Characterizing the reliability of global crop prediction based on seasonal climate forecasts

Reliable crop prediction based on seasonal climate forecasts can be achieved when a strong climate- crop relationship exists and there are reliable forecasts of the climatic constraints on crops. Here, we present global assessments of the climatic constraints on crops (maize, soybeans, rice, and wheat), the degree of the climate-crop relationship, and the reliability of seasonal forecasts of dominant climatic constraints based on statistical crop models and ensemble seasonal climate forecasts. We then classify the reliability of within-season crop prediction into four categories based on the degree of the climate-crop relationship and the reliability of the climate forecasting: (I) reliable; (II) less reliable due to the low reliability of climate forecasting; (III) not reliable due to the low reliability of climate forecasting and a weak climate-crop relationship; and (IV) less reliable due to a weak climate-crop relationship. The results showed that a strong climate-crop relationship exists in the area that produces 24-38% of the global crop production. On a global scale, 51-59% of the maize and soybean production is sensitive to soil moisture level during the reproductive growth period, whereas 47-53% of the rice and wheat production is sensitive to temperature. Due to the greater reliability of temperature forecasts, crop prediction is reliable in those areas in which the crop yield is temperature-sensitive and temperature forecasts are reliable. The categorized reliability of crop prediction indicated that improvements of soil moisture forecasts in 30-50°N during July- October and in 30-40 ? S during February-April are needed for better maize and soybean prediction, whereas improved temperature forecasts in 20-60 ° N during March-August are keys to rice and wheat prediction. This study established a novel way of assessing the reliability of crop prediction, which will enable decision-making and allow researchers to prioritize the direction of new research to improve crop prediction in a given area for global food prediction.