Developing a multi-sensor remote sensing based 30-m grassland biomass productivity map for central Nebraska
Satellite-derived growing season time-integrated Normalized Difference Vegetation Index (GSN) has been successfully used as a proxy for vegetation biomass productivity. A 250-m grassland biomass productivity map for the Greater Platte River Basin has been developed based on the relationship between Moderate Resolution Imaging Spectroradiometer (MODIS) GSN and Soil Survey Geographic (SSURGO) annual grassland productivity. The high temporal resolution and a wide range of wavelengths make the MODIS land surface products robust and reliable. However, the 250-m MODIS grassland biomass productivity map lacked local detail on ecological features (or patterns) and may result in only generalized estimation of the regional total productivity. Developing a high or moderate spatial resolution (e.g., 30-m) productivity map to better understand the regional detailed vegetation condition and ecosystem services is preferred. The long-term 30-m Landsat data provide spatial detail for characterizing human-scale processes and have been successfully used for land cover and land change studies. The main goal of this study is to develop a 30-m grassland biomass productivity map for central Nebraska leveraging MODIS and Landsat observations. A data mining approach was used to downscale the 250-m MODIS GSN to 30-m. The GSN-grassland productivity equation was applied to generate a 30-m grassland biomass productivity map for central Nebraska.
A data-driven rule-based piecewise regression MODIS-Landsat GSN mapping model was developed at a 250-m spatial resolution. The correlation coefficient (r) between the predicted GSN and the actual GSN at 250-m resolution is 0.94 with a mean absolute error of 0.023, indicating that the derived 250-m resolution GSN mapping model can successfully predict GSN across the study area. The derived 30-m MODIS-Landsat predicted GSN map provides spatially detailed biophysical, ecological, environmental, and vegetation dynamic information of the study area (e.g., center pivot systems, abandoned land, and gravel pit). The final estimated 30-m grassland biomass productivity map is seamless (with no state or county lines as with SSURGO data) and captured the detailed spatial variations of productivity in the study area (e.g., sand blow out areas and fence lines). Results from this study will be useful for regional ecosystem study and local land management practices in central Nebraska.