Remote-sensing disturbance detection index to identify spatio-temporal varying flood impact on crop production

Flooding is the most common type of natural hazards that can interrupt crop growth and reduce production. Current understanding of flood impact on crops is largely obtained from broad-scale studies without considering the influence of localized variations. Due to the highly localized features of flooding, it is essential to develop an effective and systematic approach to investigate and better understand the spatio-temporal varying flood disturbances at fine spatial scales. Based on the pixel-based time series of Enhanced Vegetation Index (EVI) data, two satellite-based flood disturbance detection indices (DIs), i.e. EVI and peak EVI, are developed to recognize the difference between the signals induced by natural variations and instantaneous/non-instantaneous flood impact in crop growth processes. To define flood impact, the actual and predicted normal values of temporal trajectories of EVI and peak EVI during the crop growing seasons are compared to detect and remove the interference from the crop’s intra-annual natural variations. A range of natural variations are considered to discern the signal induced by the crop’s inter-annual natural variations. Furthermore, recovery of crops from flooding is also considered by comparing the peak EVI during crop growing seasons to detect the final flood impact. Using the Northeast China as a case study area, we successfully demonstrate the capacity of these two DIs to identify spatio-temporal varying flood impact on crop production. The DIs also reveal positive response of crops to extreme precipitation under certain conditions. Further analysis demonstrates the non-linear relationships between flood disturbances and terrain slope, distance from rivers, and flow accumulation area, which enable the development of empirical regression models to sufficiently capture the variation of flood damage extent. The research findings confirm that the two DIs proposed in this work are useful in detecting flood disturbances to crops and facilitating informed decision-making in agricultural flood management.