2020 ESA Annual Meeting (August 3 - 6)

OOS 30 Abstract - Grass-Cast: Using grassland productivity forecast to link ecology and ecosystem management

William Parton1, Melannie Hartman1, Melannie Hartman1, Dannele Peck2, Justin D. Derner3, William Smith4, Stephen J. Del Grosso5 and Brian Fuchs6, (1)Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, CO, (2)Northern Plains Climate Hub, USDA Agricultural Research Service, Fort Collins, CO, (3)USDA-ARS, Rangeland Resources and Systems Research Unit, Cheyenne, WY, (4)School of Natural Resources and the Environment, University of Arizona, Tucson, AZ, (5)USDA-ARS, Soil Management and Sugar Beet Research Unit, Fort Collins, CO, (6)National Drought Mitigation Center, University of Nebraska-Lincoln, Lincoln, NE
Background/Question/Methods

We have developed a grassland productivity forecast tool known as “Grass-Cast” using over 30 years of historical data including weather and the satellite-derived normalized vegetation difference index (NDVI) combined with ecosystem modeling and seasonal precipitation forecasts. The Grass-Cast model has been used to predict changes in growing season grassland plant production for the U.S. Great Plains from 2017 to 2019; in 2020 we added predictions for Arizona and western New Mexico. Grass-Cast produces its forecasts on a biweekly basis from early April until the end of August with results shown on the Grass-Cast web site: http://grasscast.unl.edu/. Grass-Cast uses observed daily weather data (daily maximum and minimum temperature and precipitation) as input to the DayCent ecosystem model which predicts cumulative April to July/August actual evapotranspiration (the precipitation-related variable most correlated to plant production in this region) using the observed weather data up to the time of the forecast and projected precipitation until the end of the growing season (normal, above normal or below normal precipitation). A biweekly 3-map Grass-Cast based on these three precipitation scenarios is produced for a 10x10 km2 grid for eleven states in the Great Plains and Southwestern U.S.

Results/Conclusions

A comparison of Grass-Cast predicted plant production with MODIS NDVI estimated plant production (validation data) shows that the model accuracy increases rapidly in from May to mid-June; the correlation between early growing season forecasts and the end-of-growing season ANPP estimate is > 50% by this time. Furthermore, the end-of-season Grass-Cast and MODIS NDVI estimates of growing season plant production from 2000 to 2019 agreed for 60% to 90% of counties each year. Results from the expansion of Grass-Cast to the Southwestern U.S. show two peaks in plant production during the growing season with the May peak in plant production correlated to cumulative January to May actual evapotranspiration, and the late summer monsoon peak in plant production correlated to cumulative rainfall from June to August. Future additions to the Grass-Cast model include using MODIS NDVI observations in the spring to predict growing season cumulative NDVI and weekly changes in live grassland biomass during the growing season.