Issues in the Impacts of Climate Variability and Change on Agriculture

As the title of this special issue suggests, the commonality of the papers included herein is a focus on the interaction of agriculture with climate variability and change in the southeastern United States. The range of topics covered is considerable, from climate modeling to remote sensing to economics. All papers result from two major projects funded by NASA MTPE and the U.S. EPA NCQERA. In addition, the USDA-ERS funded part of one study concerning economics of agriculture under climate change.

We analyze the control runs and 2 × CO₂ projections (5-year lengths) of the CSIRO Mk 2 GCM and the RegCM2 regional climate model, which was nested in the CSIRO GCM, over the Southeastern U.S.; and we present the development of climate scenarios for use in an integrated assessment of agriculture. The RegCM exhibits smaller biases in both maximum and minimum temperature compared to the CSIRO. Domain average precipitation biases are generally negative and relatively small in winter, spring, and fall, but both models produce large positive biases in summer, that of the RegCM being the larger. Spatial pattern correlations of the model control runs and observations show that the RegCM reproduces better than the CSIRO the spatial patterns of precipitation, minimum and maximum temperature in all seasons. Under climate change conditions, the most salient feature from the point of view of scenarios for agriculture is the large decreases in summer precipitation, about 20% in the CSIRO and 30% in the RegCM. Increases in spring precipitation are found in both models, about 35% in the CSIRO and 25% in the RegCM. Precipitation decreases of about 20% dominate in winter in the CSIRO, while a more complex pattern of increases and decreases is exhibited by the regional model. Temperature increases by 3 to 5 °C in the CSIRO, the higher values dominating in winter and spring. In the RegCM, temperature increases are much more spatially and temporally variable, ranging from 1 to 7 °C across all months and grids. In summer large increases (up to 7 °C) in maximum temperature are found in the northeastern part of the domain where maximum drying occurs.

We use the CERES family of crop models to assess the effect of different spatial scales of climate change scenarios on the simulated yield changes of maize (Zea mays L.), winter wheat (Triticum aestivum L.), and rice (Oryza sativa L.) in the Southeastern United States. The climate change scenarios were produced with the control and doubled CO₂ runs of a high resolution regional climate model and a coarse resolution general circulation model, which provided the initial and lateral boundary conditions for the regional model. Three different cases were considered for each scenario: climate change alone, climate change plus elevated CO₂, and the latter with adaptations. On the state level, for most cases, significant differences in the climate changed yields for corn were found, the coarse scale scenario usually producing larger modeled yield decreases or smaller increases. For wheat, however, which suffered large decreases in yields for all cases, very little contrast in yield based on scale of scenario was found. Scenario scale resulted in significantly different rice yields, but mainly because of low variability in yields. For maize the primary climate variable that explained the contrast in the yields calculated from the two scenarios is the precipitation during grain fill leading to different water stress levels. Temperature during vernalization explains some contrasts in winter wheat yields. With adaptation, the contrasts in the yields of all crops produced by the scenarios were reduced but not entirely removed. Our results indicate that spatial resolution of climate change scenarios can be an important uncertainty in climate change impact assessments, depending on the crop and management conditions.

This study examines how uncertainty associated with the spatial scale of climate change scenarios influences estimates of soybean and sorghum yield response in the southeastern United States. We investigated response using coarse (300-km, CSIRO) and fine (50-km, RCM) scale climate change scenarios and considering climate changes alone, climate changes with CO₂ fertilization, and climate changes with CO₂ fertilization and adaptation. Relative to yields simulated under a current, control climate scenario, domain-wide soybean yield decreased by 49% with the coarse-scale climate change scenario alone, and by 26% with consideration for CO₂ fertilization. By contrast, the fine-scale climate change scenario generally exhibited higher temperatures and lower precipitation in the summer months resulting in greater yield decreases (69% for climate change alone and 54% with CO₂ fertilization). Changing planting date and shifting cultivars mitigated impacts, but yield still decreased by 8% and 18% respectively for the coarse and fine climate change scenarios. The results were similar for sorghum. Yield decreased by 51%, 42%, and 15% in response to fine-scale climate change alone, CO₂ fertilization, and adaptation cases, respectively — significantly worse than with the coarse-scale (CSIRO) scenarios. Adaptation strategies tempered the impacts of moisture and temperature stress during pod-fill and grain-fill periods and also differed with respect to the scale of the climate change scenario.

We examine the effect of climate scenarios generated using results from climate models of different spatial resolution on yields simulated by the deterministic cotton model GOSSYM for the southeastern U.S.A. Two related climate change scenarios were used: a coarse-scale scenario produced from results of a general circulation model (GCM) which also provided the boundary conditions to a regional climate model (RCM), from which a fine-scale scenario was constructed. Cotton model simulations were performed for three cases: climate change alone; climate change and elevated CO₂; climate change, elevated CO₂ and adaptations to climate change. In general, significant differences in state-average projected yield changes between the coarse and fine-scale scenarios are found for these three cases. In the first two cases, different directions of change are found in some sub-regions. With adaptation, yields substantially increase for both climate scenarios, but more so for the coarse-scale scenario (30% domain-average increase). Under irrigation, yield change differences between the two climate scenarios are small in all three cases, and yields are higher under irrigation (~35% domain-average increase with adaptation case) compared to dryland conditions. For the climate change alone case, differences in summer water-stress levels explain the contrasts in dryland yield patterns between the coarse and fine-scale climate scenarios.

The appropriate level of spatial resolution for climate scenarios is a key uncertainty in climate impact studies and regional integrated assessments. To the extent that such uncertainty may affect the magnitude of economic estimates of climate change, it has implications for the public policy debates concerning the efficiency of CO₂ control options. In this article, we investigate the effects that different climate scenario resolutions have on economic estimates of the impacts of future climate change on agriculture in the United States. These results are derived via a set of procedures and an analytical model that has been used previously in climate change assessments. The results demonstrate that the spatial scale of climate scenarios affects the estimates of both regional changes in crop yields and the economic impact on the agricultural sector as a whole. An assessment based on the finer scale climatological information consistently yielded a less favorable assessment of the implications of climate change. Regional indicators of economic activity were of opposite sign in some regions, based on the scenario scale. Such differences in economic magnitudes or signs are potentially important in examining whether past climate change assessments may misstate the economic consequences of such changes. The results reported here suggest that refinement of the spatial scale of scenarios should be carefully considered in future impacts research.

The purpose of the paper is to propose and test a new approach to simulating farmers’ agronomic adaptation to climate change based on the pattern of adoption of technological innovation/substitution over time widely described as a S-shaped (or logistic) curve, i.e., slow growth at the beginning followed by accelerating and then decelerating growth, ultimately leading to saturation. The approach we developed is tested using the Erosion Productivity Impact Calculator crop model applied to corn production systems in the southeastern U.S. using a high-resolution climate change scenario. Corn is the most extensively grown crop in the southeastern U.S. The RegCM limited area model nested within the CSIRO general circulation model generated the scenario. We compare corn yield outcomes using this new form of adaptation (logistic) with climatically optimized (clairvoyant) adaptation. The results show logistic adaptation to be less effective than clairvoyant adaptation in ameliorating climate change impacts on yields, although the differences between the two sets of yields are statistically significant in one case only. These results are limited by the reliance on a single scenario of climate change. We conclude that the logistic technique should be tested widely across climate change scenarios, crop species, and geographic areas before a full evaluation of its effect on outcomes is possible.

El Niño/Southern Oscillation (ENSO) is considered one of the most powerful forces driving anomalous global weather patterns. Large-scale seasonal precipitation and temperature changes influenced by ENSO have been examined in many areas of the world. The southeastern United States is one of the regions affected by ENSO events. In this study, remote sensing detection of vegetation response to ENSO phases is demonstrated with one-kilometer biweekly Normalized Difference Vegetation Index (NDVI) data (1989–1999) derived from the Advanced Very High Resolution Radiometer (AVHRR). The impacts of three ENSO phases, cold, warm and neutral, on vegetation were analyzed with a focus on two vegetation cover types, two seasons and two geographic regions within the southeastern U.S. Significant ENSO effects on vegetation were found in cropland and forest vegetation cover types based on image and statistical analysis of the NDVI data. The results indicate that vegetation condition was optimal during the ENSO neutral phase for both agricultural and natural vegetation.

Statistical methodology is devised to model time series of daily weather at individual locations in the southeastern U.S. conditional on patterns in large-scale atmosphere-ocean circulation. In this way, weather information on an appropriate temporal and spatial scale for input to crop-climate models can be generated, consistent with the relationship between circulation and temporally and/or spatially aggregated climate data (an exercise sometimes termed ‘downscaling’). The Bermuda High, a subtropical Atlantic circulation feature, is found to have the strongest contemporaneous correlation with seasonal mean temperature and total precipitation in the Southeast (in particular, stronger than for the El Niño-Southern Oscillation phenomenon). Stochastic models for time series of daily minimum and maximum temperature and precipitation amount are fitted conditional on an index indicating the average position of the Bermuda High. For precipitation, a multi-site approach involving a statistical technique known as ‘borrowing strength’ is applied, constraining the relationship between daily precipitation and the Bermuda High index to be spatially the same. In winter (the time of greatest correlation), higher daily maximum and minimum temperature means and higher daily probability of occurrence of precipitation are found when there is an easterly shift in the average position of the Bermuda High. Methods for determining aggregative properties of these stochastic models for daily weather (e.g., variance and spatial correlation of seasonal total precipitation) are also described, so that their performance in representing low frequency variations can be readily evaluated.