Kevin Tu Jesslyn F. Brown Michael Marshall 2018 <p><span>Earth observation data are increasingly used to provide consistent eco-physiological information over large areas through time. Production efficiency models (PEMs) estimate Gross&nbsp;Primary Production&nbsp;(GPP) as a function of the fraction of photosynthetically active radiation absorbed by the canopy, which is derived from Earth observation. GPP can be summed over the&nbsp;growing season&nbsp;and adjusted by a crop-specific harvest index to estimate yield. Although PEMs have many advantages over other&nbsp;crop yield&nbsp;models, they are not widely used, because performance is relatively poor. Here, a new PEM is presented that addresses deficiencies for macro-scale application: Production Efficiency Model Optimized for Crops (PEMOC). It was developed by optimizing functions from the literature with GPP estimated by&nbsp;eddy covariance&nbsp;flux towers in the United States. The model was evaluated using newly developed Earth observation products and county-level yield statistics for major crops. PEMOC generally performed better at the field and county level than another commonly used PEM, the&nbsp;Moderate Resolution Imaging Spectroradiometer&nbsp;GPP (MOD17). PEMOC and MOD17 estimates of GPP had an R</span>²<span>&nbsp;and root mean squared error (RMSE) over the growing season of 0.71–0.89 (9.87–17.47 g CO</span>₂<span> d</span>⁻¹<span>) and 0.59–0.83 (6.86–22.20 g CO</span>₂<span> d</span>⁻¹<span>) with flux tower GPP. PEMOC produced R</span>²<span>s and RMSE of 0.70 (0.52), 0.60 (0.61), and 0.62 (0.59), while MOD17 produced R</span>²<span>s and RMSE of 0.65 (0.57), 0.53 (0.66), and 0.65 (0.57) with corn,&nbsp;soybean, and winter wheat crop yield anomalies. The sample size of rice was small, so yields were compared directly. PEMOC and MOD17 produced R</span>²<span>s and RMSE of 0.53 (3.42 t ha</span>⁻¹<span>) and 0.40 (4.89 t ha</span>⁻¹<span>). The most sizeable model improvements were seen for C</span>₃<span>&nbsp;and C</span>₄<span>&nbsp;crops during emergence/senescence and peak season, respectively. These improvements were attributed to C</span>₃<span>&nbsp;and C</span>₄<span>&nbsp;partitioning, optimized temperature and moisture constraints, and an evapotranspiration-based soil moisture index.</span></p> application/pdf 10.1016/j.rse.2018.08.001 en Elsevier Optimizing a remote sensing production efficiency model for macro-scale GPP and yield estimation in agroecosystems article