Combining Seasonal Climate Scenarios with Forage Production Models to Inform Drought Decision-Making

Abstract:

Drought is one of the costliest natural disasters, impacting various sectors including agriculture. It lowers the productivity and livestock carrying capacity of grasslands, negatively impacting livestock operations and rural economics. Drought planning is crucial for livestock producers in areas of high interannual forage production variability, especially in semi-arid areas. However, a high uncertainty in seasonal climate outlooks can make drought planning challenging. Other disturbances like fire or overgrazing also have a significant effect on the amount of forage produced. When using satellite remote sensing to monitor the weather impacts on annual forage production, non-weather effects need to be minimized. We developed a forage production model using a data-driven regression tree method to estimate the amount of annual forage produced under specific weather conditions while separating the effects of other disturbances. The model uses environmental and weather variables and the remote sensing-based Normalized Difference Vegetation Index as a proxy for grassland productivity. The method was initially developed for Nebraska Sandhills upland grasslands (validation on hold-out data R² = 0.86) and has been recently applied to other semi-arid grasslands in the western U.S. The development process and validation of the forage model in various geographic locations will be introduced as well as the annual maps of forage production in years 2000 to 2018. These maps can be created only after the end of the growing season. Therefore, to help with drought planning and timely management decisions we developed a method that produces scenarios of expected annual forage production early in the growing season. Specifically, we used historical weather data for summer season and created representative maps of abnormally low, normal, and abnormally high temperature and precipitation. Using these maps, we created summer season climate scenarios that combine these variables (e.g. cool and dry, warm and wet). These scenarios, in combination with up-to-date spring weather, served as an input for the aforementioned forage production model to create estimates of the end-of-season forage production. When provided on an operational basis, these results can serve as a drought decision support tool for rangeland managers.