Assessing precipitation, evapotranspiration, and NDVI as controls of U.S. Great Plains plant production

Authors: CHEN, MAOSI - Colorado State University; PARTON, WILLIAM - Colorado State University; HARTMAN, MELANNIE - Colorado State University; Del Grosso, Stephen - Steve; SMITH, WILLIAM - University Of Minnesota; KNAPP, ALAN - Colorado State University; LUTZ, SUZY - Colorado State University; Derner, Justin; TUCKER, COMPTON - National Aeronautics And Space Administration (NASA); OJIMA, DENNIS - Colorado State University; VOLESKY, JERRY - University Of Nebraska; STEPHENSON, MITCHELL - University Of Nebraska; SCHACHT, WALTER - University Of Nebraska; GAO, WEI - Colorado State University

Submitted to: Ecosphere
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/7/2019
Publication Date: 10/8/2019
Citation: Chen, M., Parton, W.J., Hartman, M.D., Del Grosso, S.J., Smith, W.K., Knapp, A.K., Lutz, S., Derner, J.D., Tucker, C.J., Ojima, D.S., Volesky, J., Stephenson, M., Schacht, W., Gao, W. 2019. Assessing precipitation, evapotranspiration, and NDVI as controls of U.S. Great Plains plant production. Ecological Applications. 10(10). Article e02889. https://doi.org/10.1002/ecs2.2889.
DOI: https://doi.org/10.1002/ecs2.2889

Interpretive Summary: Plant growth throughout the North American Great Plains grasslands is generally considered to be water limited, with the strength of this limitation increasing as precipitation decreases. We hypothesize that plant and soil water loss (AET) from April to July is the precipitation related variable most correlated to aboveground plant growth in the Great Plains. We tested this by evaluating the relationship of plant growth to AET and precipitation. We used multi-year plant growth data from five sites ranging from semi-arid grasslands in Colorado and Wyoming to wetter grasslands in Nebraska and Kansas, and satellite-derived vegetation greenness (NDVI) data. Results from the five sites showed that cumulative April to July AET and precipitation were well correlated to annual changes in growth for all but the wettest site and that precipitation in August and September had little impact on plant growth in the drier sites. In contrast, April to July precipitation was best correlated with interannual variability in plant growth at the wettest site, with AET poorly correlated with growth at this grassland. Cumulative growing season (May to September) NDVI (iNDVI) was strongly correlated with annual plant growth at the five sites. Using iNDVI as a surrogate for plant growth, we found that county-level cumulative April–July AET was more strongly correlated to ANPP than precipitation for more than 80% of Great Plains counties, with precipitation tending to perform better in the eastern wetter portion of the Great Plains. Accounting for how different water-related variables control plant growth (AET in drier portion, precipitation in wetter portion) provides opportunity to develop spatially-explicit forecasting of forage production across the Great Plains for enhancing decision-making by land managers.

Technical Abstract: Productivity throughout the North American Great Plains grasslands is generally considered to be water limited, with the strength of this limitation increasing as precipitation decreases. We hypothesize that cumulative actual evapotranspiration water loss (AET) from April to July is the precipitation related variable most correlated to aboveground net primary production (ANPP) in the Great Plains. We tested this by evaluating the relationship of ANPP to AET, precipitation, and plant transpiration (Tr). We used multi-year ANPP data from five sites ranging from semi-arid grasslands in Colorado and Wyoming to mesic grasslands in Nebraska and Kansas, and satellite-derived normalized difference vegetation index (NDVI) data. Results from the five sites showed that cumulative April to July AET, precipitation, and Tr were well correlated (R2: 0.44–0.65) to annual changes in ANPP for all but the wettest site and that precipitation in August and September had little impact on productivity in these drier sites. In contrast, April to July precipitation was best correlated (R2=0.60) with interannual variability in ANPP in the most mesic site, with AET and Tr poorly correlated (R2<0.06) with ANPP in this grassland. Cumulative growing season (May to September) NDVI (iNDVI) was strongly correlated with annual ANPP at the five sites (R2=0.86). Using iNDVI as a surrogate for ANPP, we found that county-level cumulative April–July AET was more strongly correlated to ANPP than precipitation for more than 80% of Great Plains counties, with precipitation tending to perform better in the eastern more mesic portion of the Great Plains. Accounting for how different water-related variables control ANPP (AET in drier portion, precipitation in wetter portion) provides opportunity to develop spatially-explicit forecasting of ANPP across the Great Plains for enhancing decision-making by land managers.