Assessing large-scale patterns of gross primary production (GPP) in arid and semi-arid (ASA) areas is important for both scientific and practical purposes. Remote sensing-based models, which integrate satellite data with input from ground-based meteorological measurements and vegetation characteristics, improve spatially extended estimates of vegetation productivity with high accuracy. In this study, the authors simulated GPP in ASA areas by integrating moderate resolution imaging spectral radiometer (MODIS) data with eddy covariance and meteorological measurements at the flux tower sites using the Vegetation Photosynthesis Model (VPM), which is a remote sensing-based model for analyzing the spatial pattern of GPP in different land cover types. The field data were collected by coordinating observations at nine stations in 2008. The results indicate that in the region during the growing season GPP was highest in cropland sites, second highest in woodland sites, and lowest in grassland sites. VPM captured the temporal and spatial characteristics of GPP for different land covers in ASA areas. Further, Enhanced Vegetation Index (EVI) had a strong liner relationship with GPP in densely vegetated areas, while the Normalized Difference Vegetation Index (NDVI) had a strong liner relationship with GPP over less dense vegetation. This study demonstrates the potential of satellite-driven models for scaling-up GPP, which is a key component for studying the carbon cycle at regional and global scales.