Validation Of MODIS GPP And ET Products Under Different Land Covers Types Across North America

Satellite remote sensing products have played a significant role in studying large scale carbon and hydrological cycles. MOD 17 and MOD 16 provide 1-km resolution of continuous global estimates of terrestrial gross primary production (GPP) and evapotranspiration (ET) respectively. There has been much effort to validate the two products over the globe, most of which was focusing on one product. However, the carbon and water cycles are deeply interconnected; a combined validation is in need to better understand their processes and relationship. We chose six flux sites in North America, each of which represents a typical land cover type. The flux towers' measurements of GPP and ET are used as ground truth to validate the remote sensing products, with an aim to find and explain the uncertainties within the products for typical land covers across the continent. This could be a useful reference when using these products in models or for refinement of MODIS products algorithm. Specifically, the bias is calculated along with the root mean square error and correlation coefficient for five years. We also utilized Bayesian methodology for uncertainty analysis, and quantified the bias as a function of soil moisture content and temperature. Overall speaking, the MODIS products reflect the seasonal variation. Results also indicate that the MOD17 underestimates GPP and MOD16 overestimates ET in growing season to some extent, and the performance of the products varies with different vegetation. Studying the ability of satellites to monitor carbon and water cycling is important for our understanding and addressing global and regional climate change issues.