Modeling the Impacts of Climate Change on Hydrology and Agricultural Pollutant Runoff in California's Central Valley
TABLE OF CONTENTS
Chapter I: Introduction........................................................................................................1
C hapter II: Climate Change Sensitivity Assessment of a Highly Agricultural Watershed Using SWAT.......................................................................................................................6
C hapter III: Sensitivity of Agricultural Runoff Loads to Rising Levels of CO2 and climate change in the San Joaquin Valley watershed of California..................................57
C hapter IV: Sensitivity of Groundwater Recharge Under Irrigated Agriculture to Changes in Climate, CO2 Concentrations, and Canopy Structure................................................101
C hapter V: Uncertainty Analysis and Multi-objective Calibration of Hydrology and Water Quality in the Sacramento River Watershed, California......................................141
C hapter VI: Climate Change Uncertainty Analysis on Streamflow and Agricultural Pollutant Transport Using a Stochastic Approach in California’s Central Valley.........186
C hapter VII: Conclusions...............................................................................................224
Chapter I: Introduction 1.1 Research introduction and objectives
There is a general scientific consensus that climate change is occurring caused in some part by greenhouse gas emissions. While many studies show this occurrence, a large amount of uncertainty still exists. The state of current climate change knowledge is well-expressed by the United States Environmental Protection Agency: “As with any field of scientific study, there are uncertainties associated with the science of climate change. This does not imply that scientists do not have confidence in many aspects of climate change. Some aspects of the science are known with virtual certainty, because they are based on well-known physical laws and documented trends. Current understanding of many other aspects of climate change ranges from ‘likely’ to ‘uncertain.’ “(EPA, 2010) This uncertainty has become an increasingly important topic for water resources managers, especially for California, where a large amount of the water used for agricultural and urban environments largely come from Sierra Nevada snowmelt. Within the past 10 years, evidence that climate change will have significant effects on California’s water resources has continued to accumulate. More than 150 peer-reviewed scientific studies have on climate and water in California has been published (Kiparsky and Gleick, 2003), addressing everything from General Circulation Model downscaling to potential aquatic ecosystem adaptation techniques due to a warmer climate.
While a substantial amount of climate change work has been done in California, more work needs to be done on the potential impacts of climate change on hydrology, agricultural water use, and agricultural pollutant fate and transport within the highly agricultural Central Valley of California. This is very pertinent, as nonpoint source pollution is a major cause of water quality impairment in the Central Valley. To address the relationship between climate change and agricultural pollutant fate and transport, the use of hydrologic and water quality models are required. There is a plethora of hydrologic models that are able to simulate agricultural non-point source pollution, with only a few those also simulate pesticide fate and transport. Those models include HSPF (Hydrologic Simulation Program Fortran; Bicknell et al., 1997), SWAT (Soil and Water Assessment Tool; Arnold et al., 1998), HYDRUS (Šimunek et al., 2005), PRZM (Plant Root Zone Model; Carsel et al., 1984) and AGNPS (Agricultural NonPoint Source model; Young et al., 1987). Of these models, SWAT is the only one that is setup to simulate the effects of climate change on agricultural watersheds. SWAT includes algorithms for predicting how CO 2 concentration, precipitation, temperature, and humidity affect plant growth, evapotranspiration (ET), snow, and runoff generation. SWAT, therefore, is an effective tool for investigating climate change effects. SWAT has also been validated throughout the world (Gassman et al., 2007). SWAT was used for all large-scale watershed modeling analyses in this dissertation. One chapter in this dissertation analyzes the effects of climate change on groundwater recharger under three irrigated crops (alfalfa, almonds, and tomatoes). HYDRUS (Šimunek et al., 2005), a one-dimensional numerical model package was used to simulate the processes of water flow, root water uptake, root growth and evaporation
from the soil surface in one-dimensional variably saturated media was used for this chapter. This study will therefore address the following questions:  How will the hydrologic cycle (streamflow, surface runoff, soil interflow, groundwater recharge, evapotranspiration) change with increasing temperature and atmospheric CO 2 concentrations and changing precipitation rates for the San Joaquin and Sacramento River watersheds?  How will irrigation water use change with increasing temperature and atmospheric CO 2 concentrations and changing precipitation rates for the San Joaquin River watershed?  How agricultural pollutant fate and transport (sediment, nitrate, phosphorous, chlorpyrifos, and diazinon) change with increasing temperature and atmospheric CO 2
concentrations and changing precipitation rates for the Sacramento and San Joaquin River watersheds? 1.2 Dissertation organization
This dissertation is organized in to five major chapters. Chapter II focuses on the effects of climate change on the hydrologic cycle and irrigation water use in the San Joaquin River watershed using sensitivity scenarios. Chapter III is similar to Chapter II, but focuses on the effects of climate change on agricultural fate and transport. Chapter IV focuses on the effects of groundwater recharge under three irrigated crops (alfalfa, almonds, and tomatoes) in the San Joaquin Valley. Chapter V focuses on calibration and uncertainty analysis of the Sacramento River watershed SWAT model. Chapter VI uses a
novel method of stochastic climate analysis on the Sacramento and San Joaquin River watersheds. The entire dissertation is then summarized in Chapter VII. 1.3 References cited Arnold, J.G., Srinivasan, R, Muttiah, R.S., Williams J.R., 1998. Large area hydrologic modeling and assessment part I: model development. Journal of the American Water Resources Association 34(1), 73–89. Bicknell, B.R., J.C. Imhoff, J.L. Kittle, A.S. Donigian, Johanson, R.C., 1997. Hydrological simulation program- FORTRAN: User's Manual for Version 11 EPA/600/R-97/080 Research Triangle Park, North Carolina. Carsel, R.F., C.N., Smith, L.A., Mulkey, J.D., Dean, P., Jowise 1984. Users Manual for the Pesticide Root Zone Model (PRZM) Release 1. Gassman, P., Reyes, M., Green, C., Arnold, J., 2007. The Soil and Water Assessment Tool: Historical development, applications and future research directions. Transactions of the American Society of Agricultural and Biological Engineers 50, 1211-1250. Kiparsky, M., Gleick, P.H., 2003. Climate Change and California Water Resources: A Survey and Summary of the Literature. Pacific Institute for Studies in Development, Environment, and Security. Šimunek, J., van Genuchten, M.Th., Sejna, M., 2005. The HYDRUS-1D Software Package for Simulating the Movement of Water, Heat, and Multiple Solutes in Variably Saturated Media, Version 3.0. Department of Environmental Sciences University of California Riverside, Riverside, California, USA, 270 pp.
U.S. Environmental Protection Agency (EPA), 2010. Climate Change: State of Knowledge. http://www.epa.gov/climatechange/science/stateofknowledge.html Young, R.A., C.A. Onstad, D.D. Bosch, and W.P. Anderson. 1987. AGNPS: Agricultural non-point source pollution model: a water analysis tool, Washington DC.
1. Chapter II: Climate Change Sensitivity Assessment of a Highly Agricultural Watershed Using SWAT
2.1 Introduction Fossil fuel consumption has caused an increase in anthropogenic emissions of carbon dioxide (CO 2 ) and other greenhouse gases (IPCC, 2007). Due to higher concentrations of these gases in the atmosphere, the proportion of solar radiation hitting the Earth that is reflected back into space is reduced, leading to a net warming of the planet (Kalnay and Cai, 2003). Based on the range of emission scenarios presented to the Intergovernmental Panel on Climate Change (IPCC; IPCC, 2007), CO 2 concentrations are expected to increase from the present day concentration of approximately 330 pm to between approximately 550 and 970 ppm. The magnitude of this increase will depend on future human activities, as well as technological and economic development. For all IPCC scenarios, however, General Circulation Models (GCMs) predict that increases in atmospheric greenhouse gas concentrations will raise surface temperatures. These changes will likely affect the hydrologic cycle. Among the GCMs and emission scenarios used by the IPCC, temperatures in 2100 are expected to be between 1.1 and 6.4°C higher than temperatures in 1900, accompanied by changes in rainfall intensity and amount (IPCC, 2007). Possible changes in regional and seasonal patterns of temperature and precipitation and their implications for the hydrologic cycle are as yet poorly understood. An increase of atmospheric CO 2 will directly affect plant transpiration and growth which are inherently tied to the hydrologic cycle. Experimental evidence indicates that
stomatal conductance of some plants will decline as atmospheric CO 2 increases, resulting in a reduction of transpiration (e.g., Morison and Gifford, 1983; Morison, 1987; Hendry et al., 1993; Tyree and Alexander, 1993; Field et al., 1995; Saxe et al., 1998; Wand et al., 1999; Medlyn et al., 2001; Wullschleger et al., 2002). Research has also shown that total leaf area of many plant types may increase with increased atmospheric CO 2
concentrations (e.g., Wand et al., 1999; Prichard et al., 1999; Saxe et al., 1998), potentially offsetting the reduction of stomatal conductance. Much research has been done to elucidate the effects that climate change and increased atmospheric CO 2 concentration will have on watershed processes. Studies have reported that an increase in CO 2 while holding temperature and precipitation constant will cause increases in water yield (e.g., Aber et al., 1995; Fontaine et al., 2001; Chaplot, 2007). Using present day precipitation patterns, studies have shown that higher temperatures lead to increased evaporation rates, reductions in stream flow, and increased frequency of droughts (e.g., Schaake, 1990; Rind et al., 1990, Nash and Gleick, 1991, 1993). Labat et al. (2004) demonstrated that a temperature increase by 1°C may lead to a global runoff increase by 4% due to increased oceanic evaporation. Kamga (2001) used WatBal, a hydrologic water balance model (Yates, 1996), to show that a 1 and
3°C temperature increase and a 4 to 13% change in rainfall intensity would result in variations in annual river fluxes of -3% to +18% in Cameroon. In Africa, Legesse et al. (2003) used the Precipitation-Runoff Modeling System (PRMS) model (Leavesley, 1983) to simulate runoff, predicting a runoff decrease by 30% in response to a 10% decrease in precipitation amount. A 1.5°C increase in air temperature resulted in a runoff decrease of 15%. In a study of climate change effects on the Missouri River in the USA, Lettenmaier
et al. (1999) used output from three transient GCMs and one double CO 2 GCM to evaluate potential effects of climate change on water resources. They estimated that the Missouri River would experience a reduction in stream flow between 6% and 34%, which would greatly impact economic infrastructure along the river. All of these studies indicate that watershed processes may be very sensitive to changes in precipitation, temperature and increased atmospheric CO 2 concentrations. Despite many studies on the effects of climate change, up-to-date quantitative information on the effects of changes in precipitation and temperature on soil and water resources is still scarce. Anticipating changes in the hydrologic cycle is particularly important for regions with limited water supplies such as the San Joaquin Valley in California. This study will contribute to the collection of studies that characterize potential climate change impacts on water resources in California’s Central Valley (e.g., Gleick, 1987; Lettenmaier and Gan, 1990; USBR, 1991; Dracup and Pelmulder, 1993; USEPA, 1997, Miller et al., 1999; Wilby and Dettinger, 2003; Knowles and Cayan, 2002; Zhu et al., 2005). Most of these early studies were subject to the underlying assumption that precipitation would not be affected by regional warming which, based on multiple GCM outputs, may not be accurate (Allen et al, 2002). While all GCM model runs predict rising temperatures for California, the magnitude and direction of changes in precipitation is much less certain (Cayan, 2008). Recent studies incorporate precipitation projections from GCMs, downscaled to a higher resolution for California (e.g., Hay et al., 2000; Miller et al., 2000; Brekke et al., 2004; Dettinger, 2004; Van Rheenan, 2004; Maurer and Duffy, 2005; Maurer, 2007). These studies show great variability in projected precipitation for California. A large
amount of uncertainty of global precipitation is caused by the structure of GCMs and their underlying assumptions (IPCC, 2001). Consequently, no global climate model should be considered superior to others in predicting California precipitation. Any precipitation projections produced for California under the IPCC CO 2 scenarios should therefore be regarded as equally plausible. This study will thus consider a range of possible precipitation scenarios. Hydrologic models are often combined with climate scenarios generated from GCMs to produce potential scenarios of climate change effects on water resources. These hydrologic models provide a link between climate changes and water yields through simulation of hydrologic processes within watersheds. Hydrologic models then allow various simulations to be performed based on user needs. Confidence in the results varies greatly and largely depends on the methods and structure of the climate scenario and the hydrologic model. The Soil and Water Assessment Tool (SWAT) (Arnold et al., 1998) was used for this study. SWAT includes approaches describing how CO 2 concentration, precipitation, temperature and humidity affect plant growth, ET, snow, and runoff generation, and has often been used as a tool to investigate climate change effects. Several case studies of climate change impacts on water resources have used this model (e.g., Hanratty and Stefan, 1998; Rosenberg et al., 1999; Cruise et al., 1999; Stonefelt et al, 2000; Fontaine et al., 2001; Eckhardt and Ulbrich, 2003; Chaplot, 2007; Schuol, 2008). SWAT has been used to model portions of the San Joaquin watershed (Flay et al., 2000; Luo et al., 2008). The objective of this study was to provide a first estimate of the overall impact of climate on the hydrology of the San Joaquin River watershed, including its impact on irrigation water use by local farmers.
It is important to note that an assessment of the sensitivity of a model to climate change does not necessarily provide a projection of the likely consequences of climate change. However, such studies provide valuable insights into the sensitivity of the hydrological systems to changes in climate (Arnell et al., 2001). Wolock and McCabe (1999) also stated that sensitivity studies of temperature and precipitation variations can provide important information regarding the responses and vulnerabilities of different hydrologic systems to climate change, especially in the light of substantial uncertainty of GCM climate projections. The specific objectives of this study were to investigate the sensitivity of hydrologic variables, such as ET, water yield (in this case, synonymous with surface runoff and soil water interflow entering the adjacent stream), irrigation water use and stream flow (rate of streamflow at the watershed outlet) to climate change. To this end, we computed all hydrologic variables for 16 climate change scenarios (6 for present-day scenarios, 5 for the B1 emissions scenario (low CO 2 concentration), and 5 for the A1FI emission scenario (high CO 2 concentration)) and compared the results to a 50- year baseline scenario with a present-day climate. 2.2 Materials and Methods 2.2.1 Description of the Study Area The San Joaquin River watershed was selected for this study (Figure 2.1). According to the United States Geological Survey (USGS), the watershed consists of four hydrologic units: the middle San Joaquin and lower Chowchilla watersheds (identified by the eight-digit hydrologic unit code 18040001), the middle San Joaquin, lower Merced, and lower Stanislaus River watersheds (18040002), the upper Chowchilla and upper Fresno River watersheds (18040007), and the Panoche and San Luis Reservoir
watersheds (18040014).The USGS monitoring site at Vernalis (#11303500; Figure 2.1) was chosen as the outlet for the entire watershed. The discharge inlets of the upper San Joaquin, upper Merced, upper Tuolumne, and upper Stanislaus Rivers were defined at the USGS monitoring sites of #11251000, #11270900, #11289650, and #1130200, respectively (Figure 2.1). The total area of the watershed is 14 983 km 2 , with approximately 66% of the total area in the San Joaquin Valley, 15% in the Coastal Range, and 19% in the Sierra Nevada mountains. The watershed is highly agricultural and includes the majority of agricultural areas in the counties of Stanislaus, Merced, and Madera, and part of San Joaquin and Fresno Counties. Of the total cropland in the study area, 38% is covered by fruits and nuts, 36% by field crops (corn, tomatoes, pumpkins, watermelon, asparagus, cotton, beans, etc.), 17% by truck, nursery, and berry crops and 4% by grain crops (DWR, 2007). Agricultural pollution has become a major concern for the watershed (e.g., Foe, 1990; Foe and Connor, 1991; Crepeau et al., 1991; Foe and Sheipline, 1993; Kratzer, 1999). Researchers at the University of California at Davis found that pesticide contamination at 48 percent of the 237 San Joaquin watershed sampling sites tested exceeded the environmental safety or public health standard maintained by the State of California (CRWQCB, monitoring data). According to an assessment conducted by the California Regional Water Quality Control Board, at least 127 miles of the San Joaquin River were severely polluted by toxic metals, numerous pesticides, or additional chemicals that promote the growth of algae (SWRCB, 2002). Many portions of the San Joaquin River are listed as Impaired and Threatened Waters by the United States Environmental Protection Agency (USEPA, 2008, Section 303(d)).
The San Joaquin Valley has a Mediterranean climate with hot, dry summers and cool, wet winters. Average rainfall is approximately 200 to 300 mm with most of the precipitation falling during the period of November through April and negligible rainfall from May to October. Mean daily temperature is approximately 15°C (NOAA, 2008). Due to the arid climate, agriculture in the San Joaquin Valley critically depends on irrigation. Farmers in the San Joaquin Valley use a combination of groundwater and surface water to meet their irrigation needs. Irrigation water is mostly developed and delivered by governmental institutions, such as the State Water Project and the Central Valley Project, which sell long-term water contracts. Several irrigation districts such as Modesto or South San Joaquin then deliver the water to the end user via irrigation canals and aqueducts. Farmers manage their own groundwater usage, which to date has not been regulated. 2.2.2 The SWAT Hydrological Model SWAT is a hydrologic/water quality model developed by the United States Department of Agriculture–Agricultural Research Service (USDA-ARS) (Arnold et al., 1998). The main objective of SWAT is to predict the impact of agricultural or land management on water, sediment and agricultural chemical yields in ungauged basins. The model is a continuous-time, spatially distributed simulator of the hydrologic cycle and agricultural pollutant transport at a catchment scale. It runs on a daily time step. Major model components are weather conditions, hydrology, soil properties, plant growth and land management, as well as loads and flows of nutrients, pesticides, bacteria and pathogens. A detailed description of SWAT can be found in Nietsch et al. (2005).
In SWAT, a watershed is divided into multiple subwatersheds which are then divided into units of unique soil/land-use characteristics called hydrological response units (HRUs). These HRUs are defined as homogeneous spatial units characterized by similar geomorphologic and hydrological properties (Flugel, 1995). In SWAT, HRUs are composed of a unique combination of homogeneous soil properties, land use and slope. For example, a specific HRU land unit may contain a sandy loam, walnut orchards, and a slope of 5%. User specified land cover, soil area and slope thresholds can be applied that limit the number of HRUs in each subwatershed. For this study, only land use, soil properties and slopes that comprise over 5% of the subbasin were used for HRU definition. HRU water balance is represented by five storage components: canopy interception, snow, soil profile, shallow aquifer, and deep aquifer. Flow generation is summed across all HRUs in a subwatershed and the resulting flows are then routed through channels, ponds, and/or reservoirs to the watershed outlet Groundwater and lateral flow are treated the same way, where the amount of water entering the stream through the soil profile or streambed are summed across all HRUs. Due to the non-spatial component of the HRU, the assumption is that there is no interaction between HRUs regarding lateral flow and groundwater return flow. This assumption is not realistic due to the large-scale interactions of lateral and groundwater flow. Many studies, however, have found SWAT to be suitable for simulation lateral and groundwater return flow (Gassman et al., 2007). Predictions of surface runoff from daily rainfall are estimated based on a similar procedure as the CREAMS runoff model (Knisel, 1980). The runoff volume is estimated using the modified SCS curve number method (SCS, 1984), a value that incorporates soil,
land-use, and management information. The curve number is adjusted at each time step based on the amount of soil water present. The plant growth component of SWAT utilizes routines for plant development based on plant-specific input parameters summarized in the SWAT plant growth database. From these parameters, SWAT computes plant growth output characteristics such as biomass and leaf area index (LAI). The heat unit theory is used to regulate the plant growth cycle (Boswell, 1926; Magoon and Culpepper, 1932). In this theory, predictions of plant development can be estimated based on the amount of heat absorbed by the plant. Potential plant growth is calculated at each time step of the simulation and is based on growth under ideal growing conditions consisting of adequate water and nutrient supply and a favorable climate. In SWAT, irrigation may be scheduled by the user or automatically applied in response to a water deficit in the soil. In this study, irrigation in an HRU was automatically simulated by SWAT based on plant water stress. Depending on the subwatershed, irrigation water was either extracted from the nearby reach or a source outside the watershed. For a given irrigation event, SWAT determines the amount of water available in the source (a stream or river), and the amount of available water is compared to the amount of water needed for the specific irrigation event. Water applied to an HRU is used to fill the soil layers up to field capacity beginning with the soil surface layer and working downward (Nietsch et al., 2005). Due to the lack of data on actual irrigation water use for the study area and simulation period, the automatic irrigation operation in the SWAT was activated to estimate water amount delivered from streams or external sources and used in irrigation. The irrigation algorithm in the SWAT
model limited the irrigated water amount by the soil field capacity, implying an assumption of high efficiency in water use and water diversion. Also, irrigation water was only extracted from surface water and not groundwater. These assumptions might underestimate the agricultural drainage towards streams during intensive irrigation seasons. SWAT was used because of its capability to model the impacts of future climate conditions. For example, the calculation of ET takes into account variations of radiation- use efficiency, plant growth and plant transpiration due to changes in the atmospheric CO 2 concentrations, which is essential for any study of CO 2 -induced climate change. SWAT allows adjustment terms such as the CO 2 concentration to vary so that the user is able to incorporate GCM projections of atmospheric greenhouse gas concentrations and temperatures into the model simulations. However, SWAT does not allow incremental increases of atmospheric CO 2 concentration. The impact of the increase of plant productivity and the decrease of plant water requirements due to increasing CO 2 levels are considered following Nietsch et al. (2005). For estimation of ET, the Penman- Monteith method must be used for climate change scenarios that account for changing atmospheric CO 2 levels. This method has been modified in SWAT to account for CO 2
impacts on ET levels. 2.2.3 Implications of CO 2 Assumptions in SWAT Many studies using a wide range of plant species have confirmed that increased atmospheric CO 2 concentrations will result in a reduction of leaf stomatal conductance (e.g., Morison and Gifford, 1983; Morison, 1987; Hendry et al., 1993; Tyree and Alexander, 1993; Field et al., 1995; Saxe et al., 1998; Wand et al., 1999; Medlyn et al,