Abstract
Heavy metal pollution of croplands is a major environmental problem worldwide. Methods for accurately and quickly monitoring heavy metal stress have important practical significance. Many studies have explored heavy metal stress in rice in relation to physiological function or physiological factors, but few studies have considered phenology, which can be sensitive to heavy metal stress. In this study, we used an integrated Normalized Difference Vegetation Index (NDVI) time-series image set to extract remote sensing phenology. A phenological indicator relatively sensitive to heavy metal stress was chosen from the obtained phenological periods and phenological parameters. The Dry Weight of Roots (WRT), which directly affected by heavy metal stress, was simulated by the World Food Study (WOFOST) model; then, a feature space based on the phenological indicator and WRT was established for monitoring heavy metal stress. The results indicated that the feature space can distinguish the heavy metal stress levels in rice, with accuracy greater than 95% for distinguishing the severe stress level. This finding provides scientific evidence for combining rice phenology and physiological characteristics in time and space, and the method is useful to monitor heavy metal stress in rice.