, USA) and Smith’s Statistical Package version
2.80. Excel 2007 was used for graphing and data organization and manipulation, while
Smith’s Statistical Package was used for linear regression analysis. A p < 0.05 level of
statistical significance was used in this study. Model performance was assessed using
coefficients of determination of observed versus simulated Ic values and the root mean square
(RMSE) statistic (Mayer and Butler 1993; Carlyle-Moses et al. 2010):

√

∑

where, Ico and Icm represent the ith pair of observed and modeled Ic (mm) values,
respectively.

(3.9)

56
The performance of the interception models were also assessed using the NashSutcliffe criterion, which represents a measure of the variance of the observed data explained
by the model (Ladson 2008):
̅

∑
∑

(3.10)

̅

where, C is the Nash-Sutcliffe criterion or coefficient of efficiency, O and M are the observed
and modeled Ic, respectively, and ̅ is the mean observed Ic.

RESULTS
Plot-Scale Canopy Water Balance
During the study period precipitation fell in the form of rainfall and as mixed rain-snow
events. However, only rainfall events were considered for this study (n = 39). Data for one
of these events was not collected since the instrumentation at the plot was temporarily
removed due to the close proximity of a forest wildfire. For the remaining 38 events the
cumulative rainfall was 252.9 mm (mean = 6.7 mm, median = 5.4 mm), with individual
events ranging from 0.5 to 30.8 mm. Forty-seven percent of rainfall events had associated
depths < 5.0 mm, while those events with depths > 10.0 mm accounted for 18% of events.
Although a relatively small fraction of the number of events, large storms (> 10 mm)
accounted for ~ 52% of the cumulative rainfall. Throughfall, stemflow and Ic, totaled 221.9,
4.5, and 26.5 mm accounting for 87.7, 1.8 and 10.5 percent of the rainfall, respectively.
The relationships between event rainfall depth (mm) and the depths (mm) associated with
each of the canopy water fluxes were found to be linear (Figure 3.7a and 3.7b):

n = 38

(3.11)
(3.12)
(3.13)

where, TF and SF are throughfall (mm) and stemflow (mm), respectively, while Pg represents
rainfall depth (mm).

57

a)

Rainfall (mm)

b)

Figure 3.7: a) Event throughfall and b) event stemflow (mm) ─●─ and interception loss
(mm) ---○--- versus rainfall (mm).

58
The proportion of rainfall that became throughfall and stemflow increased, while the
proportion of rainfall that became interception loss decreased with increasing rainfall depth
(Table 3.2).

Table 3.2: Percentage of rainfall partitioned into throughfall, stemflow and canopy
interception loss for four rainfall depth classes.
Rainfall Class (mm)
<1.5
1.6 – 4.9
5.0 – 9.9
>10

TF (%)
70.3
75.6
87.7
89.9

SF(%)
0.1
0.4
1.8
2.1

Ic (%)
29.6
24.0
10.5
8.0

Modeling Results
The derived parameter values for both the Gash and Liu models are provided in Table 3.3.
The Gash and Liu models estimated cumulative Ic at 24.7 and 24.6 mm (9.8% and 9.7% of
rainfall), respectively, compared with the observed 26.5 mm (10.5% of rainfall); an
underestimate of 1.8 and 1.9 mm or 6.8 and 7.2% of the observed Ic, respectively. No
significant difference (p < 0.0001) was found between the slope of the Liu versus Gash
modeled results and unity nor was there a significant difference (p < 0.05) between the
intercept and zero (Figure 3.8).

Table 3.3: Gash and Liu model parameter values derived for the study stand.
Parameter
Value
Description

Model

Sc
Ec
R
c
Stc
Pd
ε
Pg’
Pg”

Gash, Liu
Gash, Liu
Gash, Liu
Gash, Liu
Gash, Liu
Gash
Gash
Gash
Gash

0.36
0.45
1.35
0.265
0.00
0.10
0.023
0.43
0.43

Canopy storage capacity (mm)
Mean evaporation rate from a saturated canopy (mm h-1)
Mean rainfall intensity (mm h-1)
Canopy cover fraction (dimensionless)
Trunk storage capacity (mm)
Drainage portioning coefficient (dimensionless)
Trunk: Canopy evaporation constant (dimensionless)
Depth of gross precipitation required to fill canopy storage (mm)
Depth of gross precipitation required to fill trunk storage (mm)

Observed and modeled accumulative Ic (mm) using the modeling approaches versus
accumulative rainfall (mm) is provided in Figure 3.9. The cumulative RMSE and r2 values
for the Gash and Liu modeled event-scale Ic versus observed event-scale Ic were 0.155 and

59

Figure 3.8: Liu modeled interception loss (mm) versus Gash modeled interception loss
(mm).

Accumulative Rainfall (mm)

Figure 3.9: Modeled accumulative interception loss (mm) ---- and observed interception loss
(mm) ____ as a function of accumulative rainfall (mm).

60

0.160 mm and 0.586 and 0.587, respectively. The Nash-Sutcliffe criterion value for the Gash
model was found to be 0.487, while for the Liu model it was 0.471.
An analysis was completed to determine the Gash and Liu model’s sensitivity to
changes in parameter values by recording the ratio between the model output compared to the
original estimate as a function of the change to the parameter (%) (Figure 3.10).

DISCUSSION AND CONCLUSION
Although the fraction of season-long rainfall portioned into Ic in this juvenile lodgepole pine
stand, 10.5%, is similar to that found in other environments where isolated trees dominate
(e.g., David et al. 2006), it is considerably less than that found for mature lodgepole pine
stands and lodgepole pine dominated stands in the region. Carlyle-Moses et al. (2014), for
example, found that 41% of the rainfall associated with 14 events (total depth = 50.1 mm)
during the growing season of 2008 in a mature (125 year-old) declining hybrid white spruce
(Picea glauca (Moench) Voss × P. engelmannii Parry ex Engelm.), subalpine fir (Abies
lasiocarpa (Hook.) Nutt.), and lodgepole pine (Pinus contorta var. latifolia Dougl. ex Loud.)
forest located approximately 0.85 km NW of the study plot was partitioned into Ic.
Spittlehouse (1998) found that Ic accounted for 24% of a May to October storm record
totaling 454 mm in a mature lodgepole pine dominated forest near Penticton, British
Columbia. However, that the magnitude and frequency of rain events play a large role in
determining the percentage of rainfall that is partitioned into Ic (Carlyle-Moses and Gash
2011). Using the linear regression relating Ic depth to rainfall depth for the mature spruce-firpine stand derived by Carlyle-Moses et al. (2014) and the rain record for this current study,
approximately 85.3 mm, or 33.7% of the season-long rainfall would have been partitioned
into Ic by the mature forest, representing an Ic efficiency approximately 3 times that of the
juvenile stand.
Intuitively, the lower Ic efficiency of the juvenile forest compared to that of the
mature forest is due, in no small part, to the relatively lower canopy cover fraction, 26.5%
compared to 58%. However, the composition of that cover, including the fraction of wood
relative to foliage, difference in LAI, the canopy depth, and the presence and abundance of
epiphytes, are also likely reasons for the differences in the fractioning of rain into Ic by the

61

a

b

Cmc

Figure 3.10: Sensitivity of the reformulated Gash (a) and Liu (b) models to changes in model
parameter values.

62
juvenile and mature stands (Spittlehouse, 1998). Carlyle-Moses et al. (2007) derived a
storage capacity ranging from 2.06 to 3.22 mm on a per unit canopy area basis for the
aforementioned mature lodgepole pine dominated stand near Penticton, British Columbia, a
full order of magnitude greater than the Sc value derived for the juvenile stand in this current
study (0.36 mm). Other studies in mature coniferous forests have also found relatively large
Sc values, including 4.47 mm for a Pinus armandii in the Ningxia Hui Autonomous Region
of China.(Zhongjie et al. 2010)
The mean during-event evaporation rate of the juvenile pines (0.45 mm h-1) is well
within the range (0.07 to 0.70 mm h-1) reported from other forests (Carlyle-Moses and Price,
2007); however, the juvenile pine Ec value exceeds, albeit only slightly, the bounds of the
mean ± standard deviation of 0.3 ± 0.1 mm h-1 derived by Miralles et al. (2010), which is
based on the results of twelve studies conducted using data from forests around the globe.
Additionally, Ec for the juvenile stand is within the range derived for the mature lodgepole
pine forest near Penticton, British Columbia, 0.44 to 0.77 mm h-1 (Carlyle-Moses et al.
2007). Thus, the Ec rate found for the juvenile pine stand is relatively high, considering the
relative short stature of these trees, compared to the mature stand; a possible consequence of
the increased ventilation offered by these relatively isolated trees (Carlyle-Moses et al. 2010;
Teklehaimanot et al. 1991). However, as David et al. (2006) note, stand density or canopy
cover may not affect the rate of Ic on a per unit of crown-projected area basis and that the
reformulated Gash model (and Liu model) assumes that E decreases linearly with decreasing
canopy cover. Nonetheless, Ic is relatively high when expressed on a per canopy cover
fraction basis (39.6%) and is due in no small part to the relatively large Ec value derived for
this stand.
Very little difference was found between the performance of the reformulated Gash
and reformulated Liu models, in keeping with the results of Carlyle-Moses and Price (2007)
in a mature pine-oak stand in northeastern Mexico. Error analysis for the current study
suggests that both the reformulated Gash and Liu models are particularly sensitive to
variations in R and Ec, and much less so for variations in Sc. Similar findings have been
found for other forest and plantation environments (e.g., Carlyle-Moses and Price 2007;
Carlyle-Moses et al. 2010; Muzylo et al. 2012). Since, for a given stand, Ec and the ratio of
Ec: R varies largely with wind speed and vapour pressure deficit (see Pereira et al., 2009;

63
Carlyle-Moses and Gash, 2011; Llorens, 1998), it is not surprising to find the relatively large
errors associated with the two models at the rainfall event scale with both the Gash and Liu
models not able to explain approximately 41% of the variation in observed Ic.
In summary, the current study suggests that Ic is more reduced in juvenile stands
compared to their mature counterparts and that this reduction is due to the decreased storage
capacity offered by these younger canopies. Evaporation during rainfall from juvenile
canopies is still appreciable and may be a consequence of the increased proportion of the
canopy exposed to wind during events. The partitioning of water by the canopies of these
juveniles stands also differs from mature forests in that stemflow represents a greater
proportion of total rainfall than it does in older stands.
Although the fraction of rainfall partitioned by the canopy is still small, representing
less than 2% over the study period, the concentrated input of this water (see Mckee and
Carlyle-Moses 2010) may have important implications for site hydrology that remain to be
explored. Additionally, the fate of the increased input of total canopy drainage (throughfall +
stemflow), which is approximately is estimated to be 15 to 20% of growing-season rainfall,
on the hydrology of these juvenile forests and how this compares with the water stores and
fluxes in mature stands also requires further study.
REFERENCES
Aston AR. 1979. Rainfall interception by eight small trees. Journal of Hydrology 42: 383–
396.
B.C. Ministry of Forests and Range. 2008. Biogeoclimatic ecosystem classification
subzone/variant map for the Arrow Boundary Forest District, Boundary Subunit,
Southern Interior Forest Region. 1:250 000. Research Branch, Victoria, B.C.
Calder, IR. 1990. Evaporation in the uplands. John Wiley & Sons Ltd., Chichester, 166 p.
Carlyle-Moses DE, Lishman CE, McKee AJ. 2014. A preliminary evaluation of throughfall
sampling techniques in a mature coniferous forest. Journal of Forestry Research 25:
407-413
Carlyle-Moses DE, Gash JHC. 2011. Rainfall Interception Loss by Forest Canopies. In:
Forest Hydrology and Biogeochemistry: Synthesis of Past Research and Future
Directions, Levia DF, Carlyle-Moses DE, Tanaka T (eds.). Ecological Series 216,
Springer-Verlag: Heidelberg, Germany; 407-423

64
Carlyle-Moses DE, Park AD, Cameron JL. 2010. Modelling rainfall interception loss in
forest restoration trials in Panama. Ecohydrology 3: 272-283
Carlyle-Moses DE, Price AG. 2007. Modelling canopy interception loss from a Madrean
pine-oak stand, Northeastern Mexico. Hydrological Processes 21: 2572-2580
Carlyle-Moses DE, Schimpf GR and DL Spittlehouse. 2007. “Influence of rainfall event
separation time on the analytical modelling of canopy interception loss from a mature
lodgepole pine (Pinus contorta var. latifolia) stand” In: Program and Abstracts of the
Joint Canadian Meteorological and Oceanographic Society, the Canadian Geophysical
Union, and the American Meteorological Society Annual Meetings, St. John’s,
Newfoundland, Canada, 28 May –1 June.
Coops, NC, Waring, RH, Wulder, MA, White, JC. 2009. Prediction and assessment of bark
beetle-induced mortality of lodgepole pine using estimates of stand vigour derived
from remotely sensed data. Remote Sensing of Environment. 113: 1058-1066
Dhar A, Wang JR, Hawkins CDB. 2015. Interaction of Trembling Aspen and Lodgepole Pine
in a Young Sub-Boreal Mixedwood Stand in Cnetral British Columbia. Open Journal
of Forestry 5: 129-138.
David TS, Gash JHC, Valente F, Pereira JS, Ferreira MI, David JS. 2006. Rainfall
interception by an isolated evergreen oak tree in a Mediterranean savannah.
Hydrological Processes 20: 2713-2726
Gash JHC. 1979. An analytical model of rainfall interception in forests. Quarterly Journal of
the Royal Meteorological Society 105: 43-55
Gash JHC, Lolyd CR, Lachaud G. 1995. Estimating sparse forest rainfall interception with an
analytical model. Journal of Hydrology 170: 79-86

Helvey JD, Patric JH. 1965. Canopy and litter interception of rainfall by hardwoods of
eastern United States. Water Resources Research 1: 193-206
Klaassen, W, Bosveld, F, Water, E (1998) Water storage and evaporation as constitutes of
rainfall interception. Journal of Hydrology 212–213: 36-50
Ladson A. 2008. Hydrology: An Australian introduction. Oxford University Press. New York
Liu S. 1997. A new model for the prediction of rainfall interception in forest canopies.
Ecological Modeling 99: 151-159
Liu S. 2001. Evaluation of the Liu model for predicting rainfall interception in forests worldwide. Hydrological Processes 15: 2341-2360

65
Llorens P. 1998. Rainfall interception by a Pinus sylvestris forest patch overgrown in a
Mediterranean mountainous abandoned area II. Assessment of the applicability of
Gash's analytical model. Journal of Hydrology 199: 346-359.
Mayer DG. And Butler DG. 1993. Statistical validation. Ecological Modelling 68: 21-32
McKee A ,Carlyle-Moses DE. 2010. Stemflow: A potentially important point source of
water for growth. FORREX Forum for Research and Extension in Natural
Resources:11-12
Miralles DG, Gash JH, Holmes TRH, de Jeu RAM and AJ Dolman. Global canopy
interception from satellite observations. Journal of Geophysical Research 115: Di6122
Moore, RD, Winkler, RD, Carlyle-Moses, DE. 2008. Watershed response to the McLure
forest fire: presentation summaries from the Fishtrap Creek workshop, March 2008.
Streamline Watershed Manage Bulletin 12: 1-11
Muzylo A, Llorens P., Valente F., Keizer J.J., Domingo D, and Gash J.H.C. 2009. A review
of rainfall interception modelling. Journal of Hydrology 370: 191–206
Pereira FL, Gash JHC, David JS, David TS, Monteiro PR and F Valente. 2009. Modelling
interception loss from evergreen oak Mediterranean savannas: Application of a treebased modelling approach. Agricultural and Forest Meteorology 149: 680-688
Price AG, Carlyle-Moses DE. 2003. Measurement and modelling of growing-season canopy
water fluxes in mature mixed deciduous forest stand, southern Ontario, Canada.
Agricultural and Forest Meteorology 119: 69-85
Rice AV, Thormann MN, Langor DW. 2007. Mountain pine beetle associated blue-stain
fungi cause lesions on jack pine, lodgepole pine, and lodgepole x jack pine hybrids in
Alberta. Canadian Journal of Botany 85: 307-315
Roth, BE, Slatton, KC, Cohen, MJ. 2007. On the potential for high-resolution lidar to
improve rainfall interception estimates in forest ecosystems. Frontiers in Ecology and
the Environment 5: 421-428
Rutter, AJ, Kershaw, KA, Robins, PC. 1971. A predictive model of rainfall interception in
forests I: derivation of the model from observations in a plantation of Corsican pine.
Agricultural Meteorology 9: 367-384
Rutter, A, Morton, A. Robin, P., 1975. A Predictive Model of Rainfall Interception in
Forests. II. Generalization of the Model and Comparison with Observations in Some
Coniferous and Hardwood Stands. Journal of Applied Ecology 12: 367–380
Spittlehouse DL. 1998. Rainfall interception in young and mature conifer forests in British
Columbia. Proceedings of the 23rd Conference on Agricultural and Forest

66
Meteorology, 2-6 November 1998, Albuquerque, New Mexico: American
Meteorological Society.
Teklehaimanot, Z, Jarvis, PG. 1991. Rainfall interception and boundary layer conductance in
relation to tree spacing. Journal of Hydrology 123: 261-278
Valente, F., David, J.S., Gash, J.H.C., 1997. Modelling interception loss for two sparse
eucalypt and pine forests in central Portugal using reformulated Rutter and Gash
analytical models. Journal of Hydrology 190: 141–162
Winkler RD, Boon S, Zimonick B, Baleshta K. 2010. Assessing the effects of post-pine
beetle forest litter on snow albedo. Hydrological Processes 24:803–812
Zhongjie S, Yanhui W, Lihong X, Pentao Y, Jixi G and Z Linbo. 2010. Fraction of incident
rainfall within the canopy of a pure stand of Pinus armandii with revised Gash model
in the Liupan Mountains of China. Journal of Hydrology 385: 44-50

67
CHAPTER 4
CONCLUSION
A juvenile lodgepole pine (Pinus contorta Douglas ex Louden var. latifolia Engelm. ex S.
Watson) stand in south-central British Columbia was examined, in order to determine the
magnitude of point and structural TF as well as SF at the rainfall event and growing-season
temporal scales. Chapter 2 evaluated if TF (both point and structural) and SF spatial
variability exhibits temporal persistence and if so to determine what the influence, if any,
certain tree and storm characteristics may have on this stability. The present study has found
that TF is modified by the canopies of juvenile lodgepole pine in a systematic fashion, with
this understory precipitation input typically increasing and its associated spatial heterogeneity
decreasing with increasing distance from the tree bole. Tree and storm characteristics, such
as tree size and direction of storm origin, were found to explain some of the variability in the
spatial delivery of TF, however, the role of other variables that are likely drivers of this
variability in this forest type, including wind speed and direction as well as wood cover, still
need to be assessed. The spatial variability of TF exhibits temporal persistence that
marginally degrades with time across the study period, probably as a result of slight changes
to the canopy during growth across the time span of the study (five months). Temporal
persistence was also found to degrade as a consequence of differences in the certain
meteorological characteristics, such as the 30-minute rain intensity, suggesting that alike
storms produce alike TF spatial patterns. However, it is not clear why differences in intrastorm break duration and or number of breaks among events resulted in a strengthening of
persistence. Nonetheless, much of the variability in the persistence of TF spatial patterns
from one event to the next in this study was not explained. Again, wind speed and direction
may be important factors that required future study.
Finally, Chapter 2 provided some recommendations for sampling strategies for future
research in similar forests environments. Our results suggest that future TF sampling in these
juvenile forests be done with relatively denser networks of TF gauges in the inner- and midcanopy zones as compared with the canopy-periphery and areas outside of the canopy due to
the increased TF variability observed in the former zones. Areas outside of the canopy,
however, are still influenced by the canopies of juvenile pine, probably as a result of the

68
effective catch of inclined rainfall, and thus still need to be sampled. The prevailing direction
of storm origin should also be considered when establishing TF sampling networks.
Chapter 3 dealt with interception modelling, specifically to evaluate the performance
of the reformulated Gash (Valente et al., 1997) and reformulated Liu (Carlyle-Moses and
Price, 2007) Ic models at both event and season-long time scales, as well as to determine the
quantitative importance of throughfall, stemflow and Ic. Furthermore, this chapter assessed if
the data collected supports a ‘water-box’ or an ‘exponential wetting approach’ of the canopy
saturation process. The current study suggests that Ic is more reduced in juvenile stands
compared to mature stands and that this reduction is due to the decreased storage capacity
offered by these younger canopies. Evaporation during rainfall from juvenile canopies is still
appreciable and may be a consequence of the increased proportion of the canopy exposed to
wind during events.

LIMITATIONS
The initially proposed thesis involved every aspect of forest hydrology, ranging from
measuring rainfall, throughfall (5472 TF points), stemflow (239 +516), soil moisture content
(4224 TDR points), meteorological station data, infiltration rates (96 minidisk infiltrometers
tests), macropore study using urinine tracer dye (15 tracer dye experiments) as well as
measuring groundwater levels (two monitoring wells). The sum of these measured forest
hydrology parameters resulted in tens of thousands of data points, which proved to be
overwhelming for the scope of the MSc. Thesis. All the data was collected for the above
mentioned parameters of forest hydrology, however only throughfall, stemflow and rainfall
data were analyzed during this study.
Due to the sheer scope of the research project, logistical and financial restraints
reduced the number of plots to create the dataset required for developing the interception
models, and to determine the magnitude of point and structural TF as well as SF at the
rainfall event and growing-season temporal scales. Additionally, TF (both point and
structural) spatial variability was evaluated to determine if it exhibited temporal persistence
at only one plot.

69
The inclusion of more samples from other similar aged tree stand locations would
have resulted in more comprehensive interception models which could be compared to one
another.

APPLICATION OF RESEARCH
One possible practicable application of Chapter 3 can be to help provide accurate
interception modelling data for the proposed BC Water Development Tool, which is based
off of the Oil and Gas Commission NorthEast Water Tool (NEWT, 2015). NEWT uses
gridded climate, topography and land cover information to develop a simple water balance
model (Q=P-ET). As these GIS tools evolve and become more in-depth, interception models
can be used in conjunction with forest land cover to provide more accurate and area specific
interception and evapotranspiration values. Additionally, interception loss and SF funnelling
ratios can be used to further our ability to predict flooding, better control of water allocation
for agricultural purposes and to provide more regulated surface water runoff control in
forested and cut-block areas.

FUTURE STUDY DIRECTIONS
Tree and storm characteristics, such as tree size and direction of storm origin, were found to
explain some of the variability in the spatial delivery of TF, however, the role of other
variables that are likely drivers of this heterogeneity in this forest type, including wind speed
and direction as well as wood cover, still needs to be addressed.
The systematic nature of the spatial heterogeneity of TF and its temporal persistence
suggests that directly or indirectly TF influenced hydrological, biogeochemical, and
ecological aspects of these young forests may also have associated spatiotemporal trends.
Further study to establish if such linkages exist is required. Additionally, spatiotemporal
patterns in TF and associated hydroecological processes, if they exist, should be assessed at
different stages of growth so that the impact of forest disturbance regimes, such as the
mountain pine beetle, and subsequent re-growth may be better understood.
The partitioning of water by the canopies of these juveniles stands also differs from mature
forests. Based on the results of this current study in a juvenile lodgepole pine forests and the
results of past studies in mature lodgepole pine forests, it may be expected that interception

70
loss will increase while throughfall and stemflow will decrease as the stand ages (Figure 4.1).
This is owed to an increase in storage capacity as the tree ages, primarily to the relatively
higher canopy cover fraction of the mature stand, as well as thicker and coarser bark (CarlyleMoses et al.2007). Because of the dependence of evaporation and storage capacity on

meteorological conditions, further study is needed, especially studies in which interception
loss, thoughfall and stemflow are measured over a chronosequence of stands of varying ages
under the same meteorological conditions so that the role of canopy characteristics can truly
be identified.

Juvenile Lodgepole
Pine

Mature Lodgepole
Pine

Rainfal
l

Rainfal
l

Ic

Sc

TF

Ic
SF

TF
Sc

SF

Figure 4.1: Conceptual model comparing a juvenile and mature lodgepole pine tree
rainfall partitioning.
Even though the fraction of rainfall partitioned by the canopy is small, representing
less than 2% over the study period, the concentrated input of this water (see McKee and

71
Carlyle-Moses 2010) may have important implications for site hydrology that remain to be
explored. Additionally, the fate of the increased input of total canopy drainage (throughfall +
stemflow), which is approximately 15 to 20% of growing-season rainfall, on the hydrology
of these juvenile forests and how this compares with the water stores and fluxes in mature
stands also requires further study.

REFERENCES
BC Oil and Gas Commission. 2015. Northeast Water Tool. Retrieved from
http://geoweb.bcogc.ca/apps/newt/newt.html
Carlyle-Moses DE, Price AG. 2007. Modelling canopy interception loss from a Madrean
pine-oak stand, Northeastern Mexico. Hydrological Processes 21: 2572-2580
Carlyle-Moses DE, Schimpf GR , DL Spittlehouse. 2007. “Influence of rainfall event
separation time on the analytical modelling of canopy interception loss from a mature
lodgepole pine (Pinus contorta var. latifolia) stand” In: Program and Abstracts of the
Joint Canadian Meteorological and Oceanographic Society, the Canadian Geophysical
Union, and the American Meteorological Society Annual Meetings, St. John’s,
Newfoundland, Canada, 28 May –1 June.
McKee A, Carlyle-Moses D. 2010. Stemflow: A potentially important point source of water
for growth. FORREX Forum for Research and Extension in Natural Resources: 11-12
Valente, F., David, JS, Gash, JHC. 1997. Modelling interception loss for two sparse
eucalyptus and pine forests in central Portugal using reformulated Rutter and Gash
analytical models. Journal of Hydrology 190: 141–162