A Robust Protocol for Estimating Crop Yield Potential at Regional to National Scales.

World population will increase 35%, by 2050, which is projected to require a 70-100% rise in food production.  Given limited arable land and freshwater resources and need to conserve natural ecosystems, how and where to concentrate agriculture and where to most effectively conserve nature become critical issues. Estimates of crop yield potential (Yp) help identify where best to intensify agriculture. Previous estimates of Yp are not necessarily transparent, robust or realistic. Some use rudimentary crop models and monthly-average weather data, which is too coarse for accurate estimates of crop Yp. Others rely on historical weather datasets of short duration, or use crop models that require a large number of genotype-specific parameters difficult to estimate or obtain.  Our purpose was to improve estimation of Yp at regional and national levels by developing a protocol using models that are crop-specific, publically available, have been vigorously validated, and, as far as possible, requires few cultivar-specific parameters. Five steps are involved: 1) Selection of reference weather stations (RWS) for adequate geospatial coverage of production area for the crop in question, 2) estimation of missing weather data and quality control, 3) gathering precise information about local crop management practices, 4) simulation of Yp for each RWS 5) estimation of national or regional Yp based on weighting of RWS Yp estimates by proportion of total production they represent. Using this method we estimated Yp of rice in China and U.S. maize, and Yp-rainfed of U.S. maize and wheat in Germany. Coverage of total harvested area by 100 km RWS buffers represented 40, 60, 49, and 74% of total harvested area, respectively. Research in progress will report comparisons of Yp estimates using the above protocol with other methods. Results suggest that current average yields of these crops are 70-85% of estimated national Yp with current management practices.