The researchers used MODIS data to estimate plant productivity at different resolutions. They compared leaf area index with vegetation indices like NDVI and EVI to predict plant growth. GPP estimates based on NDVI at 250m resolution had around 9.12% error, while GPP based on EVI at 1000m resolution had around 5.76% error. GPP estimates at 5600m resolution had the highest errors, around 18.47% for NDVI and 13.02% for EVI. Despite some errors, the global resolution can still be used for modeling with an error of around 20%.