Scaling Near-Surface Remote Sensing To Calibrate And Validate Satellite Monitoring Of Grassland Phenology
Date of Award
Doctor of Philosophy (PhD)
Earth System Science & Policy
Phenology across the U.S. Great Plains has been modeled at a variety of field sites and spatial scales. However, combining these spatial scales has never been accomplished before, and has never been done across multiple field locations. We modeled phenocam Vegetation Indices (VIs) across the Great Plains Region. We used coupled satellite imagery that has been aligned spectrally, for each imagery band to align with one another across the phenocam locations. With this we predicted the phenocam VIs for each year over the six locations.Using our method of coupling the phenocam VIs and the meteorological data we predicted 38 years of phenocam VIs. This resulted in a coupled dataset for each phenocam site across the four VIs. Using the coupled datasets, we were able to predict the phenocam VIs, and examine how they would change over the 38 years of data. While imagery was not available for modeling the 38 years of weather data, we found weather data could act as an acceptable proxy. This means we were able to predict 38 years of VIs using weather data. A main assumption with this method, it that no major changes in the vegetation community took place in the 33 years before the imagery. If a large change did take place, it would be missed because of the data lacking to represent it. Using the phenocam and satellite imagery we were able to predict phenocam GCC, VCI, NDVI, and EVI2 and model them over a five-year period. This modeled six years of phenocam imagery across the Great Plains region and attempted to predict the phenocam VIs for each pixel of the satellite imagery. The primary challenge of this method is aggregating grassland predicted VIs with cropland. This region is dominated by cropland and managed grasslands. In many cases the phenology signal is likely driven by land management decisions, and not purely by vegetation growth characteristics. Future models that take this into account may provide a more accurate model for the region.
Burke, Morgen W.v., "Scaling Near-Surface Remote Sensing To Calibrate And Validate Satellite Monitoring Of Grassland Phenology" (2021). Theses and Dissertations. 4159.