Spatial and Interannual Variability of Drought Stress in U.S. Grasslands

Abstract:

Severe, prolonged droughts are predicted to occur more frequently due to global climate change. We used a combination of satellite imagery and eddy covariance data to evaluate a numerical model of drought impacts on the productivity (plant growth) of grasslands in the U.S. Great Plains. We compared simulations using the Simple Biosphere Model (SiB4) to observations during a period that included extreme drought in various temperate grassland sites in the U.S. Great Plains from 2010-2012. Changes in phenology (seasonal greening and browning) were evaluated over this three-year time period at various grassland sites by comparing simulated leaf area index (LAI) from SiB4 to Moderate Resolution Imaging Spectroradiometer (MODIS) data from the Terra and Aqua satellites. The net ecosystem exchange (NEE) of carbon, gross primary production (GPP), latent heat flux (LH) and sensible heat flux (H) values from SiB4 and flux tower sites were then analyzed over the same three-year time period. These selected variables were chosen to aid in the evaluation of the onset and duration of drought in the model compared to observations. Observed west-to-east gradients in grassland productivity were generally well-captured by the model. Growing season onset tends to be too late in the model relative to observations and peak LAI tends to last too long in the autumn. Changes in seasonal plant allocation mitigated some of these problems. Response to drought, indicated in the observations by reduced LAI and GPP and elevated Bowen ratio, was also simulated by the model. In several cases, simulated drought responses were weaker than observed. We explore reasons for this discrepancy.