Journal Pre-proofs
Research papers
Changes in Rainfall Partitioning and Canopy Interception Modeling after Pro‐
gressive Thinning in Two Shrub Plantations in the Semiarid Loess Plateau in
China
Xiaotao Niu, Jun Fan, Mengge Du, Zijun Dai, Ruihua Luo, Hongyou Yuan,
Shougang Zhang
PII: S0022-1694(23)00241-X
DOI: https://doi.org/10.1016/j.jhydrol.2023.129299
Reference: HYDROL 129299
To appear in: Journal of Hydrology
Received Date: 23 May 2022
Revised Date: 25 December 2022
Accepted Date: 15 February 2023
Please cite this article as: Niu, X., Fan, J., Du, M., Dai, Z., Luo, R., Yuan, H., Zhang, S., Changes in Rainfall
Partitioning and Canopy Interception Modeling after Progressive Thinning in Two Shrub Plantations in the
Semiarid Loess Plateau in China, Journal of Hydrology (2023), doi: https://doi.org/10.1016/j.jhydrol.
2023.129299
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
? 2023 Published by Elsevier B.V.
1
Changes in Rainfall Partitioning and Canopy Interception Modeling after Progressive Thinning
in Two Shrub Plantations in the Semiarid Loess Plateau in China
Xiaotao Niu Conceptualization Investigation Methodology Software Visualization
Writing - original draft Writing - review & editinga,b,c,d, Jun Fan Conceptualization Data
curation Formal analysis Funding acquisition Methodology Project administration
Supervision Validation Writing - original draft Writing - review & editinga,b,d,,
fanjun@ms.iswc.ac.cn, Mengge Du Formal analysis Writing - review & editingd, Zijun Dai
Methodology Writing - review & editingd, Ruihua Luo Validation Investigationd,
Hongyou Yuan Investigationd, Shougang Zhang Investigationd
aThe Research Center of Soil and Water Conservation and Ecological Environment, Chinese Academy
of Sciences and Ministry of Education, Yangling, Shaanxi 712100, China
bInstitute of Soil and Water Conservation, Chinese Academy of Sciences and Ministry of Water
Resources, Yangling, Shaanxi 712100, China
cUniversity of Chinese Academy of Sciences, Beijing 100049, China
dState Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Northwest A&F
University, Yangling, Shaanxi 712100, China
Corresponding author.
Highlights
2
A five-year experiment with two thinning intensities in two dense shrub plantations.
Effects of thinning on rainfall partitioning and canopy parameters were quantified.
The observed drought year did not have significant impact in the HT plots.
Revised Gash model performed an underestimated interception in two shrubs (RE＜ 20%).
Gash model performed better in thinning plots than control plots.
Abstract
Understanding the interaction between the hydrological cycle and vegetation management strategies,
such as thinning, is essential to improve watershed management and support ecological services.
However, it remains unclear how thinning affects the key components of hydrological cycle of dense
shrublands in the semiarid regions. Thus, the purpose of this study was to analyze and evaluate the
effect of thinning on rainfall partitioning and interception simulations in dense shrublands. In a 5-year
field experiment, we compared rainfall partitioning in two re-vegetated shrublands (Caragana
korshinskii and Salix psammophila) in the Chinese Loess Plateau under two thinning intensities
(moderate thinning [MT] and heavy thinning [HT] refer to the removal of 25% and 50% of the
branches, respectively) or no thinning (NT) (control). We modeled canopy interception losses (I) in the
two thinning treatments and control treatment using the revised Gash model. The results showed that
under MT and HT, the throughfall (TF) rate increased by about 12% and 20%, respectively, compared
to NT. The stemflow (SF) and observed I rates decreased by about 26% and 33%, respectively, under
MT, and the corresponding values for HT were about 50% and 52%, respectively. The observed I rate
decreased proportional to the percentage of biomass removed from the C. korshinskii and S.
psammophila plots. The results also revealed a significant linear correlation between the plant area
index (PAI) value and the canopy water balance of the two shrub plantations (R2>0.83, P<0.05). The
3
performance of the revised Gash model (i.e., relative error [RE] < 20%) was satisfactory according to
the Nash?Sutcliffe model efficiency (NSE) coefficient (0.34-0.71). Based on the RE values, the
performance of the revised Gash model was better when applied to the plots subjected to MT and HT
(RE≤3%) than NT (RE=9.3%) for S. psammophila. Changes in the canopy storage capacity and canopy
evaporation rate strongly affected changes in simulated interception loss. The model can facilitate
water management in semiarid shrub plantations by accurately simulating the effect of thinning on
interception loss.
Keywords: Rainfall partitioning, Interception, Revised Gash model, Thinning, Shrubland
1. Introduction
Vegetation restoration based on unreasonable density is a fact that currently exists in many arid and
semiarid areas (Cao et al., 2011; Farley et al., 2005; Feng et al., 2016; Molina & del Campo, 2012).
Highly dense vegetation consumes soil moisture and increases interception of limited precipitation in
arid and semiarid regions (Chen et al., 2010; Christina et al., 2017; Ge et al., 2022; Robinson et al.,
2006; Shao et al., 2018). As a result, insufficient water supply to supplement evapotranspiration, which
ultimately aggravates soil drought conditions and leads to the formation of a dry layer in the soil. The
conditions as mentioned above make the vegetation unsustainable, with vegetation decline (partial
death) or stunting, where trees remain small for decades (Jia et al., 2019; Navarro-Cerrillo et al., 2019;
Wang et al., 2010; Zhang et al., 2020). Under these conditions, the water conservation function of the
vegetation cover is extremely limited. Therefore, studying changes in key components of the
4
hydrological cycle after reducing vegetation density in water-limited regions is necessary to improve
watershed management and support ecological services.
In arid and semi-arid areas of the Loess Plateau, like the northern areas of the Qinghai-Tibetian Plateau
and the Greater Hinggan Mountains in China, The vegetation in these areas is dominated by shrubs
(Zhang et al., 2022). Due to the nature conservation policy of the local government, these shrubs have
seldom been managed since their established, resulting in unmanaged dense shrub canopies that
intercept more rainfall and consume more water. Drought-induced branch dieback has recently been a
frequent phenomenon in these plantations (Ma et al., 2020). This suggests that the current structural
characteristics of these shrub regions may not be conducive to alleviating regional drought, and these
areas need more rainfall to preserve and supplement local water resources. It is necessary to allow as
much rainfall as possible to reach the surface and be stored in the soil, rather than being lost to
evaporation through interception. As a result, shrub plantations could be reduced in density at a large
spatial scale to increase resilience in the face of severe drought stress and future climate change.
Thinning is a common vegetation management practice aimed at reducing the canopy density and
improving the quality of the remaining vegetation (Gebhardt et al., 2014; Grunicke et al., 2020;
Momiyama et al., 2021; Sun et al., 2015). In recent decades, experimental evidence from di?erent
forest ecosystems has demonstrated that thinning can promote the vitality of residual trees and reduce
long-term stress due to competition for water, nutrients, and light. Thinning can also increase the
resilience and resistance of forest trees to severe drought stress and thus may be an effective approach
to climate adaptation (Gebhardt et al., 2014; Grunicke et al., 2020; Navarro-Cerrillo et al., 2019; Wang
et al., 2019). In addition, thinning affects hydrological processes, including evaporation, transpiration,
5
rainfall partitioning, soil water content, and surface ?ow, in vegetated ecosystems (Gebhardt et al.,
2014; del Campo et al., 2019; Sun et al., 2014). Canopy interception, which accounts for 5?40% of
total rainfall, is an important part of the shrub water cycle (Yue et al., 2021). It determines the amount
of net rainfall reaching the soil and is a key component in the hydrological cycle. However, few studies
have explored practical strategies for managing highly dense shrublands, with most focusing on forests
(Grunicke et al., 2020; Sun et al., 2015).
In recent decades, many observational studies have focused on canopy interception of shrubs in arid
and semiarid areas at the individual plant scale (An et al., 2022; Domingo et al., 1998; Jian et al., 2018;
Li et al., 2008; Tonello et al., 2021; Yang et al., 2019a; Yuan et al., 2017; Zhang et al., 2015).
Extrapolating the results of these studies to heterogeneous landscapes is a challenge. In this context,
interception needs to be quantified at the stand or plot scale rather than at the individual plant scale due
to plot-level measurements of interception have the advantage that the results can be scaled up, for
example, using traditional cover methods (Snyder et al., 2021). Furthermore, plot-level experimental
results are generally applicable to a watershed, and they cannot be confidently applied where
conditions are markedly different, particularly regarding rainfall regimes and vegetation types. While
interception models allow the extrapolation of measurements in space and time and can be used to
predict the effects of climate and land cover change on water resources (Magliano et al., 2022).
Therefore, model simulations continue to be the main focus of vegetation canopy interception research.
Widely used models for estimating interception losses are the original Gash model and the revised
version of this model (Muzylo et al., 2009). The original Gash model, a simpli?ed version of the Rutter
model (Gash., 1979; Rutter et al., 1975). The Gash model represents rainfall input as a series of discrete
6
storms separated by intervals suf?ciently long for the canopy and stems to dry completely—this
assumption is possible because of the rapid drying of forest canopies. Each storm is then divided into
three subsequent phases—canopy wetting-up, saturation, and drying. This separation emphasizes the
relative importance of the climate against plant structure. Gash et al. (1995) proposed the sparse
versions of Gash model to adjust the original model formulations to forest stands with signi?cant open
spaces between the tree canopies. A crucial change is that in the sparse versions, the evaporation rate
from wetted surfaces is no longer calculated for the entire plot area but only for the area covered by the
canopy. This change overcame a poor boundary condition in the original models whereby the modeled
canopy failed to wet up beyond a certain degree of sparseness. As might be expected, the Gash sparse
model gives better results than the original version in terms of modelling error. More than 76% of the
model applications resulted in errors below 10%, with an important contribution of model
performances with errors under 5% (51% of the applications). In contrast, the Gash original model the
figure was 27%. The better performance of the Gash sparse model may be due to the conceptual
changes introduced in this version but it may also be caused by many of the applications not being duly
validated. It should also be mentioned that the original version has mainly been applied to temperate
climates, whereas the sparse version has been applied mainly to tropical and semiarid climates, also
with good results (Deng et al., 2022; Herbst et al., 2006; Junqueira Junior et al., 2019; Limousin et al.,
2008; Ma et al., 2019; 2020). However, this model has rarely been applied to shrubs compared to
forests in semiarid areas. The lack of research in this area can be attributed to the fact that evaporation
from wet shrubland canopies, unlike wet forest canopies, is not a net water loss (David et al., 2005) and
because of the dif?cult techniques of water ?ow measurement (Dunkerley, 2000). Domingo et al. (1998)
showed that interception models could be successfully applied to shrubs, despite their structural
7
differences from forests. Recently, several studies have measured rainfall partitioning after changes in
forest structure and derived interception parameters for model applications (Grunicke et al., 2020; Ma
et al., 2020; Sun et al., 2015). Most of these studies have been based on measurement data only over
several months, one growing season, or one year. Hence, the model parameters are derived from very
short datasets. Few long-term studies (i.e., > 1 year) have focused on shrub structural changes. Few
long-term studies (i.e., > 1 year) have focused on the effect of structural changes in the shrubland
canopy over time on interception loss.
The aims of this study were to (1) quantify the effects of vegetation changes on rainfall partitioning in
control and thinned plots of two shrub species (Caragana korshinskii and Salix psammophila) and (2)
clarify changes in interception processes using the revised Gash model based on the determination of
interception parameters. With these aims in mind, we conducted a 5-year experiment in which we
measured rainfall partitioning in two re-vegetated shrublands (C. korshinskii and S. psammophila) in
the Chinese Loess Plateau subjected to two thinning intensities (moderate thinning [MT] and heavy
thinning [HT]) and no thinning (NT). The revised Gash model modeled canopy interception losses in
the control and thinned treatments. This study’ results can help to predict and manage ecohydrological
change in water-limited ecosystems in the context of shifting shrub cover and climate conditions.
2. Methods
2.1 Study area
This study was conducted in the Liudaogou watershed (38.78°N, 110.35°E; 1,200 m altitude; 6.89 km2)
of Shenmu County, Shaanxi Province, China (Figure. 1). The watershed is located in an area of the
8
Loess Plateau known as a “water-wind erosion crisscross region.” The climate in the region is
mid-temperate semiarid. The average annual precipitation between 2003 and 2021 was 457 mm, with
approximately 83% of precipitation occurring between June and October. The potential evaporation is
1,200 mm y-1, and the average annual temperature is 9.6° C. January and July are the coldest and
warmest, with average monthly temperatures of -6.4 and 23.5° C, respectively. Aeolian sandy soil and
loess are typical soil types in this catchment. Two shrub species, C. korshinskii and S. psammophila,
are widely planted for ecological restoration in the catchment. Rainfall and high-temperature seasons
are synchronized, with plant growth in early April and plant senescence in late October.
The present experiment commenced in 2017, and monitoring continued until 2021. Two plantations of
C. korshinskii (30 × 10 m) and S. psammophila (15 × 6 m) on the level ground were selected. Each plot
was further subdivided into three equal plots of 10 × 10 m and 5 × 6 m (Figure. 1). C. korshinskii and S.
psammophila were planted in 2013, and the average plant heights at the start of the study were 165 and
290 cm, respectively. Both C. korshinskii and S. psammophila have an inverted-cone canopy and are
multi-stemmed shrubs with no trunk or branches extending obliquely from the base.
In June 2019, for C. korshinskii, one plot was subjected to MT, one was subjected to HT, and one was
left (NT) as a control. Likewise, for S. psammophila, one plot was subjected to MT, one was subjected
to HT, and one (NT) was left as a control (Figure. 1). In the MT plots, 25% of the branches were
removed, whereas 50% of the branches were removed in the HT plots. All thinning operations were
conducted by local villagers using branch shears to minimize soil disturbance on the plots. All twigs
and branches from the thinned shrubs were removed from the plots. The number of branches in each
plot post-treatment is shown in Figure 2. Despite the significant difference in the number of
9
branches/base diameters in the three plots under the three various densities for two shrubs, the
frequency distribution was not significantly different.
2.2 Rainfall, throughfall, and stem?ow measurements
Gross precipitation (Pg, mm) was measured using an Onset? (Onset Computer Corp., Bourne, MA,
U.S.A.) RG3-M tipping bucket rain gauge (0.2 mm per tip) located 3 m outside the plots in the study
area. As automatic rain gauges generally underestimate rainfall, three manual rain gauges (20 cm
diameter) were placed around it for calibration and were read immediately after each rainfall event. In
this way, rainfall characteristics, such as the rainfall duration (h) and average rainfall intensity (mm h-1),
were calculated. A rain event was defined as a period with more than 0.2 mm of total Pg, separated by
at least 6 h without rain.
Throughfall (TF, mm) was measured using 13 rain gauges (20 cm diameter) at each study plot. To
overcome spatial variability in TF, the rain gauges at each plot were randomly placed every year. The
average TF was computed from all functioning rain gauges. To reduce evaporative loss from the rain
gauges, TF was measured within 2 h after rainfall ended during the daytime to reduce evaporative loss
from the rain gauges. If a rainfall event ended at night, TF was measured early the next morning.
Stem?ow (SF, mm) was measured on 12 representative branches at each study plot, with three
branches from branch basal diameter (BD) categories of < 5, 5?10, 10?15, and 15?18 mm. SF was
collected using aluminum foil collars. Each collar was fitted around the entire branch circumference
and near the base of the branch and sealed with neutral silicone caulk (Figure. 1). A 0.8 cm diameter
PVC hose was used to guide the SF into a container fitted with a lid. The collars and hoses were
10
checked periodically for leaks and blockages, respectively. After obtaining these measurements, SF
was returned to the branch base to alleviate unnecessary drought stress on the sample branches. The SF
depth after each rainfall event was calculated in each shrub plot as follows:
(1)1=0
niVSF??
(2)dVPA
where SFv is the SF volume (L) of a plot after a rainfall event, Vi is the SF volume (ml) collected from
individual stems in a plot, n is the total number of stems in a plot, SFd is the SF depth (mm), and PA is
the plot area (m2).
The observed interception loss (I, mm) was calculated using Equation (3):
(3)g()IPTFS???
One-way ANOVA (post hoc Tukey''s tests) was used to recognize the significant differences (p < 0.05)
in rainfall partitioning variables across treatments in the same year and the same treatments in different
years. Statistical analysis was performed in SPSS (IBM SPSS Statistics, version 25.0).
2.3 Revised Gash analytical model
2.3.1 Description of the model
In the revised Gash analytical model, Pg is expressed as a series of discrete rainfall events in which the
canopy has sufficient time to dry before another rainfall event begins (Gash et al., 1995). Each rainfall
11
event includes three phases: wetting, saturation, and cessation. The total canopy interception was
obtained by summing the interception losses of the canopy during each phase of a rainfall event.
Rainfall events were analyzed separately from a threshold value of Pg’ to determine whether the canopy
was saturated. To calculate interception loss, the revised analytical model incorporates two types of
data: canopy and climatic parameters. The canopy parameters are the canopy storage capacity (S),
canopy cover (c), free TF coe?cient (p) (calculated using Equation (4)), trunk storage capacity (St),
and rainfall fraction diverted into SF (pt). The climatic parameters are Pg, mean evaporation rate ( , E
mm h?1), and mean rainfall intensity ( , mm h?1) during each rainfall event. The amounts of rainwater R
required to saturate the canopy and trunk fully were calculated using Equations (5) and (6),
respectively:
(4)1pc??
(5)??g''ln/ccRPSE
(6)''/ttp?
where Pg’ and Pt’ are the amounts of rainwater required to saturate the canopy and trunk fully,
respectively; (mm h?1) is the mean evaporation rate per unit area of canopy cover; Sc (mm) is the cE
canopy storage capacity per unit area of cover.
The revised analytical model divides simulated interception loss into ?ve components, and the equation
used to calculate each component is shown in Table 1.
Table 1. Components of simulated interception loss in the revised Gash model.
12
Components of simulated interception loss Formula
1 For m small storms insuf?cient to saturate the canopy (Pg < Pg’) ,1gimcP??
For n storms (Pg > Pg’) sufficient to saturate the canopy
2 Wetting the canopy ('')gcnS?
3 Wet canopy evaporation during storm events ??,''1giEPiR?
4 Evaporation after storm cessation cnS
5 Evaporation from trunks for q storms that saturate trunks (Pg > Pt’) ,1ttgiqqP????
2.3.2 Estimation of the model parameters
The S was obtained as a linear relationship between Pg and the sum of TF and SF, where the S was the
value of Pg when the sum of TF and SF was zero (Wallace & McJannet, 2008). Canopy cover was
estimated using hemispherical photographs taken by ?sheye webcams during the growing season, and
data were processed using CAN-EYE software (version 6.3) (Niu et al., 2021; Weiss & Baret, 2014).
The St and pt were estimated using the method of Gash and Morton (1978) as the negative intercept and
slope of the linear regression between Pg and SF. The was estimated using the method of Buttle E
and Farnsworth (2012) and calculated using Equation (7):
, (7) =EaR?
where “a” is the slope of the linear regression between Pg and observed I for Pg ≥ Pg’ and R
represents the mean rainfall intensity is less than 10 mm h-1.
13
2.3.3 Model evaluation
The performance of the revised Gash analytical model was evaluated using relative error (RE, %), the
Nash?Sutcliffe model ef?ciency (NSE) metric (Nash & Sutcliffe, 1970), and the root mean square error
(RMSE). The RE (%) was the difference between the values estimated by the model and the measured
values. Muzylo et al. (2009) classified the performance of the interception rainfall model according to
the RE as follows: poor (RE > 30%), fair (10% < RE ≤ 30%), good (5% < RE ≤ 10%), very good (1%
< RE ≤ 5%), and extremely good (RE ≤ 1%). The NSE measures the performance of each interception
model as compared to the mean and can be used to assess the predictive power of the model. The
RMSE was used to quantify the agreement between the measured data and the model predictions.
These three metrics are more suitable used to evaluate hydrological models.
2.3.4 Sensitivity analysis
To explore the relative importance of the parameters in the revised Gash model, ?ve parameters (S, c,
Pt, St, and ) were subjected to a sensitivity analysis. In the analysis, the values of these /ER
parameters were increased or decreased by up to 50% of their original values. The simulated results
were then compared with the actual measurement data.
The methodology in this study is further presented in a flowchart in Figure 3.
14
Figure 1. The study area in the Liudaogou catchment of Shenmu County, Shaanxi Province, China (a)
and C. korshinskii and S. psammophila plots subjected to no thinning (NT), moderate thinning (MT), or
heavy thinning (HT) by branch removal (b). Collection of gross rainfall (Pg) outside the plots (c) and
collection of throughfall (TF) (d) and stemflow (SF) (e).
15
Figure 2. The number and frequency distributions of branch basal diameter (BD) in the shrub plots
under no thinning (NT), moderate thinning (MT), and heavy thinning (HT) in 2019.
Figure 3. Methodological flowchart.
3. Results
3.1 Rainfall characteristics
The daily precipitation amounts, intensities, and durations for the 2017?2021 are shown in Figure 4. In
total, 674.2 mm of precipitation occurred in 2017, 637.2 mm in 2018, 436.5 mm in 2019, 359.3 mm in
16
2020, and 320.0 mm in 2021. According to the average annual precipitation (457.0 mm) between 2003
and 2021, 2017 and 2018 were classified as wet years, whereas 2019, 2020, and 2021 were classified as
normal, dry, and extremely dry years, respectively. Rainfall was unevenly distributed throughout the
year, with a mean of 88.0% occurring during the rainy season (May?October) in 2017?2021.
Individual rainfall events ranged from 0.2 to 92.5 mm, with an average of 7.6 mm. The rainfall
intensity ranged from 0.2 to 40.7 mm h-1, with an average of 4.2 mm h-1. The rainfall duration ranged
from 0.1 to 26.6 h, with an average of 3.1 h. As shown in Table 2, rainfall events less than 10 mm
account for about 75% of the rainfall events during 2017-2021, and about 90% have rainfall intensities
less than 10 mm h-1. The wet years are mainly characterized by a few more rainfall events greater than
40 mm compared to the dry years.
During the experimental periods (purple dashed lines, Figure. 4a) in the 5 years from 2017 to 2021, 116
rainfall events were recorded, with a total of 1,587.8 mm. Among them, 780.4 mm (49.1% of total Pg)
in the pre-thinning period (2017?2018) and 807.4 mm (50.9% of total Pg) in the post-thinning period
(2019?2021). The frequency distributions of event size and intensity during the pre-thinning and
post-thinning periods are shown in Figure 5. Generally, small rainfall events (low depths) were more
frequent and contributed to a lower total Pg percentage than large rainfall events (Figure. 5a, c).
Generally, the event frequency of low-intensity rainfall events (range: 0?5 mm h-1) was higher (> 50%
of total events) than the frequency of high-intensity rainfall events in the pre-thinning and post-thinning
periods (Figure. 5b, d). This suggested that lower-intensity rainfall events accounted for a higher
percentage of total Pg.
17
Figure 4. Daily precipitation distribution for 2017?2021 (purple dashed lines = the experimental
periods).
18
Figure 5. Frequency distribution of rainfall events in di?erent ranges of rainfall amounts (a, c) and
rainfall intensities (b, d) during both the pre-thinning (2017?2018) and post-thinning (2019?2021)
periods in the study region.
19
Table 2. Rainfall classification according to rainfall amount and intensity during observation years.
Rainfall classification Observation years n <2 2-5 5-10 10-20 20-40 >40
Amount (mm) 2017 86 34 (39.5%) 25 (29.1%) 8 (9.3%) 11 (12.8%) 3 (3.5%) 5 (5.8%)
2018 64 28 (43.8%) 6 (9.4%) 11 (17.2%) 9 (14.1%) 6 (9.3%) 4 (6.2%)
2019 61 25 (41.0%) 11 (18.0%) 13 (21.3%) 7 (11.5%) 3 (4.9%) 2 (3.3%)
2020 59 28 (47.4%) 13 (22.0%) 6 (10.2%) 7 (11.9%) 5 (8.5%) 0
2021 48 20 (41.7%) 8 (16.6%) 8 (16.6%) 9 (18.8%) 3 (6.3%) 0
2017-2021 318 135 (42.4%) 63 (19.8%) 46 (14.5%) 43 (13.5%) 20 (6.3%) 11 (3.5%)
Intensity (mm h-1) 2017 86 46 (53.5%) 21 (24.4%) 9 (10.4%) 6 (7.0%) 4 (4.7%) 0
2018 64 30 (46.9%) 23 (35.9%) 5 (7.8%) 4 (6.3%) 2 (3.1%) 0
2019 61 30 (49.2%) 20 (32.8%) 5 (8.2%) 3 (4.9%) 3 (4.9%) 0
2020 59 37 (62.7%) 12 (20.3%) 2 (3.4%) 3 (5.1%) 5 (8.5%) 0
2021 48 30 (62.5%) 11 (22.9%) 4 (8.3%) 1 (2.1%) 2 (4.2%) 0
2017-2021 318 173 (54.4%) 87 (27.4%) 24 (7.5%) 18 (5.7%) 16 (5.0%) 0
20
3.2 Rainfall partitioning measurements
The changes in annual rainfall partitioning during the pre- and post-thinning periods are shown in
Figure 6 and Table 3. In 2017?2018, before thinning, there were no significant differences in the ratios
of TF, SF, and I to Pg in the C. korshinskii and S. psammophila plots subjected to NT, MT, and HT. In
2019?2021, after thinning, both thinned plots had a higher TF rate than the plots subjected to NT
(Figure. 6a, b). In both the C. korshinskii and S. psammophila plots, the TF rate increased by about 12%
(MT) and 20% (HT) compared to the plots subjected to NT. In contrast, the SF and I rates significantly
decreased with thinning intensity (Figure. 6c, d, e, f). In the C. korshinskii plots, the total SF decreased
by 21% (MT) and 49% (HT), and the total I decreased by 37% (MT) and 54% (HT) compared to the
plot subjected to NT. Corresponding values for S. psammophila were 31% (MT) and 50% (HT) for the
SF rate and 29% (MT) and 50% (HT) for the I rate. For both shrub species, although the ratio of I to Pg
was highest under the NT treatment in 2021, the plots in the thinned treatments had already reached
68.5% (MT) and 48.6% (HT) of this value for C. korshinskii, 73.0% (MT) and 67.2% (HT) for S.
psammophila. It is important to emphasize that in the plots subjected to HT, the effects of the
extremely dry year (2021) on total I cannot be determined.
The relationship between Pg and canopy water balance (TF, SF, and I) based on rainfall event data is
shown in Figure 7. The event-based canopy water balance amount increased linearly with Pg (P < 0.01).
The ratio of canopy water balance to Pg varied greatly throughout the study period and stabilized
gradually under rainfall events greater than 20 mm.
21
Figure 6. Annual ratio of throughfall (TF), stemflow (SF), and canopy interception loss (I) to gross
rainfall (Pg) in the C. korshinskii and S. psammophila plots before (2017?2018) and after thinning
(2019?2021). In the figure, a capital letter indicates a signi?cant difference between the thinning
treatments (NT, MT, or HT), and a lower-case letter indicates a signi?cant difference between years
within the same treatment (p < 0.05).
22
Table 3. Observed rainfall amount, throughfall, stemflow, and interception loss during pre- (2017-2018) and post-thinning (2019-2021) periods in control and treated shrub
plots (2017 and 2018 were wet years, 2019, 2020, and 2021 were normal, dry, and extremely dry years, respectively).
Shrub species Plots Study period Rainfall amount (mm) Throughfall (mm) Stemflow (mm) Interception (mm)
C. korshinskii NT Pre-thinning 780.4 560.9 (71.9%) 77.1 (9.9%) 142.4 (18.2%)
Post-thinning 807.4 605.3 (75.0%) 80.3 (9.9%) 121.8 (15.1%)
MT Pre-thinning 780.4 553.7 (71.0%) 77.9 (9.9%) 148.8 (19.1%)
Post-thinning 807.4 667.6 (82.7%) 61.5 (7.6%) 78.3 (9.7%)
HT Pre-thinning 780.4 562.7 (72.1%) 75.7 (9.7%) 142.0 (18.2%)
Post-thinning 807.4 708.7 (87.8%) 41.9 (5.2%) 56.8 (7.0%)
S.psammophila NT Pre-thinning 780.4 553.8 (71.0%) 55.0 (7.0%) 171.6 (22.0%)
Post-thinning 807.4 601.2 (74.5%) 56.8 (7.0%) 149.4 (18.5%)
MT Pre-thinning 780.4 554.9 (71.1%) 55.2 (7.1%) 170.3 (21.8%)
Post-thinning 807.4 661.2 (81.9%) 39.9 (4.9%) 106.3 (13.2%)
HT Pre-thinning 780.4 554.1 (71.0%) 54.5 (7.0%) 171.8 (22.0%)
Post-thinning 807.4 696.7 (86.3%) 30.4 (3.7%) 80.3 (10.0%)
23
Figure 7. Relationship between gross rainfall (Pg) and canopy water balance using rainfall event data
from 2019 to 2021 in the C. korshinskii and S. psammophilas plots subjected to no thinning (NT),
moderate thinning (MT), or heavy thinning (HT). (a) Throughfall (TF) amount, (b) stemflow (SF)
amount, (c) canopy interception loss (I) amount, (d) ratio of TF to Pg, (e) ratio of SF to Pg, and (f) ratio
of I to Pg for C. korshinskii. (g) TF amount, (h) SF amount, (i) I amount, (j) ratio of TF to Pg, (k) ratio
of SF to Pg, and (l) ratio of I to Pg for S. psammophila. p < 0.05, p < 0.01
24
3.3 Model parameterization
Model input parameters may provide insights into the quality of the model’s performance and be useful
in the model’s result interpretation, so they should be given priority consideration. From the linear
regressions of Pg and rainfall partitioning components (Figure. 6), the study derived the required
parameter values in the revised Gash model applied to the C. korshinskii and S. psammophila plots
under NT, MT, and HT are shown in Tables 4 and 5. The estimated canopy parameters (S, PAI, c, Pt,
and St) and climatic parameter ( ) gradually decreased as thinning intensity increased. For example, E
the S of the C. korshinskii and S. psammophila plots decreased from 0.90 ± 0.05 to 0.40 ± 0.10 mm,
and the decreased from 0.30 ± 0.02 to 0.13 ± 0.03. Speci?cally, the S, Pt, and St values of C. E
korshinskii were slightly larger than those of S. psammophila in the plots under NT, MT, and HT.
Parameters for the observed periods agree with the development of vegetation structure (Table 4). The
thinning treatments reduced the measured parameters of vegetation structure, plant area index (PAI)
and c, to about 73% (MT) and 50% (HT) of the corresponding values of the control (NT) (Table 4). In
2017?2018 (two wet years), the relative PAI increase (rPAI-I, i.e., the change in the PAI divided by the
initial PAI) was within 10% of all plots. In 2019?2020 (a normal and dry year, respectively), the rPAI-I
declined to a lesser extent in the thinned plots (MT and HT) compared with the plots subjected to NT,
especially in the C. korshinskii plots. In 2021, under particularly dry conditions, the degree of decrease
in the rPAI-I of the plots subjected to HT was reduced by about half compared with that of the plots
under NT and MT. Although the PAI of the NT treatment was the highest in 2021, the plots on the
thinned treatments had already reached about 85% (MT) and 70% (HT) of this value for two shrubs.
Based on 5 years of measurement data, there was a simple linear relationship between the PAI values
25
and canopy water balance of the two shrub plantations (Figure. 8). According to the findings, the TF
rate decreased in accordance with an increase in the PAI in the two shrub plantations, whereas SF and I
rates increased in accordance with an increase in the PAI.
26
Table 4. The parameter values in the revised Gash analytical model applied to the C. korshinskii and S. psammophila plots subjected to no thinning (NT), moderate thinning
(MT), or heavy thinning (HT) during 2017?2021. Data obtained in the NT plots of the C. korshinskii and S. psammophila in 2017?2019 were used for calibration, and data
obtained in these plots in 2020?2021 were used for validation.
C. korshinskii S.psammophilaPlots Years
S (mm) PAI c p Pt St (mm) TF—E
—
R S (mm) PAI c p P
t St (mm) TF
—
E
NT 2017 0.82 2.33 0.85 0.15 0.100 0.121 0.24 2.30 0.78 1.64 0.71 0.29 0.078 0.098 0.35
2018 0.94 2.64 0.92 0.08 0.105 0.128 0.28 2.38 0.87 1.80 0.76 0.24 0.072 0.100 0.36
2019 0.85 2.52 0.90 0.10 0.108 0.119 0.26 2.43 0.85 1.77 0.73 0.27 0.076 0.106 0.33
2020 0.82 1.99 0.79 0.21 0.094 0.109 0.24 1.69 0.80 1.62 0.70 0.30 0.082 0.108 0.32
2021 0.68 1.47 0.65 0.35 0.074 0.087 0.22 2.02 0.58 1.16 0.60 0.40 0.059 0.077 0.29
MT 2017 0.86 2.35 0.83 0.17 0.103 0.112 0.23 2.30 0.76 1.65 0.70 0.30 0.078 0.094 0.34
2018 0.97 2.55 0.91 0.09 0.103 0.121 0.29 2.38 0.81 1.79 0.72 0.28 0.079 0.107 0.35
2019 0.65 1.74 0.65 0.35 0.080 0.099 0.17 2.43 0.60 1.27 0.53 0.47 0.051 0.075 0.29
2020 0.60 1.64 0.64 0.36 0.079 0.106 0.14 1.69 0.58 1.21 0.52 0.48 0.050 0.075 0.28
2021 0.48 1.27 0.55 0.45 0.058 0.076 0.14 2.02 0.40 0.86 0.45 0.55 0.043 0.065 0.25
HT 2017 0.86 2.34 0.82 0.18 0.099 0.111 0.25 2.30 0.75 1.67 0.70 0.30 0.075 0.090 0.36
2018 0.98 2.49 0.92 0.08 0.102 0.116 0.26 2.38 0.84 1.78 0.75 0.25 0.076 0.096 0.38
2019 0.42 1.14 0.47 0.53 0.060 0.086 0.13 2.43 0.38 0.87 0.38 0.62 0.042 0.071 0.22
2020 0.41 1.07 0.45 0.55 0.058 0.084 0.12 1.69 0.32 0.85 0.37 0.63 0.041 0.070 0.21
2021 0.40 0.93 0.44 0.56 0.048 0.072 0.11 2.02 0.30 0.76 0.35 0.65 0.037 0.063 0.19
27
Table 5. The calibration parameters in the revised Gash model for the C. korshinskii and S.
psammophila plots under the NT, MT, and HT treatments. The data were derived from June, July, and
September 2019?2021 datasets.
Shrubs Plots S (mm) c Pt St (mm) TF—E
C. korshinskii NT 0.75 0.74 0.090 0.105 0.28
MT 0.52 0.59 0.065 0.085 0.14
HT 0.40 0.44 0.050 0.075 0.10
S.psammophila NT 0.76 0.68 0.075 0.100 0.32
MT 0.46 0.48 0.046 0.070 0.22
HT 0.32 0.37 0.038 0.065 0.16
Figure 8. Relationship between plant area index (PAI) values and canopy water balance of the C.
korshinskii and S. psammophila plots for 2017?2021. The ratio of (a) TF to Pg, (b) SF to Pg, and (c) I to
Pg. p < 0.05
3.4. Rainfall interception modeling
The performance of the revised Gash model for the entire monitoring period (2017?2021) in the C.
korshinskii and S. psammophila plots subjected to NT is illustrated in Figure 9 and Table 6. Comparing
the simulated I with the measured values showed that the revised Gash model severely underestimated
I in wet years (2017 and 2018), especially in the presence of an individual rainfall event greater than 40
mm, such as that occurred in 2019, despite this classified as a normal year. In dry years (2020 and
28
2021), when there were many individual rainfall events of less than 10 mm, the revised Gash model
overestimated I. Applying the method proposed by Muzylo et al. (2009), the performance of the revised
Gash model in simulating I in both shrub plantations was classified as “fair” for both shrubs throughout
the observation period, “poor” in wet and extremely dry years, and “fair” in normal and dry years. But
in 2020 (dry year), the performance of the revised model in simulating I by S. psammophila was
classified as “very good.”
In the NT, MT, and HT plots of C. korshinskii and S. psammophila, individual rainfall events observed
during June, July, and September of 2019?2021 were used to calibrate the revised Gash model. The
calibrated models were then validated using measurements obtained in August and October 2019?2021.
The measured and simulated total I according to the revised Gash model are summarized in Table 7
and plotted in Figure 10. The I was underestimated to varying degrees in the calibration period, such as
in NT plots, with underestimation of 3.0% for C. korshinskii and 14.5% for S. psammophila. As the
degree of thinning increased, the degree of underestimation gradually increased. In contrast to the
results of the revised Gash model in the calibration period, in the validation period, the revised model
overestimated I, with the degree of overestimation decreasing in accordance with the thinning intensity.
Both root mean square error (RMSE) and NSE values gradually decreased with an increase in the
thinning intensity in both the calibration and validation periods. The performance of the revised Gash
model in all the plots varied from “fair” to “good” to “very good,” with a RE of < 20.0%. As shown in
Figure 10, the simulated values of individual rainfall events were slightly higher than the measured
values when the rainfall amounts were small. When the rainfall amounts were large, the simulated
values were lower than the measured values. Based on the correlation (R2) of the simulated and
observed values, the validation period was slightly better than the calibration period (Figure. 10).
29
Changes in interception components were compared in the six plots of the two shrub species (Figure.
11). The sum of evaporation during and after rainfall consistently accounted for the largest component
of estimated interception loss (about 90%). With an increase in the thinning intensity, evaporation
during rainfall gradually decreased in the C. korshinskii plots (from 51.6% to 42.9%) but increased in
the S. psammophila plots (from 55.1% to 57.1%). In contrast, evaporation after rainfall gradually
increased in the C. korshinskii plots (from 37.4% to 44.8%) but decreased in the S. psammophila plots
(from 35.4% to 30.6%).
To explore the relative importance of the parameters in the revised Gash model, a sensitivity analysis
of ?ve parameters (S, c, Pt, St, and ) was conducted (Figure. 12). and S were the most /ER/ER
sensitive parameters in the C. korshinskii and S. psammophila plots, followed by c, St, and Pt.
Figure 9. Accumulated total observed and simulated interception loss during the observation period
(2017?2019 data for calibration and 2020?2021 data for validation) in the C. korshinskii and S.
psammophila plots under the no thinning (NT) treatment according to the revised Gash model.
30
Table 6. Comparison of observed and simulated interception loss during the observation period
(2017?2021) in the C. korshinskii and S. psammophila plots under the no thinning (NT) treatment
according to the revised Gash analytical model.
Years Plots Observed I (mm) Simulated I (mm) RE (%) Classification
C. korshinskii NT 265.8 223.2 -16.0 Fair2017-2021
S. psammophila
NT
319.1 255.1 -20.0 Fair
C. korshinskii NT 70.4 38.6 -45.1 Poor2017
(calibration) S. psammophila
NT
69.6 44.0 -36.8 Poor
C. korshinskii NT 69.4 46.0 -33.7 Poor2018
(calibration) S. psammophila
NT
94.8 55.8 -41.1 Poor
C. korshinskii NT 59.4 43.2 -27.3 Fair2019
(calibration) S. psammophila
NT
66.5 49.2 -25.9 Fair
C. korshinskii NT 45.3 56.2 24.0 Fair2020
(validation) S. psammophila
NT
61.8 63.1 2.2 Very good
C. korshinskii NT 21.3 39.1 84.0 Poor2021
(validation) S. psammophila
NT
26.5 43.0 62.5 Poor
31
Figure 10. Observed interception loss versus simulated interception loss at the event-based scale
according to the revised Gash model in the calibration (June, July, and September 2019?2021) and
validation (August and October 2019?2021) periods in the C. korshinskii and S. psammophila plots
under no thinning (NT), moderate thinning (MT), or heavy thinning (HT).
32
Table 7. Comparison of observed and simulated interception loss according to the revised Gash analytical model in the calibration (June, July, and September 2019?2021)
and validation (August and October 2019?2021) periods in the C. korshinskii and S. psammophila plots under no thinning (NT), moderate thinning (MT), or heavy thinning
(HT).
Calibration period Validation period
C. korshinskii S.psammophila C. korshinskii S.psammophilaComponents of interception
NT MT HT NT MT HT NT MT HT NT MT HT
Simulated I (mm) 72.1 46.6 34.0 82.3 54.0 39.9 60.2 34.2 25.5 63.9 44.8 31.6
Observed I (mm) 74.3 50.4 36.8 96.2 66.8 49.7 51.7 29.5 22.1 58.4 43.5 31.1
Relative error (%) -3.0 -7.5 -7.6 -14.5 -19.2 -19.7 16.4 15.9 15.4 9.3 3.0 1.4
RMSE (mm) 1.34 0.45 0.23 1.35 1.19 0.87 1.68 0.25 0.13 0.81 0.69 0.43
NSE 0.44 0.36 0.34 0.48 0.39 0.38 0.49 0.47 0.46 0.71 0.70 0.67
Classification Very Good Good Good Fair Fair Fair Fair Fair Fair Good Very Good Very good
33
Figure 11. Interception components were estimated using the revised Gash model for 2019?2021 in
control (NT) and thinned (MT and HT) shrub plots.
34
Figure 12. Sensitivity analysis of the canopy parameters c, S, Pt, and St and climatic parameter /ER
in the revised Gash model under no thinning (NT), moderate thinning (MT), or heavy thinning (HT) in
the C. korshinskii and S. psammophila plots. Note: The change in c should be less than or equal to 1.
4. Discussion
4.1 Thinning effects on rainfall partitioning
The TF rate gradually increased, while the SF and observed I rates decreased with increasing thinning
intensity on an annual scale (Figure. 6). Although the gross amount of incident rainfall directly
determines the magnitudes of TF and SF (Figure. 7) and levels of saturation of canopy and stem
surfaces (Carlyle-Moses, 2004; Levia et al., 2010), many studies have shown that the factors affecting
rainfall partitioning patterns include meteorological variables, such as rainfall intensity, air temperature
and potential evapotranspiration (Crockford & Richardson, 2000; Llorens & Domingo, 2007; Staelens
et al., 2008), and the actual contributions are also dependent on vegetation composition and structure
(Crockford & Richardson, 2000; Sadeghi et al., 2020; Zhang et al., 2023). Shrub structures (e.g.,
canopy cover and stand density) strongly affect variations in rainfall partitioning (Chang et al., 2022;
Yue et al., 2021; Zhang et al., 2021). Thinning alters the structures of shrub plots, and can serve as an
important method for regulating the redistribution of water resources in watersheds with shrubland
cover. Thus, examining changes in various components of the shrub water cycle, such as transpiration
and evapotranspiration, as a result of thinning is important to improve understanding of the processes
underlying changes in the water yield.
35
In the present study, changes in rainfall partitioning due to thinning were consistent with a simple
linear relationship between the PAI and canopy water balance values of the two shrub plantations based
on 5 years of measurement data (Figure. 8). Thus, the linear relationship can be used as a useful tool for
predicting net precipitation and observed I rates. When the PAI of the C. korshinskii and S.
psammophila plots was the same, the TF rates of C. korshinskii was significantly higher than that of S.
psammophila, whereas the I rates of C. korshinskii was significantly lower than that of S. psammophila,
there was no significant difference in the SF rates of the C. korshinskii and S. psammophila plots
(Figure. 8). However, in studies conducted in regions with climatic variables similar to those in the
present study, when assessing at the individual plant level, not the stand level, the SF rate of C.
korshinskii was significantly higher than that of S. psammophila and there was no significant difference
in the TF rate (Yuan et al., 2017; Yang et al., 2019b). These findings can be explained by the canopy
projection area of different shrubs. The canopy projection area of C. korshinskii is smaller than that of
S. psammophila. In addition, as the stand density increases, the degree of overestimation of TF and
underestimation of SF and I increases at the individual plant level. As a result, measurements of I at the
individual plant level may underestimate results at the stand level at biome and global scales. Thus, the
role of rainfall partitioning in hydrological processes needs to be studied at the plant community level.
Results from the 3 years after thinning of this study indicate that more heavily thinning treatments had
more pronounced effects on shrub growth, net precipitation, and interception, especially in dry years
(Table 4). However, the effects of thinning intensity are subject to the interplay among various factors.
When shrub plots become dense over time, the canopy intercepts more rain, potentially shortening the
soil water recharge process (Snyder et al., 2021). In water-limited regions, abiotic and biotic factors
affect the soil water content. Vegetation depends on soil moisture, and soil moisture is affected by
36
water uptake by plants, shading, SF, and I (Llorens & Domingo, 2007; Metzger et al., 2017;
Rodriguez-Iturbe, 2000). If a reduction in net precipitation (sum of TF and SF) is not balanced by a
reduction in evaporation due to canopy shading, high densities of shrubs may have a significant impact
on the water budget of the respective ecosystem. Comparative studies of different thinning intensities
can help us to determine the most suitable thinning method for shrub plots to reduce drought stress. For
example, Gebhardt et al. (2014) indicated that repeated moderate thinning was a better option than the
heavy thinning because of heavy thinning induced the progressive development of understory, which
not only competed for resources with trees but also hindered natural regeneration. In this study, the
understory of S. psammophila plot was sparse, so its effects are expected to be minor. Still, for C.
korshinskii plot, the role of the understory in the longer term could become important, affecting the
difference between the two thinning treatments. This further emphasizes the need to examine the
long-term effects of the thinning treatments in the two shrub plots in this study.
4.2 Interception parameters
The S of the C. korshinskii and S. psammophila plots remained around or below 1 mm throughout the
monitoring period (Table 4), which agrees well with ?ndings obtained for other semiarid shrub species
(Zhang et al., 2018). The S of forests is significantly higher than that of shrubs in similar areas. For
example, Ma et al. (2019) reported a S value of 1.34 for Roinia pseudoacacia and 1.43 for Pinus
tabuliformis. Other studies reported that climatic variables (e.g., rainfall intensity and wind speed) and
canopy traits (e.g., cover, height, and leaf area index) influenced S (Carlyle-Moses & Gash, 2011). In
this study, SF was relatively low compared to the values obtained for other parts of rainfall partitioning
in the C. korshinskii and S. psammophila plots (Table 4). The SF-related parameters St and Pt (around
37
0.1 mm) in two shrub plots were much lower than those reported by Zhang et al. (2018) (0.55 mm for
St and 0.68 mm for Pt), but measured SF rates agreed with those found in previous studies in semiarid
regions (Yuan et al., 2017; Yang et al., 2019b; Li et al., 2008; Yue et al., 2021). In this study, the
observed mean values of St and Pt were slightly higher in the C. korshinskii plots than in the S.
psammophila plots, these findings can be explained by the lower threshold of precipitation (0.9 mm for
C. korshinskii vs. 2.1 mm for S. psammophila) and beneficial leaf traits of C. korshinskii versus those
of S. psammophila (Yuan et al., 2017).
In the present study, changes in canopy structure induced by thinning altered interception parameters
(e.g., S, p, and ), thereby influencing I (Table 4). As reported previously, the S and p also lead to E
spatial variability in TF (Loustau et al., 1992; Sun et al., 2015). In this study, in the three thinning years,
there was a greater reduction in S, St, and Pt in the shrub plots subjected to NT and MT than HT.
During 2019?2021, p also increased under NT (0.25 and 0.13 for C. korshinskii and S. psammophila,
respectively) and MT (0.10 and 0.08 for C. korshinskii and S. psammophila, respectively) with
decreasing rainfall. In the plots subjected to HT, p slightly increased and remained at a high level (0.03
for both C. korshinskii and S. psammophila). Both light exposure and aerodynamic conductance of
branches increase with an increase in canopy openness caused by thinning, thus changing
meteorological conditions, such as temperature, humidity, and wind speed, all of which control
evaporation of water intercepted and stored within the canopy (Pypker et al., 2005).
The revised Gash model performance (10% < RE ≤ 20%) improved over the entire observation
period compared to weaker Gash model performance (24% < RE < 84%) for a single year (Table 4).
Fluctuations in model result from the years (see Figure. 9 and Table 6) reflect data availability within a
38
single year rather than vegetation change. Generally, the reliability of derived model parameters
depends on the dataset (i.e., measurement duration) from which the parameters are obtained. In this
study, the parameters derived from three different hydrological years (normal, dry, and extremely dry)
yielded better simulation results than the parameters derived from only one type of hydrological year
(Tables 6 and 7). This clearly shows that the performance of the revised Gash analytical model depends
on reliable parameterization. Including a range of parameters in the revised model from a large number
of hydrological years can enable estimations of rainfall interception over time, which can be used in
shrub water balance studies in arid and semiarid regions.
4.3 Performance of the revised Gash analytical model
The revised Gash analytical model provided good estimates of the total I in both the C. korshinskii and
S. psammophila plots (Tables 6 and 7), and it also captured canopy density variation. Based on the
RMSE, the revised Gash model performed better after thinning than before. In contrast, the NSE values
were markedly better before thinning than after thinning (Table 7). This is not surprising, as intensive
thinning greatly reduces variability in I and TF, and these models are not designed to measure rainfall
events in the open field (Shinohara et al., 2015). In terms of modeling error, the underestimation of
simulated total I using the revised Gash model was similar under the NT treatment in both the C.
korshinskii and S. psammophila plots (Table 6) and within the range reported by other studies (Fan et
al., 2014; Fathizadeh et al., 2018; Junqueira Junior et al., 2019; Limousin et al., 2008; Shinohara et al.,
2015). Muzylo et al. (2009) concluded that the expected modeling error in the prediction of
interception loss can be as high as 20%. In this study, I was slightly overestimated in the control and
thinned plots for small rainfall events during the calibration and validation periods. For high rainfall
39
events, the predicted I was far from the 1:1 line in both the C. korshinskii and S. psammophila plots
(Figure. 10). The revised Gash model severely underestimated I as rainfall increased (Figure. 9). Thus,
this model should be used with caution in areas with high rainfall events (Fathizadeh et al., 2018;
Limousin et al., 2008; Sadeghi et al., 2015). Based on meteorological data collected using automatic
rain gauges in the study area in the past ten years, single rainfall events with heavy rainfall amounts (>
40 mm) account for about 30% of the total rainfall. Therefore, the effect of such rainfall events should
be considered in future research to improve the performance of the model.
The thinning treatments altered the the interception components in the present study. Evaporation
during storm events and the post-storm period accounted for the largest amount of interception loss
before and after thinning (Figure. 11), which is consistent with the findings of previous studies (Sun et
al., 2015; Ma et al., 2019; 2020). Changes in S and strongly affected I (Figure. 12). The degree of E
decrease in the S of the C. korshinskii plots under MT and HT relative to that under NT was smaller
than the degree of decrease in . However, in the S. psammophila plots, the decrease in the S of the E
plots subjected to MT and HT relative to that in the plot subjected to NT was greater than the decrease
in (Table 4). Therefore, S was the main cause of the decrease in I in the thinned plots of C. E
korshinskii, whereas the was the main cause in the thinned plots of S. psammophila. E
Quanti?cation of interception components before and after thinning can help to improve understanding
of changes in interception processes, as well as underlying processes of peak ?ows, in?ltration, and
water resources in this ecosystem.
Typically, model error refers to the results of model validation rather than calibration (Muzylo et al.,
2009). This paper shows performance of the revised Gash model in S. psammophila plots than in C.
40
korshinskii plots during the validation period. This indicates the revised Gash model for interception
loss modeling can be more appropriate for the S. psammophila plots than for the C. korshinskii plots in
terms of RE and NSE values (Table 7). As noted in previous studies, the revised version has been
extensively applied in semiarid climates where sparse vegetation cover is common, and performs well
(Fan et al., 2014; Muzylo et al., 2009). Interception modeling is important for estimating the water
balance and thus plays a crucial role in hydrological simulations and determining eco-hydrological
services (Junqueira Junior et al., 2019). Future work should focus on operationalizing canopy
interception models for routine use by shrub managers to optimize shrub management strategies.
5. Conclusions
Thinning increased net precipitation reaching the soil surface and reduced canopy interception. In the
plots subjected to HT, signi?cant effects of the extremely dry year could not be identi?ed in rainfall
partitioning measurements and canopy parameters, such as the PAI and S. This suggests that HT allows
both C. korshinskii and S. psammophila plots to adapt better to dry weather conditions.
The interception parameters (S, c, Pt, St, and ) decreased with thinning intensity, and changes in S E
and strongly affected simulated I. Of note, the decrease in simulated I in the thinned plots of C. E
korshinskii was mainly caused by the S, whereas it was mainly caused by the in the S. E
psammophila plots.
The revised analytical Gash model provided a good estimate of the total I of the two shrub species (RE
< 20%). However, when evaluating the I of individual rainfall events, the simulation ability and
accuracy of the model decreased significantly, with moderate overestimation of low rainfall events (<
41
10 mm) and severe underestimation of high rainfall events (> 40 mm). These over- and
underestimations were mainly caused by uncertainty in the input parameters. Using data from a type
hydrological year or growing season to derive canopy parameters is not recommended. Parameters that
can reliably estimate I can only be generated using data with varying rainfall amounts. Moreover, as
intensive thinning greatly reduces the variability in I, the revised analytical Gash model performed
better after thinning than before thinning based on the RE values.
Although the effects of thinning on rainfall partitioning in this study lasted for three years, the effective
length of time for thinning remains to be addressed. In addition, further studies should also evaluate the
effects of other important water cycle elements (soil moisture, transpiration, etc.) to address the effects
on the whole water cycle and to improve the implementation of tree plantations in water-limited
regions.
Conflicts of interests
The authors declare that there are no conflicts of interest.
Acknowledgements
This work was supported by the Strategic Priority Research Program of the Chinese Academy of
Sciences (XDA23070202) and the National Natural Science Foundation of China (No. 41977016).
References
An, J.X., Gao, G.Y., Yuan, C., Pinos, J., Fu, B.J., 2022. Inter- and intra-event rainfall partitioning
dynamics of two typical xerophytic shrubs in the Loess Plateau of China. Hydrology and
Earth System Sciences. 26, 3885-3900. https://doi.org/10.5194/hess-26-3885-2022
42
Buttle, J.M., Farnsworth, A.G., 2012. Measurement and modeling of canopy water partitioning in a
reforested landscape: the Ganaraska Forest, southern Ontario, Canada. Journal of Hydrology.
466-467, 103-114. https://doi.org/10.1016/j.jhydrol.2012.08.021
Cao, S., Chen, L., Shankman, D., Wang, C., Wang, X., Zhang, H., 2011. Excessive reliance on
a?orestation in China''s arid and semi-arid regions: Lessons in ecological restoration.
Earth-Science Review. 104, 240-245. https://doi.org/10.1016/j.earscirev.2010.11.002
Carlyle-Moses, D., 2004. Throughfall, stemflow, and canopy interception loss fluxes in a semi-arid
Sierra Madre Oriental matorral community. Journal of Arid Environments. 58, 181-202.
https://doi.org/10.1016/S0140-1963(03)00125-3
Carlyle-Moses, D.E., Gash, J.H.C., 2011. Rainfall interception loss by forest canopies. Forest
Hydrology and Biogeochemistry: Synthesis of Past Research and Future Directions. Springer
Ecological Studies.
Chang, Q.Y., Zwieback, S., DeVries, B., Berg, A., 2022. Application of L-band SAR for mapping
tundra shrub biomass, leaf area index, and rainfall interception. Remote Sensing of
Environment. 268, 112747. https://doi.org/10.1016/j.rse.2021.112747
Chen, L.D., Wang, J.P., Wei, W., Fu, B.J., Wu, D.P., 2010. Effects of landscape restoration on soil
water storage and water use in the Loess Plateau region, China. Forest Ecology Management.
259, 1291-1298. https://doi.org/10.1016/j.foreco.2009.10.025
Christina, M., Nouvellon, Y., Laclau, J. P., Stape, J. L., Bouillet, J. P., Lambais, G. R., Maire, G. I.,
2017. Importance of deep water uptake in tropical eucalypt forest. Functional Ecology. 31,
509-519. https://doi.org/10.1111/1365-2435.12727
Crockford, R., Richardson, D., 2000. Partitioning of rainfall into throughfall, stemflow and interception:
effect of forest type, ground cover and climate. Hydrological Processes. 14, 2903-2920.
https://doi.org/10.1002/1099-1085(200011/12)14:16/17<2903::AID-HYP126>3.0.CO;2-6
David, J., Valente, F., Gash, J., 2005. Evaporation of intercepted rainfall. In: Anderson, M. (Ed.),
Encyclopedia of Hydrological Sciences. John Wiley and Sons. Ltd., pp. 627-634 (Chapter 43).
del Campo, A.D., González-Sanchis, M., Molina, A.J., García-Prats, A., Ceacero, C.J., Bautista, I.,
2019. Effectiveness of water-oriented thinning in two semiarid forests: The redistribution of
increased net rainfall into soil water, drainage and runoff. Forest Ecology Management. 438,
163-175. https://doi.org/10.1016/j.foreco.2019.02.020
Deng, J., Yu, Y., Shao, J., Lu, S., Liu, F., Li, Z., Shi, X., 2022. Rainfall interception using the revised
Gash analytical model for Pinus sylvestris var. mongolica in a semi-humid region of NE
China. Ecological Indicators. 143, 109399. https://doi.org/10.1016/j.ecolind.2022.109399
43
Domingo, F., Sanchez, G., Moro, M., Brenner, A., Puigdefabregas, J., 1998. Measurement and
modelling of rainfall interception by three semi-arid canopies. Agricultural and Forest
Meteorology. 91, 275-292. https://doi.org/10.1016/S0168-1923(98)00068-9
Dunkerley, D., 2000. Measuring interception loss and canopy storage in dryland vegetation: a brief
review and evaluation of available research strategies. Hydrological Processes. 14, 669-678.
https://doi.org/10.1002/(SICI)1099-1085(200003)
Fan, J., Oestergaard, K.T., Guyot, A., Lockington, D.A., 2014. Measuring and modeling rainfall
interception losses by a native Banksia woodland and an exotic pine plantation in subtropical
coastal Australia. Journal of Hydrology. 515, 156-165.
https://doi.org/10.1016/j.jhydrol.2014.04.066
Farley, K.A., Jobbagy, E.G., Jackson, R.B., 2005. E?ects of a?orestation on water yield: a global
synthesis with implications for policy. Global Change Biology. 11, 1565-1576.
https://doi.org/10.1111/j.1365-2486.2005.01011.x
Fathizadeh, O., Hosseini, S.M., Keim, R.F., Boloorani, A.D., 2018. A seasonal evaluation of the
reformulated Gash interception model for semi-arid deciduous oak forest stands. Forest
Ecology Management. 409, 601-613. https://doi.org/10.1016/j.foreco.2017.11.058
Feng, X., Fu, B., Piao, S., Wang, S., Ciais, P., Zeng, Z., Lü, Y., Zeng, Y., Li, Y., Jiang, X., Wu, B.,
2016. Revegetation in China’s Loess Plateau is approaching sustainable water resource limits.
Nature Climate Change. 6, 1019-1022. https://doi.org 10.1038/NCLIMATE3092
Gash, J., 1979. An analytical model of rainfall interception by forest. Quarterly Journal of the Royal
Meteorological Society 105, 43-55.
Gash, J.H.C., Morton, A.J., 1978. An application of the Rutter model to the estimation of the
interception loss from Thetford Forest. Journal of Hydrology. 38, 49-58.
https://doi.org/10.1016/0022-1694(78)90131-2
Gash, J.H.C., Lloyd, C.R., Lachaud, G., 1995. Estimating sparse forest rainfall interception with an
analytical model. Journal of Hydrology. 170, 79-86.
https://doi.org/10.1016/0022-1694(95)02697-N
Ge, J., Fan, J., Yang, X., Luo, Z., Zhang, S., Yuan, H., 2022. Restoring soil water content after alfalfa
and Korshinsk peashrub conversion to cropland and grassland. Hydrological Processes. 36,
e14539. https://doi.org/10. 1002/hyp.14539
Gebhardt, T., H?berle, K.-H., Matyssek, R., Schulz, C., Ammer, C., 2014. The more, the better? Water
relations of Norway spruce stands after progressive thinning. Agricultural and Forest
Meteorology. 197, 235-243. https://doi.org/10.1016/j.agrformet.2014.05.013
44
Grunicke, S., Queck, R., Bernhofer, C., 2020 Long-term investigation of forest canopy rainfall
interception for a spruce stand-sciencedirect. Agricultural and Forest Meteorology. s 292-293,
108125. https://doi.org/10.1016/j.agrformet.2020.108125
Herbst, M., Roberts, J.M., Rosier, P.T.W., Gowing, D.J., 2006. Measuring and modelling the rainfall
interception loss by hedgerows in southern England. Agricultural and Forest Meteorology.
141, 244-256. https://doi.org/10.1016/j.agrformet.2006.10.012
Jia, Y.H., Li, T.C., Shao, M.A., Hao, J.H., Wang, Y.Q., Jia, X.X., Zeng,C., Fu, X., Liu, B., Gan, M.,
Zhao, M., Ju, X.N., 2019. Disentangling the formation and evolvement mechanism of
plants-induced dried soil layers on China''s Loess Plateau. Agricultural and Forest
Meteorology. 269-270, 57-70. https://doi.org/10.1016/j.agrformet.2019.02.011
Jian, S.Q., Zhang, X.L., Li, D., Wang, D., Wu, Z.N., Hu, C.H., 2018. The effects of stemflow on
redistributing precipitation and infiltration around shrubs. Journal of Hydrology and
Hydromechanics. 66, 79-86. https://doi.org/10.1515/johh-2017-0043
Junqueira Junior, J.A., de Mello, C.R., de Mello, J.M., Scolforo, H.F., Beskow, S., McCarter, J., 2019.
Rainfall partitioning measurement and rainfall interception modelling in a tropical
semi-deciduous atlantic forest remnant. Agricultural and Forest Meteorology. 275, 170-183.
https://doi.org/10.1016/j.agrformet.2019.05.016
Levia, D.F., Van Stan, J.T., Mage, S.M., Kelley-Hauske, P.W., 2010. Temporal variability of stemflow
volume in a beech-yellow poplar forest in relation to tree species and size. Journal of
Hydrology. 380, 112-120. https://doi.org/10.1016/j.jhydrol.2009.10.028
Li, X.Y., Liu, L.Y., Gao, S.Y., Ma, Y.J., Yang, Z.P., 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. https://doi.org/10.1016/j.agrformet.2008.05.003
Limousin, J.M., Rambal, S., Ourcival, J.M., Joffre, R., 2008. Modelling rainfall interception in a
mediterranean Quercus ilex ecosystem: Lesson from a throughfall exclusion experiment.
Journal of Hydrology. 357, 57-66. https://doi.org/10.1016/j.jhydrol.2008.05.001
Llorens, P., Domingo, F., 2007. Rainfall partitioning by vegetation under Mediterranean conditions. A
review of studies in Europe. Journal of Hydrology. 335, 37-54.
https://doi.org/10.1016/j.jhydrol.2006.10.032
Loustau, D., Berbigier, P., Granier, A., Moussa, F.E., 1992. Interception loss, throughfall and stemflow
in a maritime pine stan. 1. Variability of throughfall and stemflow beneath the pine canopy.
Journal of Hydrology. 138, 449-467. http://dx.doi.org/10.1016/0022-1694(92)90130-N
Ma, C., Li, X., Luo, Y., Shao, M., Jia, X., 2019. The modelling of rainfall interception in growing and
dormant seasons for a pine plantation and a black locust plantation in semi-arid Northwest
China. Journal of Hydrology. 577, 123849. https://doi.org/10.1016/j.jhydrol.2019.06.021
45
Ma, C., Luo, Y., Shao, M., 2020. The modelling of rainfall interception in growing and dormant
seasons for a pine plantation and a black locust plantation in semi-arid Northwest China.
Journal of Hydrology. 590, 125234. https://doi.org/10.1016/j.jhydrol.2020.125234
Magliano, P.N., Whitworth-Hulse, J.I., Cid, F.D., Leporati, J.L., Van Stan, J. T., Jobb′agy, E.G., 2022.
Global rainfall partitioning by dryland vegetation: Developing general empirical models.
Journal of Hydrology. 607, 127540. https://doi.org/10.1016/j.jhydrol.2022.127540
Metzger, J.C., Wutzler, T., Dalla Valle, N., Filipzik, J., Grauer, C., Lehmann, R., Küsel, K., 2017.
Vegetation impacts soil water content patterns by shaping canopy water fluxes and soil
properties. Hydrological Processes. 31, 3783-3795. https://doi.org/10.1002/hyp.11274
Molina, A.J., del Campo, A.D., 2012. The e?ects of experimental thinning on throughfall and stem?ow:
A contribution towards hydrology-oriented silviculture in Aleppo pine plantations. Forest
Ecology Management. 269, 206-213. https://doi.org/10.1016/j.foreco.2011.12.037
Momiyama, H., Kumagai, T., Egusa, T., 2021. Model analysis of forest thinning impacts on the water
resources during hydrological drought periods. Forest Ecology Management. 499, 119593.
https://doi.org/10.1016/j.foreco.2021.119593
Muzylo, A., Llorens, P., Valente, F., Keizer, J.J., Domingo, F., Gash, J.H.C., 2009. A review of rainfall
interception modelling. Journal of Hydrology. 370, 191-206.
https://doi.org/10.1016/j.jhydrol.2009.02.058
Nash, J., Sutcliffe, J., 1970. River flow forecasting through conceptual models; part I-A discussion of
principles. Journal of Hydrology. 10, 282-290. https://doi.org/10.1016/0022-1694(70)90255-6
Navarro-Cerrillo, R.M., Sánchez-Salguero, R., Rodriguez, C., Duque Lazo, J., Moreno-Rojas, J.M.,
Palacios-Rodriguez, G., Camarero, J.J., 2019. Is thinning an alternative when trees could die
in response to drought? The case of planted Pinus nigra and P. Sylvestris stands in southern
Spain. Forest Ecology Management. 433, 313-324.
https://doi.org/10.1016/j.foreco.2018.11.006
Niu, X.T., Fan, J., Luo, R.H., Fu, W., Yuan, H.Y., Du, M.G., 2021. Continuous estimation of leaf area
index and the woody-to-total area ratio of two deciduous shrub canopies using ?sheye
webcams in a semiarid loessial region of China. Ecological Indicators. 125, 107549.
https://doi.org/10.1016/j.ecolind.2021.107549
Pypker, T.G., Bond, B.J., Link, T.E., Marks, D., Unsworth, M.H., 2005. The importance of canopy
structure in controlling the interception loss of rainfall: examples from a young and an
old-growth Douglas-fir forest. Agricultural and Forest Meteorology. 130, 113-129.
http://dx.doi.org/10.1016/j.agrformet.2005.03.003
46
Robinson, N., Harper, R.J., Smettem, K.R.J., 2006. Soil water depletion by eucalyptus spp. integrated
into dryland agricultural systems. Plant and Soil. 286, 141-151.
https://doi.org/10.1007/s11104-006-9032-4
Rodriguez-Iturbe, I., 2000. Ecohydrology: A hydrologic perspective of climate-soil-vegetation
dynamies. Water Resources Research. 36, 3-9. https://doi.org/10.1029/1999WR900210
Rutter, A.J., Morton, A.J., Robins, P.C., 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. https://doi.org/10.2307/2401739
Sadeghi, S.M.M., Gordon, D.A., Van Stan II, J.T. 2020. A global synthesis of throughfall and stemflow
hydrometeorology. In J. T. Van Stan, II, E. Gutmann, & J. Friesen (Eds.), Precipitation
partitioning by vegetation, a global synthesis (pp. 49-70). Springer Nature Switzerland AG.
Sadeghi, S.M.M., Attarod, P., Van Stan, J.T., Pypker, T.G., Dunkerley, D., 2015. Efficiency of the
reformulated Gash''s interception model in semiarid afforestations. Agricultural and Forest
Meteorology. 201, 76-85. https://doi.org/10.1016/j.agrformet.2014.10.006
Shao, M.A., Wang, Y.Q., Xia, Y.Q., Jia, X.X. 2018. Soil drought and water carrying capacity for
vegetation in the critical zone of the Loess Plateau: A review. Vadose Zone Journal. 17,
170077. https://doi. org/10.2136/vzj2017.04.0077
Shinohara, Y., Levia, D.F., Komatsu, H., Nogata, M., Otsuki, K., 2015. Comparative modeling of the
effects of intensive thinning on canopy interception loss in a Japanese cedar (Cryptomeria
japonica D. Don) forest of western Japan. Agricultural and Forest Meteorology. 214-215,
148-156. https://doi.org/10.1016/j.agrformet.2015.08.257
Snyder, D.K., Stringham, T.K., Snyder, K.A., 2021. Rainfall interception by mountain big sagebrush
(Artemisia tridentata spp. vaseyana): Dryland shrub canopy cover affects net precipitation.
Hydrological Processes. 36, e14441. https://doi.org/10.1002/hyp.14441
Staelens, J., De Schrijver, A., Verheyen, K., Verhoest, N.E. 2008. Rainfall partitioning into throughfall,
stemflow, and interception within a single beech (Fagus sylvatica L.) canopy: Influence of
foliation, rain event characteristics, and meteorology. Hydrological Processes. 22, 33-45.
https://doi.org/10.1002/hyp.6610
Sun, X., Onda, Y., Otsuki, K., Kato, H., Hirata, A., Gomi, T., 2014. The effect of strip thinning on tree
transpiration in a Japanese cypress plantation (Chamaecyparis obtusa Endl.) plantation.
Agricultural and Forest Meteorology. 197, 123-135. http://dx.doi.org/10.1016/j.
agrformet.2014.06.011
Sun, X.C., Onda, Y.C., Kato, H., Gomi, T., Komatsu, H., 2015. Effect of strip thinning on rainfall
interception in a Japanese cypress plantation. Journal of Hydrology. 525, 607-618.
http://dx.doi.org/10.1016/j.jhydrol.2015.04.023
47
Wallace, J., McJannet, D., 2008. Modelling interception in coastal and montane rainforest in northern
Queensland. Australia. Journal of Hydrology. 348, 480-495.
https://doi.org/10.1016/j.jhydrol.2007.10.019
Wang, Y., Wei, X.H., del Campo, A.D., Winkler, R., Liu, J.P., Li, Q., Liu, W.Y., 2019. Juvenile
thinning can effectively mitigate the effects of drought on tree growth and water consumption
in a young pinus contorta stand in the interior of british columbia, canada. Forest Ecology and
Management. 454, 117667. https://doi.org/10.1016/j.foreco.2019.117667
Wang, Y.Q., Shao, M.A., Shao, H.B., 2010. A preliminary investigation of the dynamic characteristics
of dried soil layers on the Loess Plateau of China. Journal of Hydrology. 381, 9-17.
https://doi.org/10.1016/j.jhydrol.2009.09.042
Weiss, M., Baret, F., 2014. CAN-EYE V6.313 User Manual.
http://www6.paca.inra.fr/can-eye/Documentation-Publications/Documentation
Yang, X.G., Chen, L., Wang, L., Wang, X., Gu, J.L., Qu, W.J., Song, N.P., 2019a. Dynamic
rainfall-partitioning relationships among throughfall, stemflow, and interception loss by
Caragana intermedia, Journal of Hydrology. 574, 980-989.
https://doi.org/10.1016/j.jhydrol.2019.04.083
Yang, X.L., Shao, M.A., Wei, X.R., 2019b. Stemflow production differ significantly among tree and
shrub species on the Chinese Loess Plateau. Journal of Hydrology. 568, 427-436.
https://doi.org/10.1016/j.jhydrol.2018.11.008
Yuan, C., Gao, G.Y., Fu, B.J., 2017. Comparisons of stemflow and its bio-/abiotic influential factors
between two xerophytic shrub species. Hydrology and Earth System Sciences. 21 (3),
1421-1438. https://doi.org/10.5194/hess-21-1421-2017
Yue, K., De Frenne, P., Fornara, D.A., Van Meerbeek, K., Li, W., Peng, X., Ni, X.Y., Peng, Y., Wu,
F.Z., Yang, Y.S., Pe?uelas, J., 2021. Global patterns and drivers of rainfall partitioning by
trees and shrubs. Global Change Biology. 00, 1-8. https://doi.org/10.1111/gcb.15644
Zhang, Q.F., Lv, X.Z., Yu, X.X., Ni, Y.X., Ma, L., Liu, Z.Q., 2022. Species and spatial differences in
vegetation rainfall interception capacity: A synthesis and meta-analysis in China. Catena. 213,
106223. https://doi.org/10.1016/j.catena.2022.106223
Zhang, S.Y., Li, X.Y., Jiang, Z.Y., Li, D.Q., Lin, H., 2018. Modelling of rainfall partitioning by a
deciduous shrub using a variable parameters Gash model. Ecohydrology. 11, e2011.
https://doi.org/10.1002/eco.2011
Zhang, Y.F., Yuan, C., Chen, N., Levia, D.F., Chen, N., 2023. Rainfall partitioning by vegetation in
China: A quantitative synthesis. Journal of Hydrology. 617, 128946.
https://doi.org/10.1016/j.jhydrol.2022.128946
48
Zhang, Y.F., Wang, X.P., Hu, R., Pan, Y.X., Paradeloc, M., 2015. Rainfall partitioning into throughfall,
stemflow and interception loss by two xerophytic shrubs within a rainfed re-vegetated desert
ecosystem, northwestern China. Journal of Hydrology. 527, 1084-1095.
https://doi.org/10.1016/j.jhydrol.2015.05.060
Zhang, Y.F., Wang, X.P., Pan, Y.X., Hu, R., 2020. Relative contribution of biotic and abiotic factors to
stemflow production and funneling efficiency: A long-term field study on a xerophytic shrub
species in Tengger Desert of northern China. Agricultural and Forest Meteorology. 280,
107781. https://doi.org/10.1016/j.agrformet.2019.107781
Zhang, Y.F., Wang, X.P., Pan, Y.X., Hu, R., Chen, N., 2021. Global quantitative synthesis of effects of
biotic and abiotic factors on stemflow production in woody ecosystems. Global Ecology and
Biogeography. 00, 1-11. https://doi.org/10.1111/geb.13322
|
|
