Extracting heavy metal stress indicators from remote sensing imagery using WOFOST model and wavelet packet decomposition algorithm

Abstract

Heavy metal pollution of crops seriously endangers food security and indirectly threatens human health. Direct measures in the fields and laboratories through on-site sample collection, testing, and analysis are time-consuming and labor intensive, thereby prohibiting their applications in large-scale monitoring. Remote sensing techniques provide an alternative means through examining above-ground vegetation status, e.g. leaf area index (LAI). Heavy metals, however, are typically accumulated in the root of crops, which may also be affected by a large number of external environmental factors besides heavy metals. The objective of this paper, therefore, is to identify heavy metal stress indicators of crops through integrating LAI extraction from remote sensing imagery, weight of rice roots (WRT) estimation by the World Food Study (WOFOST) model, and heavy metal stress indicator identification with the wavelet packet decomposition (WPD) method. First, LAI was retrieved from the HJ CCD data over three continuous years through constructing a relationship between LAI and normalized difference vegetation index (NDVI). Next, dry weight of rice roots (WRT) curves over these three continuous years were estimated using the WOFOST model with multi-temporal LAIs are inputs. Finally, a component (e.g. cfs 14) was identified to represent the heavy metal pollution status with the Wavelet Packet Decomposition (WPD) of the WRT curves of these three years. Validation results suggest that the identified component can successfully represent different levels of heavy metal stress.