The potential of in-situ phenology data to estimate satellite driven gross primary productivity of rice in Arkansas

The Vegetation Photosynthesis Model (VPM) is a simplistic and powerful method to measure gross primary productivity for large regions and sites lacking in-situ instruments. However, the use of biome specific and constant parameters across the whole growing season for an annual crop like rice, which quickly transitions through different maturation phases, can increase the uncertainty in the model at both the site and region scales. These challenges highlight the need for calibration and validation of model parameters at both scales. Previous studies have found that the parameters maximum light use efficiency (LUEmax) and optimum air temperature (Topt) are important for accurate estimation of gross primary productivity (GPP) in the VPM. Thus, an improved version of the model is needed, and several studies have found better results considering site information and customizing the model to phenological VPM (PVPM) by including phenological LUEmax and Topt values into the model structure. However, most of the studies have been applied to the site-scale only. In this study, we leverage the potential of 16 site-years of in-situ data (via the eddy covariance (EC) technique) to estimate LUEmax at 8 day intervals and to build a relationship of the phenological nature of LUEmax and Topt from rice fields based on the number of days after planting in Arkansas. We make use of phenological information to modify the PVPM model and use this relationship across the state's rice production region to have a robust estimate of the GPP associated with Arkansas rice cultivation. Preliminary results validated against 16 site years indicate that at site scale, the modified PVPM performs better than VPM by having higher R2, better slope, and lesser root mean square error and mean absolute error. The mean spatial pattern of GPP at the state scale has shown that rice fields located in the north and mid latitudes are more productive than rice fields located in other regions of Arkansas. GPP estimates based on greater accuracy can help address its underlying meteorological and soil factors, derive a relationship with crop yield, investigate crop response to a changing climate, and render a benchmark for validating simulation output from environmental models.