Forecasting global crop yields based on El Nino Southern Oscillation early signals

Abstract

CONTEXTThe El-Niño Southern Oscillation (ENSO), one of the most well-known climate modes, can lead to large-scale climate variability and subsequent crop loss, posing a severe risk to global food security. OBJECTIVEThe study's main goal was to examine the synchronous impacts of ENSO and the probability of simultaneous ENSO–related crop loss on the global yields of major crops and investigate the predictability of finer-scale variation in crop yields based on ENSO-related large-scale climate precursors. METHODSHere, using updated crop census data for ∼12,000 political units, the study first investigated the synchronous impact of ENSO on yield variability of major crops (i.e., maize, rice, wheat, and soybean) using Synthetic Analysis and bootstrap method, and then estimated the probability of simultaneous crop loss in the top five crop-producing countries by copula approach. Finally, multiple regression was developed to identify the best forecast model, the corresponding ENSO indices, and the lead time for each political unit based on pre-occurred ENSO indices. RESULTS AND CONCLUSIONSThe results show that 12.8% (2.1%), 13.4% (6.4%), 11.8% (10.2%), and 8.4% (18.3%) of wheat, rice, maize, and soybean harvest areas were significantly negatively (positively) associated with El Niño, respectively; and 7% (11.7%), 20.2% (3.4%), 5.8% (5.6%), and 14% (6.4%) with La Niña. El Niño reduced global-mean crop yield by 1.32%, 1.33%, and 0.37% for wheat, rice, and maize, respectively, but increased it for soybean by 1.9%. La Niña reduced the global mean yield for rice (2.1%), maize (1.5%), and soybean (1.3%) but increased it for wheat (1.0%). Rice (6.6%) had the highest probability of simultaneous loss during the El Niño phase, whereas La Niña is soybean (5.9%). Based on the early ENSO signals, crop yield could be reliably forecasted for ∼32.05%, ∼42.2%, ∼21%, and ∼ 26.37% of global harvest areas, with R2 being 0.24, 0.26, 0.24, and 0.23 and a lead time of 1–12 months, for wheat, rice, maize, and soybean, respectively. The results suggest that although the reliable yield prediction based on ENSO indexes alone can be developed in a limited proportion of harvest areas, it is skillful in the ENSO-sensitive regions. SIGNIFICANCEThe findings improved the understanding of ENSO-induced crop yield variability and developed novel approaches to forecast global crop yields based on early ENSO signals.