An Agricultural Reference Index for Drought
Abstract
A number of drought indices are available to quantify agricultural drought. But, no index exists that is simple, generic, and based on agriculturally important aspects, namely, soil, plant, and atmosphere (SPA). To account for these aspects, Agricultural Reference Index for Drought (ARID) was developed. Being generic and SPA-based, ARID is intended to work as a general indicator of water stress for all crops and estimate yield loss from drought. This study was conducted to evaluate ARID based on comparison with other drought indices and estimation of soil moisture, assess yield prediction ability of ARID-based yield models, and explore the potential of forecasting ARID. To evaluate ARID based on comparison with the water stress factor (WSPD) calculated by the Decision Support System for Agrotechnology Transfer, a widely-used crop model, and seven drought indices (Crop Moisture Index, Keetch and Byram Drought Index, Lawn and Garden Moisture Index, Palmer Drought Severity Index, Palmer Z-index, Standardized Precipitation Index, Water Requirement Satisfaction Index), average values of each index and WSPD were computed for three stages of maize for each of sixteen locations in southeast USA. Using the stage-wise average values, drought indices were compared with WSPD. To evaluate ARID based on soil moisture, daily values of ARID-estimated and observed soil moisture contents were compared for Citra, FL, with one year data. To assess ARID-yield relationship, ARID-based yield models were developed for cotton, maize, peanut, and soybean. Then, historical rainfed yields of these crops from several locations in the region were compared with the yields estimated by the ARID-based yield models. To explore the possibility of forecasting drought using ARID, ENSO and climate indices were used. While in the former method, monthly ARID values were separated into three ENSO categories and averaged by phases; in the latter, regression was done to estimate 1-, 2-, and 3-month lead values of ARID from the past values of some significant climate indices. Finally, prediction errors of each method were compared across months and five locations in the region. Among the drought indices compared, ARID showed the highest correlations and least errors with WSPD in all locations, suggesting that ARID can estimate water stress better than done by the other indices. Similarly, a good agreement was found between the ARID-estimated and observed soil moisture contents (d-index = 0.90), indicating ARID can provide good estimations for soil moisture. Also, the ARID-based yield models performed reasonably well (relative RMSE: 13 to 22%). The predicting abilities of the two methods varied depending on months and locations. For instance, the prediction errors of the climate index-based method for Jan, Feb, Aug, and Sep were less than those of the ENSO-based method for Liveoak, FL, suggesting the possibility of improved forecasting for these months with climate indices. To conclude, as ARID provided good estimations for soil water status and the ARID-based yield models made reasonable predictions of crop yields, ARID may be used as a reliable tool to quantify agricultural droughts. Applying a specific method for a specific month or location, ARID may be used as a potential tool to forecast drought.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2009
- Bibcode:
- 2009AGUFMNH41A1231W
- Keywords:
-
- 0468 BIOGEOSCIENCES / Natural hazards