Agrometeorological crop yield forecasting methods

This paper presents an introduction to agrometeorological crop yield forecasting (ACYF) methods. Such methods attempt to quantitatively assess the effect of weather vagaries on regional crop yields before harvest. The paper insists on the fact that all ACYF approaches are eventually calibrated against traditional yield statistics obtained through sampling procedures. Three methods are covered: descriptive methods, regression methods and crop simulation methods. Descriptive methods simply classify weather conditions according to one or several variables to identify conventional thresholds separating groups of significantly different yields. Regression methods derive equations relating crop yield with weather variables. Finally, simulation methods, the most complex ones currently in use, attempt to analytically describe the physical and physiological impact of environment and management conditions on crop development, growth and yield. A large section of the paper deals with methodological options and questions. Most of them are linked with the differences in temporal and spatial scales of the model inputs and outputs, and apply equally to the three methods. Inputs cover a large spectrum of data sources and types, from weather stations (points) and satellites (pixels) to soils (polygons) and direct crop observations at the field level. The integration of the data requires techniques of aggregation and dis-aggregation which are now available, although their implementation is fraught with methodological problems. In practice, methods will be selected based on their performance under a range of different conditions and their cost of implementation.