Yang Chen1, Randall J. Donohue2, Tim R. McVicar2, François Waldner3, Gonzalo Mata1, Noboru Ota1, Alireza Houshmandfar1, Kavina Dayal4, Roger A. Lawes1
1 CSIRO Agriculture and Food, Underwood Ave, Floreat WA 6014, Australia; firstname.lastname@example.org, 2 CSIRO Land and Water, GPO Box 1700, Canberra, ACT 2061, Australia; 3 CSIRO Agriculture and Food, 306 Carmody Rd, St Lucia QLD 4067, Australia; 4 CSIRO Agriculture and Food, College Rd, Sandy Bay TAS 7005, Australia.
Inter-annual national crop yields fluctuate due to area planted, management strategies, prevailing weather conditions, weeds, diseases, and pests. High temporal and spatial variability in grain production makes nationwide crop yield prediction challenging. The present study developed a model combining semi-physical and empirical approaches to estimate yield of major crops (i.e., canola, wheat and barley) across the Australian dryland wheatbelt using a remote sensing (RS) based radiation use efficiency approach and meteorology-driven Stress Indices (SI). Crop specific stress indices (e.g., drought, heat and cold stress) in critical months (e.g., anthesis and grain-filling) were used to explain the impact of highly variable (in both space and time) actual grain yield across the wheatbelt. The present model, Crop-SI, reduces the field-scale prediction error rate by ~20% when compared with existing models for nationwide crop yield simulations. Our finding have improved the predictive capability of RS-based models for crop yield for a wide range of variability in meteorological conditions by incorporating rainfall and temperature into the simulation and provides new insights for the next-generation of nationwide agricultural yield models.