National Scale Maize Yield Estimation by Integrating Multiple Spectral Indexes and Temporal Aggregation

Maize yield in China accounts for more than one-fourth of the global maize yield, but it is challenged by frequent extreme weather and increasing food demand. Accurate timely estimation great significance to crop management security. Commonly applied vegetation indexes (VIs) are mainly used as they can reflect greenness vegetation. However, environmental pressures growth development difficult monitor evaluate. Indexes water content, pigment nutrient elements biomass have been developed indirectly explain influencing factors with extant studies assessing VIs, climate content factors. Only a few attempted systematically evaluate sensitivity these indexes. The spectral indexes, combined effect temporal aggregation data need be evaluated. Thus, this study proposes novel evaluation method integrating multiple data. In particular, were calculated publicly available (remote sensing images data) from Google Earth Engine platform, county-level yields 2015 2019 estimated using random forest model. Results showed that normalized moisture difference index (NMDI) most sensitive estimation. Furthermore, potential adopting especially NMDI_NDNI, was verified. Compared whole-growth period eight-day time series, vegetative reproductive into periods. obtained R2 reached 0.8. This provide feature knowledge references assessments research.