Detecting and modeling topography‐induced differences in plant productivity across a grassland ecosystem gradient

Understanding spatial variation in grassland productivity is critical for ecological modeling and management. The Great Plains of North America, a vast grassland region east of the Rocky Mountains, exhibits strong gradients in aboveground net primary production (ANPP) linked to precipitation and finer scale topoedaphic variation. Swales typically have 40%–80% greater ANPP than adjacent ridges due to deeper, nutrient‐rich soils with higher water‐holding capacity. We used the DayCent‐UV model to simulate interannual ANPP differences between ridges and swales at three grassland sites across a west‐to‐east precipitation gradient. This DayCent‐UV version includes higher N availability to plants in the swales. We also evaluated estimates of ANPP from the Rangeland Analysis Platform and the Rangeland Carbon Tracking and Management framework, two satellite‐driven process‐based models that use normalized difference vegetation index (NDVI) to estimate productivity. These models better captured swale dynamics than DayCent did but overestimated ridge ANPP. Finally, we evaluated the performance of remote sensing‐derived vegetation indices (NDVI, enhanced vegetation index [EVI], near‐infrared reflectance of vegetation [NIRv]) in capturing interannual biomass variability across topoedaphic positions. To overcome saturation and cloud limitations in the most humid sites, we incorporated Sentinel‐1 radar backscatter as an alternative proxy for vegetation biomass, which successfully detected both interannual and spatial differences and showed potential for identifying woody encroachment. Our results demonstrate a methodological framework that integrates process‐based modeling with multisource remote sensing to improve detection of fine‐scale productivity variation in grassland ecosystems.