Collaborative inversion heavy metal stress in rice by using two-dimensional spectral feature space based on HJ-1 A HSI and radarsat-2 SAR remote sensing data

Abstract

Accurate and extensive monitoring of heavy metal pollution levels in rice fields is crucial to agricultural production and food safety.Most previous studies used optical remote sensing data to monitor heavy metal stress in rice in which way the remote sensing features are not diversified and the monitoring accuracy is relatively low. Compared with that optical remote sensing can monitor features like color and cell internal structure, microwave remote sensing can monitor the morphology and geometrical features. The complementary characteristics of multi-source remote sensing data are reflected in two aspects. On the one hand, the high spectral resolution characteristics of hj-1a HIS and the high spatial resolution characteristics of radarsat-2 SAR satellite are used for feature fusion to obtain the image data of high altitude spectral resolution; on the other hand, the two remote sensing data can depict the vegetation characteristics from different angles. In order to fully extract the characteristics of crop stress, the model was constructed by utilizing the complementary characteristics of multi-source remote sensing data. This paper synergized optical and microwave remote sensing to construct the crop heavy metal stress monitoring model. To this end, certain rice-growing areas polluted by heavy metal in Suzhou city were selected as the experimental areas, where ASD spectrum data, biochemical parameters and heavy metal content data of rice were collected during critical growth periods on one hand, HJ-1 A HIS and Radarsat-2 SAR satellite data were obtained almost simultaneously on the other hand. Based on heavy metal stress-responsive mechanism of rice, NVI (R598, R508) and SVI(HV,VH,HH,VV), spectral indexes sensitive to heavy metal stress, were extracted from the optical and radar data respectively to construct a two-dimensional feature space, based on which an optical-and-radar-remote-sensing-combined model for monitoring heavy metal stress in rice was constructed. In this paper, the study and main conclusions are as follows:(1) It used spectral characteristics analysis combined with statistic methods or random forest algorithm to build the canopy chlorophyll index NVI, which is sensitive to chlorophyll content changes of rice under heavy metal stress, and a heavy metal stress level inversion model was built based on hyperspectral HSI data accordingly. Also, it used statistic method to build the microwave index SVI, which is sensitive to biomass changes of rice under heavy metal stress, and a heavy metal stress level inversion model was built based on microwave SAR images as a result. (2) A combination of biochemical and morphological parameters, both responding to heavy metal stress, was synergized with NVI and SVI, sensitive to chlorophyll and biomass changes respectively, to build a two-dimensional feature space. Heavy metal stress levels were classified in this space. Thus, a synergetic model to retrieve heavy metal stress based on multi-source remote sensing data was developed. In this paper, the innovation lies in a model constructed by synergizing optical and radar remote sensing data to monitor heavy metal stress in rice based on a multi-dimensional spectral feature space and this model can be applied to monitoring multiple environmental stresses in crops.