Changes in Climate or Vegetation Printer-friendly Version Interactive Discussion Do Changes in Climate or Vegetation Regulate Evapotranspiration and Streamflow Trends in Water-limited Basins? Hessd Changes in Climate or Vegetation Printer-friendly Version Interactive Discussion

Interactions between climate change, vegetation, and soil regulate hydrological processes. In this study, it was assumed that vegetation type and extent remained fixed and unchanged throughout the study period, while the effective rooting depth (Z e) changed under climate change scenarios. Budyko's hydrological model was used to 5 explore the impact of climate change and vegetation on evapotranspiration (E) and streamflow (Q) on the static vegetation rooting depth and the dynamic vegetation rooting depth. Results showed that both precipitation (P) and potential evapotranspiration (E p) exhibited negative trends, which resulted in decreasing trends for dynamic Z e scenarios. Combined with climatic change, decreasing trends in Z e altered the partitioning 10 of P into E and Q. For dynamic scenarios, total E and Q were predicted to be −1.73 and 28.22 %, respectively, greater than static scenarios. Although climate change regulated changes in E and Q, the response of Z e to climate change had a greater overall contribution to changes in hydrological processes. Results from this study suggest that with the exception of vegetation type and extent, Z e scenarios were able to alter wa-15 ter balances, which in itself should help to regulate climate change impacts on water resources.


Introduction
Partitioning of precipitation (P ) on land surfaces into evapotranspiraiton (E ) and streamflow (Q) reflects a hydrologic response to land use and climate forcing.Given that this impacts water availability globally (Xu et al., 2013), understanding such a process is critical for water resource management.In the past, a large number of studies have been conducted to quantify the impact of climate or vegetation changes on catchment water balances.Evidence shows that changes in climatic conditions, resulting from P and potential evapotranspiration E p , for example, have a considerable impact on Q (e.g., IPCC, 2007;Oudin et al., 2009;Liang et al., 2013).Additionally, changes in Figures land use type can dramatically alter water balances on different scales, such as the reduction in Q observed on the Loess Plateau, China, which the Grain for Green program has shown to exist.Thus, quantifying the impact of climate and vegetation change on Q remains a challenge in hydrological sciences.
Along with complex, physically based distributed hydrological models (bottom-up approach), a simple coupled water and energy balance model (top-down approach), such as Budyko's hydrological model, has attracted considerable attention in recent years (e.g., Zhang et al., 2001;Yang et al., 2007;Brümmer et al., 2012;Donohue et al., 2012).According to Budyko's assumption (1974), available water and energy are the primary factors that determine the rate of E , which also controls the partitioning of P into E and Q.Because the original version of Budyko's assumption only included climatic variables, an adjustable parameter has been incorporated into the model to reflect the influence of watershed characteristics (e.g., Fu et al., 1981;Choudhury, 1999;Zhang et al., 2001;Yang et al., 2007).Even though this watershed characteristics parameter (n) has been investigated by a number of studies (Zhang et al., 2001;Yang et al., 2009;Donohue et al., 2010), its relation to physical attributes remains obscure (Gerrits et al., 2009;Donohue et al., 2012;Liu and McVicar, 2012).By combining equations by Choudhury (1999)  This study employed the improved version of Budyko's hydrological model with the addition of an ecohydrological adjustable parameter to assess impacts of climate and vegetation changes on E and Q.By incorporating this adjustment parameter, Budyko's model can be expressed as follows (e.g., Choudhury's equation (1999)): where n is a catchment-specific parameter (dimensionless), which reflects the influence of catchment characteristics on the partitioning of P between E and Q.By combining the equation provided by Porporato et al. (2004) and Choudhury (1999) Ignoring changes in storage, the steady state water balance model can be expressed as follows (Donohue et al., 2011;Roderick and Farquhar, 2011;Liu and McVicar, 2012): through the Loess Plateau and the North China Plain before finally emptying into the Bohai Sea.Impacted by climate change, most YRB regions have exhibited decreasing P trends (Liu et al., 2008;McVicar et al., 2002;Nakayama, 2011).Along with climate change, alterations in physiological characteristics (e.g., changes in Z e ) can also have an impact on changes in hydrological processes, although little effort has been focused on this topic to date.In this study, it was assumed that vegetation type and extent were fixed and remained unchanged throughout the study period, while the effective rooting depth changed under the influence of climatic change.Two scenarios were developed for this study according to Budyko's assumption.For the static Z e scenario, the effective rooting depth for 1961 was fixed throughout a simulation period between 1961 and 2010.For the dynamic Z e scenario, the effective rooting depth was influenced by the specific water and energy conditions of each grid cell, in accordance with specific changes in climatic conditions.Data from the National Climate Center of the China Meteorological Administration (CMA) were used to investigate impacts of climate change on water resources.The Yellow River Conservancy Commission (YRCC) provided Q data between 1961 and 2010.Monthly E p was calculated by means of monthly wind speed, daylight hours, relative humidity, and average air temperature using the Penman equation (Shuttleworth, 1993).Normalized difference vegetation index (NDVI) data were obtained from the Global Land Cover Facility (http://www.glcf.umd.edu/), and were used to calculate the fraction of photosynthetically active radiation (PAR) absorbed by vegetation (f PAR ).

Changes to ecohydrological processes
YRB P and E p temporal trends  are provided in Fig. 1.On a basin scale, the average slope for P was −0.96 mm a −2 , with a range from −2.37 mm a −2 to 1.03 mm a −2 (Fig. 1a), while the average slope for E p was −0.13 mm a −2 , with a range Introduction

Conclusions References
Tables Figures

Back Close
Full from −3.38 to 1.47 mm a −2 (Fig. 1b).E p and P exhibited increasing trends, with an average increase of 0.004 mm a −2 .The vegetation fractions for trees (Fig. 1c) and grass (Fig. 1d) were calculated from f PAR (Donohue et al., 2009), which was necessary for calculating Z e .
According to conclusions that state that the higher the P the deeper the Z e (Schenk and Jackson, 2002;Donohue et al., 2012), Z e was calculated for YRB, a large waterlimited basin (data provided in Fig. 2).Averaged static Z e (Fig. 2a) (1961 was used to set the base condition of Z e ) ranged from 89 to 2245 mm, with an average of 381 mm, while averaged Z e throughout 1961-2010 ranged from 82 to 1818 mm.Z e was influenced by decreasing P trends, resulting in a decreasing trend with a slope of −0.12 mm a −2 .

Changes in streamflow and evapotranspiration
Modeled time series of E are provided in Fig. 3. Results showed similar trends to observed E (calculated using P − Q).Furthermore, the Nash-Sutcliffe model efficiency (NSE) coefficient reached up to 0.85 for the dynamic Z e scenario.As a result of the overestimation of high E (i.e., 1961 and 1964) and the underestimation of high E (i.e., 2002), modeled E under the dynamic Z e scenario exhibited a negative trend (−0.81 mm a −2 ), and thus was in opposition to observations (0.23 mm a −2 ).Modeled E under the static Z e scenario also exhibited a negative trend, with a slope of −0.78 mm a −2 .
Relative differences in modeled annual total Q and E between the static Z e and dynamic Z e scenarios are provided in Fig. 4. Results showed that (i) for the dynamic Z e scenario (Fig. 4a), total E was predicted to be 1.73 % smaller than the static Z e scenario, while, conversely, total Q (Fig. 4b) was predicted to be 28.22 % greater than the static Z e scenario, and (ii) decreasing trends were detected in most areas of the basin for E , while increasing trends were detected in most areas of the basin for Q.Introduction

Conclusions References
Tables Figures

Back Close
Full Temporal trends in E calculated from both dynamic and static Z e scenarios are respectively provided in Fig. 5a and b.Results showed that (i) temporal trends for both dynamic and static Z e scenarios exhibited similar spatial patterns; i.e., most areas of the basin contributed to negative (decreasing) trends, and only the northwestern area contributed to positive (increasing) trends in E , (ii) temporal trends in E under the dynamic Z e scenario ranged from −2.86 to 1.41 mm a −2 , with an average increase of −0.80 mm a −2 (Fig. 5a), while temporal trends in E under the static Z e scenario ranged from −2.70 to 1.41 mm a −2 , with an average increase of −0.81 mm a −2 (Fig. 5b), and (iii) significant trends (P < 0.05) in E under dynamic (Fig. 5c) and static Z e scenarios (Fig. 5d) showed similar patterns for YRB, while the extent per area showed significant increasing trends under the static Z e scenario (Fig. 5d).
The relative contribution of climate was mapped as the trend of modeled E under the dynamic Z e scenario divided by modeled E under the static Z e scenario for each grid cell (Fig. 6a) from which the relative contribution of vegetation was obtained (Fig. 6b).
Results showed that climate regulated temporal trends in E , while changes in Z e only contributed slightly to changes in E .Using the differential of E or Q to the variables (e.g., ∂E/∂P ) multiplied by changes in variables (e.g., dP ), the contribution of different variables can be obtained (∂E/∂P ×dP ).For example, following Donohue et al. (2012), the "typical variability" observed for each variable between 1961 and 2010 was determined (represented by the standard deviation of annual values).Results showed that (i) changes in P caused the greatest variability in E (or Q), generally followed by variability in Z e , α, and E p , (ii) changes in P contributed more to changes in E compared to Q, and (iii) summed contributions of climatic variables (P , E p , and α) to E and Q were larger than Z e , especially for E .Introduction

Conclusions References
Tables Figures

Back Close
Full

Discussion
As demonstrated by results from this study (Fig. 1), both P and E p showed decreasing trends, while E p /P showed a slight increasing trend (0.004 mm a −2 ) throughout the 1961-2010 period.Results were consistent with those reported by Liu and McVicar (2012).Decreasing trends in P (with an average increase of −0.96 mm a −2 ) or increasing trends in E p /P resulted in decreasing trends in Z e (data provided in Fig. 2).
Strong interactions are known to exist between climate, vegetation, and soil properties that lead to specific hydrologic partitioning on a catchment scale (Troch et al., 2013).Inevitably, it is the available water (P ) and energy (represented by E p ) that regulate vegetation patterns.Degradation in vegetation influenced by decreasing P has been reported in YRB (e.g., Xin et al., 2008).In particular, changes in vegetation extent and type (mainly resulting from human activity) are major causes of Q change (Li et al., 2007;Liu et al., 2009).For example, changes in vegetation patterns as a result of land use changes (e.g., such as determined by the Grain for Green program in the Loess Plateau) inevitably alter hydrological processes, and result in a decrease in Q (McVicar et al., 2007;Cao et al., 2011).On the one hand, numerous studies concluded that vegetation change was the main cause of hydrological process change, e.g., such as observed changes in Q in the Yiluo River basin (Liu et al., 2009).On the other hand, other studies reported that climate variability in certain regions influences surface hydrology more significantly compared to land use changes, e.g., such as observed in the Heihe River basin, China (Li et al., 2009), as well as the upper Mississippi River basin (Frans et al., 2013).
In combination with climate change, this study explored how vegetation impacts hydrological processes from an alternative aspect, i.e., assessing Z e response to climate change and its impacts on Q and E .According to Budyko's assumption, the greater the P , the deeper the rooting depth (Schenk and Jackson, 2002).Furthermore, modeled E under dynamic and static Z e scenarios exhibited negative trends, while observed E exhibited positive trends (Fig. 3).Vegetation structure profoundly regulates the annual Introduction

Conclusions References
Tables Figures

Back Close
Full surface hydrological cycle, and is a cause and consequence of surface water balances (Gentine et al., 2012).In this study, following Donohue et al. (2012), Z e in combination with α and κ was used to calculated n (1.81 average).Results were similar to n (n = 1.76) calculated from a nonlinear fitted model by Liu and McVicar (2012).Given that Z e data are scarce, validation of modeled Z e is difficult to obtain for larger basins (Donohue et al., 2012).From the n calculated for each grid cell, it was shown that simulated E and Q fitted well with observed values (provided in Fig. 3).Alteration of Z e contributed more to changes in Q and E (Fig. 4 and Table 1), which indicated that the response of vegetation to climatic change can alter the partitioning of P into E and Q. Budyko's hydrological model with the addition of ecohydrological parameters (such as Z e , α, and κ) captured affects of watershed characteristics on the partitioning of P into E and Q. Results can also reflect relative contributions provided in Fig. 5 and Fig. 6.For example, Fig. 5c and d show that significant temporal trends in E yield different results between dynamic and static Z e scenarios.Figure 6 shows that climate change regulates temporal trend changes in E , which are consistent with results provided in Table 1.Furthermore, given that soil, topography, vegetation, and climate are intrinsically interconnected, Gentine et al. 1.For YRB, both P and E p exhibited negative trends, while E p /P exhibited positive trends, resulting in a decreasing Z e trend under the dynamic Z e scenario throughout the 1961-2010 investigation.
2. Simulated E under the dynamic and static Z e scenarios exhibited negative trends, with an average increase of −0.81 and −0.78 mm a −2 , respectively.For the dynamic scenario, total E and Q were respectively predicted to be −1.73 and 28.22 % greater than the static scenario, which exhibited obvious spatial variation.
3 Full  Full Discussion Paper | Discussion Paper | Discussion Paper | and Porporato et al. (2004),Donohue et al. (2012) deduced the relationship between n and ecohydrological processes (such as storm depth α, plant-available soil water holding capacity κ, and the effective rooting depth Z e ), which offers new insight into understanding the response of hydrological processes to impacts of climate change and vegetation.While numerous studies have investigated impacts of climate and vegetation on hydrological processes in the past, few have explored impacts of vegetation on hydrological processes from the point of view of the response of vegetation to climate change.The objective of this study was therefore to explore the temporal trends in E and Q, and to assess the relative contribution of climate and vegetation change to E and Q.
) Equations (2) and (3) constitute the Budyko-Choudhury-Porporato model (BCP model).By incorporating ecohydrological parameters (Z e , α, and κ), this model can be used to assess the impact of climate change on E or Q.The Yellow River basin (YRB) is approximately 5400 km long, with a basin drainage area of 7.95 × 10 5 km 2 .Its headwaters originate from the Tibetan Plateau, flowing Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 3.3 Relative contribution of climatic and vegetation change on E and Q Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | (2012) attempted to use the Budyko curve to explain ecohydrological controls of soil water balances.Further research should focus more attention on mechanisms of watershed parameters and improve the accuracy of the Budyko's hydrological model, as it relates to different temporal and spatial scales.5 Conclusions Climate change and vegetation impacts on Q and E can be explored using Budyko's hydrological model with the addition of adjustable ecohydrological parameters (such as Z e , α, and κ).According to the "typical variability" of different variables, climate change and vegetation impacts were obtained for the dynamic and static Z e scenarios investigated.The following conclusions can be drawn from this study: Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Li, L. J., Zhang, L., Wang, H., Wang, J., Yang, J. W., Jiang, D. J., Li, J. Y., and Qin, D. Y.: Assessing the impact of climate variability and human activities on streamflow from the Wuding River basin in China, Hydrol.Process., 21, 3485-3491, 2007.Li, Z., Liu, W. Z., Zhang, X. C., and Zheng, F. L.: Impacts of land use change and climate variability on hydrology in an agricultural catchment on the Loess Plateau of China, J. Hydrol.Discussion Paper | Discussion Paper | Discussion Paper | Fig. 1 Temporal trends in P (a) and E p (b) (mm a -2 ), and the cover fraction for grass (c) and trees (d).

Figure 1 .
Figure 1.Temporal trends in P (a) and E p (b) (mm a −2 ), and the cover fraction for grass (c) and trees (d).

Fig. 4
Fig. 4 Modeled percent differences in mean annual total E (a) and Q (b) between static Z e (Z e for 1961 was fixed throughout the 1961-2010 simulation period) and dynamic Z e (Z e was influenced by specific water and energy conditions for each grid cell in accordance with specific climate change conditions).

Figure 4 .
Figure 4. Modeled percent differences in mean annual total E (a) and Q (b) between static Z e (Z e for 1961 was fixed throughout the 1961-2010 simulation period) and dynamic Z e (Z e was influenced by specific water and energy conditions for each grid cell in accordance with specific climate change conditions).

Figure 5 .Fig. 6
Figure 5. Temporal trends in E under dynamic Z e (a) and static Z e (b) scenarios, and regions exhibiting significant E slopes (P < 0.05) for dynamic Z e (c) and static Z e (d) scenarios as determined by the Mann-Kendall method.
. As anticipated, although climate change regulates changes in E and Q, Z e response to climate change contributed more to changes in hydrological processes for this water-limited region.Results indicated that with the exception of vegetation type and extent, Z e scenarios were able to alter the partitioning of P into E and Q. Introduction

Table 1 .
Summaries of E and Q sensitivities to changes in ecohydrological conditions throughout the study period.Shown for each variable and zone are E and Q sensitivity coefficients, observed variability per variable, and typical variability in E caused by driving variables.dZ e is the difference between tree and grass Z e modeled for the basin.