Agroecosystem, or agricultural ecosystems, is the important anthropogenic ecosystem to meet the human demand for food, fiber, and feed, and it covers approximately 40-50% of the earth’s land surface. Accurate estimates of agricultural land use and land cover and Gross Primary Production (GPP) are indispensable for global food security and understanding variations in the terrestrial carbon budgets. This dissertation aimed to strengthen the capacities of remote sensing to produce the high-quality products of crop type maps and primary productivity on large regional scales. In chapter 2, we designed simple algorithms to identify paddy rice of two different phenological phases (flooding/transplanting and ripening) at regional scales using 30-m multi-temporal Landsat images. Sixteen Landsat images from 2010 - 2012 were used to generate the paddy rice map in the Sanjiang Plain, northeast China - one of the intensive paddy rice cultivation regions in Northern Asia. The user and producer accuracies of paddy rice on the resultant Landsat-based paddy rice map were 90% and 94%, respectively, and was an improvement over the paddy rice dataset derived through visual interpretation and digitalization on the fine-resolution satellite images and traditional agricultural census data. Chapter 3 evaluated the capacities of the temporal MODIS vegetation indices and the satellite-based Vegetation Photosynthesis Model (VPM) to describe phenology and model the seasonal dynamics of GPP for savanna woodlands in Southern Africa on the site level. The results showed that the VPM-based GPP estimates tracked the seasonal dynamics and interannual variation of GPP estimated from eddy covariance measurements at flux tower sites. This study suggests that the VPM is a valuable tool for estimating GPP of semi-arid and semi-humid savanna woodland ecosystems in Southern Africa. Chapter 4 assessed the accuracies of air temperature and downward shortwave radiation of the North America Regional Reanalysis (NARR) by the National Centers for Environmental Prediction (NCEP), and evaluated impacts of the accuracies of regional climate inputs on the VPM-based GPP estimates for U.S. Midwest cropland. The results implied that the bias of NARR downward shortwave radiation introduced significant uncertainties into the VPM-based GPP estimates, suggesting that more accurate surface radiation datasets are needed to estimate primary production of terrestrial ecosystems at regional and global scales. Chapter 5 presented independent and complementary analyses of the impact of 2012 flash drought on productivity in the U.S. Midwest using multiple sources of evidences, i.e., in-situ AmeriFlux CO2 data, satellite observations of vegetation indices and solar-induced chlorophyll fluorescence (SIF), and scaled ecosystem modeling. The results showed that phenological activities of all biomes advanced 1-2 weeks earlier in 2012 compared to other years of 2010-2014, and the drought threatened the U.S. Midwest agroecosystems. The growth of grassland/prairie and cropland was suppressed from June and it didn’t recover until the end of the growing season. As the frequency and severity of droughts have been predicted to increase in future, this study provides better insights into the impacts of flash droughts on vegetation productivity and carbon cycling of major biomes in the U.S. Midwest.