This paper begins with a review of regional impacts of climate change to California agriculture

Prior to Smith and Mendelsohn , several notable studies examined the state of the knowledge of climate assessments at the US level . In particular, Lewandrowski and Schimmelpfennig integrate the knowledge from both programming and econometric studies of the agricultural sector. Other reviews have focused on the technical details of the different modeling approaches without discussing the results of the various studies . Following the pioneering work of Smith and Mendelsohn , this paper also focuses on California. The state is a leader in agricultural production, with $53.5 billion in sector cash receipts in 2014. California accounts for roughly 2/3 of US fruit/nut production, and 1/3 of US vegetable production . Roughly 1/3 of California cropland, or 9 million acres, is irrigated , making the state’s agricultural sector highly vulnerable to changes in groundwater and surface water supply . Several programming and econometric studies have been published after Smith and Mendelsohn , that operationalize the concept of adaptation . It is followed by a review of the results from recent programming and econometric studies.

The final section synthesizes the results from these studies, addressing lessons learned about vulnerability,adaptation,plastic plant pots and adaptive capacity; and how these relate to economic welfare and efficiency.Observational studies indicate that average daily temperature and daily minimum temperatures, particularly during the winter season, have increased in California . Average daily temperature in the US Southwest for the previous decade has been higher than any decade observed in the previous century . Barnett et al. find that daily minimum temperatures in winter have increased between 0.28– 0.43 C per decade from 1950–1999. Not just magnitude, but an increased rate of warming has been observed. Karl et al. suggest that the US Southwest has experienced the most rapid rate of warming in the nation. Observed precipitation patterns are fundamentally more complex and variable than temperature, exhibiting a high degree of variability across space and time. Trenberth et al. indicate that annual precipitation has decreased in the southwestern United States for the period 1901–2005. Consistent with scientific theory, empirical research suggests that warmer climates, such as those projected for the Southwest, will lead to more extreme precipitation intensity and frequency , particularly during the winter season . Since annual precipitation is projected to decline , more extreme events do not translate into higher total rainfall for a given year. Instead, it is projected that light precipitation — an important source for soil moisture and groundwater recharge — will concomitantly decline.

Between 1901 and 2010, the areal extent of drought increased in the southwestern United State . Some have attributed the increasing expanse of drought, particularly in the previous decade, to warmer temperatures . Others have suggested that it is due to changes in atmospheric circulation . In addition to temperature and precipitation, CO2 fertilization is another climate change pathway affecting agriculture. Increased atmospheric carbon dioxide stimulates photosynthesis, leading to increased plant productivity and decreased water and nutrient use . Benefits from elevated CO2 concentrations depend upon plant type and irrigation level. C3 photosynthetic plants will benefit more than C4 plants , and dryland cropping systems will benefit more than irrigated systems . The extent to which CO2 fertilization mitigates climate-induced water scarcity in the field still lacks scientific consensus, and there is debate on the extent to which simulating CO2 effects actually reproduces the results in free air carbon dioxide enrichment experiments . Agricultural impacts from climate change are rooted in complex pathways. Assessments of crop impacts due to climatic change fall under two, broad categories: process-based and statistical models. Process-based models simulate physiological development, growth and yield of a crop on the basis of interaction between environmental variables and plant physiological processes . Statistical crop models impute a relationship between historic crop yield and climate variables, often in order to project the impact on yield under future climate scenarios.

Process-based models remain the gold standard in crop modeling as one is able to study the relationship between weather and all phases of crop growth in a range of weather possibilities, even those lying outside the historical record . California field crops have been modeled using DAYCENT . Both studies highlight resilience of alfalfa yield under A2 scenario by end of the century, whereas 5 other crops exhibit a decline. Jackson et al. also find alfalfa yield to be particularly resilient to early and repeated heat waves during May–July. Lee et al. also run climate projections with and without a CO2 fertilization effect on seven field crops in the Central Valley of California. They assume a CO2 increase of 350 ppmv from 1990 levels enhances net primary production by 10% for all crops except alfalfa and maize. They find that CO2 fertilization increases crop yields 2–16% above the model without CO2 effects under the high-emissions scenario by the end of the 21st century. There is a much smaller yield increase under the low-emissions scenario. Lobell and Field use two estimation methods in studying the effects of temperature and precipitation on perennial crop yields. Their model includes 72 potential weather predictor variables for each crop, such as monthly averages for max and min temperature and their corresponding squares. They find that cherries and almonds are harmed by future warming out of a set of 20 perennial crops in their analysis. Crop-level adaptations — such as adjusting the planting and harvesting date , and substituting between different crop varieties — have been included to a limited extent in crop models. However, these cannot account for the broad range of decision making at the farm-level under which many of the negative effects of climate change could be partially offset with input and output substitutions, improving information, and effective water institutions. Thus, economic models are necessary to capture a broader range of responsive decision-making as the climate changes.Recently, adaptations specific to California agriculture have been studied using three economic programming models: the Statewide Agricultural Production model, Central Valley Production Model , and the US Agricultural Resources Model . Capturing the decision-making process is an important part of modeling. In programming models,blueberry pot the farmer’s decision is captured by the objective function. The main decision variable in these models is acres of land allocated to a region-specific crop mix. The farmer responds to reductions in water availability and yield by adjusting crop acreage. Exogenous adaptations include institutional , socioeconomic , and technological change . Calibration through positive mathematical programming also captures decision-making by preserving observed crop mix allocation decisions . SWAP employs a PMP cost function to the capture the decision of bringing an additional unit of land into production . Both CVPM and USARM have also been calibrated using PMP . CVPM studies have also generated synthetic crop share data from Monte Carlo runs using a base water supply and groundwater depth with random perturbations. Crop adaptation equations are then derived from a multinomial logit regression of this CVPM-generated synthetic crop share data . In order to represent climate-induced changes in water supply, many mathematical programming models are linked to hydrological management models, such as the California Value Integrated Network , Water Evaluation and Planning , CalSim-II, and C2VSim. CALVIN is a generalized network flow-based optimization model that minimizes economic operating and scarcity costs of water supply, subject to water balance, capacity, and environmental constraints for a range of operational and hydrologic conditions .

CALVIN has the potential to incorporate several basin-level adaptations to water allocation rules such as contract changes, markets and exchanges, water rights, pricing, and water scarcity levels. However, it has limited ability to represent important physical phenomena, such as stream-aquifer interactions and groundwater flow dynamics under different climate and water management scenarios . WEAP has many of the same water management features as CALVIN and CalSim-II. WEAP includes demand priorities and supply preferences in a linear programming framework to solve the water allocation problem as an alternative to multi-criteria weighting or rule-based logic. It is different because analysis in the WEAP framework comes directly from the future climate scenarios and not from a perturbation of historical hydrology as with the other models. Unlike CALVIN and CalSim-II, WEAP only has a simplified representation of the rules guiding the State Water Project and Central Valley Project systems . CalSim-II is also very similar to CALVIN and WEAP . C2VSim is a multi-layer, distributed integrated hydrologic model that could represent pumping from multiple aquifer layers, effects on groundwater flow dynamics, and stream-aquifer interaction . Recent programming studies focus on how certain adaptations may affect costs under relatively extreme cases of water scarcity. These studies thus assess how these adaptations may offset costs under worst-case-scenarios of water supply reductions. Given that reduction in statewide agricultural water use due to the current drought is estimated at 6% , studies on 40–70% flow reduction should be interpreted with caution. The subsequent studies are organized according to magnitude of water supply/flow reduction. Studies on 5–6% reduction in water supply reveal the heavy fallowing and groundwater use . Howitt et al. find that a 6.6 maf deficit in surface water caused by the current drought is largely substituted by 5.1 maf of additional groundwater. This is estimated to cost an additional $454 million in pumping. In addition to over-pumping groundwater, farmers adjust by fallowing crop land. The overwhelming majority of the 428,000 acres estimated fallowed in 2014 are in the Central Valley, where the majority of fallowed acres belong to field crops. However, they project that fallowing will decrease by 43% by 2016, suggesting a trend toward stabilization. Frisvold and Konyar use USARM to examine the effects of a 5% reduction in irrigation water supply from the Colorado River on agricultural production in southern California. In particular, they are able to compare the potential value-added of additional adaptations that include changing the crop mix, deficit irrigation, and input substitution to a “fallowing only” model. They find that these additional adaptations have the potential to reduce costs of water shortages to producers by 66% compared to the “fallowing only” model.1 Medellin-Azuara et al. examine the extent to which more flexible2 versions of California water markets could reduce water scarcity costs under a 27% statewide reduction in annual stream flow. They compare agricultural water scarcity in the year 2050 under two scenarios: 1. Baseline: population growth and resulting levels of agriculture to urban land transfer, 2. Warm-dry: includes population pressure and climatic changes under GFDL CM2.1 A2. Under the warm-dry scenario, even with optimized operations, water scarcity and total operational costs increase by $490 million/year, and statewide agricultural water scarcity increases by 22%. If water markets are restricted to operate only within the four CALVIN sub-regions, statewide water scarcity costs increase by 45% and 70% for the baseline and warm-dry scenarios, respectively. Marginal opportunity costs of environmental flows increase under the warm-dry scenario, with particularly large percentage increases for the Delta Outflow and American River. Medellin-Azuara et al. conduct a similar analysis, adding the comparison with a warm-only 2050 scenario. The agricultural sector water scarcity costs rise by 3% from the baseline to warm-only scenario, versus an increase of 302% from the baseline to the warm-dry scenario.3 Indeed the greater hydrological impact of the warm-dry scenario results in significantly greater scarcity costs than the warm-only scenario. Using the CALVIN model runs from Medellin-Azuara et al. , MedellinAzuara et al. analyze adaptations at the farm-level, including adjustments in crop acreage , and to a more limited extent, yield-enhancing technology . Similar to the 2008 paper, the model compares economic losses between a baseline scenario and a warm-dry scenario . Results reveal an anticipated decline in acreage of low-value crops , which is particularly severe due to the large reduction in water availability. For example, pasture acreage is reduced by 90% across 3 out of 4 agricultural regions. The results also suggest that statewide agricultural revenues decline at a proportionately lower level than the reduction in water availability . Their model also captures the complexity between crop demand and climate-induced supply reduction. Although the demand for high-valued orchard crop increases, production decreases due to the negative impact on yield from temperature increases.The resulting price increase cannot compensate for the decrease in supply, and gross revenue still declines. Two studies examine the impacts of more extreme reductions in water supply .