Deriving The Characteristic Scale For Heavy Metal Stress Monitoring In Rice Based On The Remote Sensing Data

The objective of this paper was to identify the optimal characteristic scale for monitoring heavy metal stress in rice by remote sensing based on target characteristics and observation approach.In terms of target characteristics,the farmland with different heavy metal stress level in Zhuzhou City of Hunan Province was selected,we explored the response characteristics of rice biochemical parameters under heavy metal stress and then selected the sensitive index.Considering the observation approach,the spatial and spectral response characteristics of rice heavy metal stress sensitive features were studied based on the multi-spectral remote sensing data and the ground observation hyperspectral data,meanwhile,the corresponding feature scale recognition method was established from the perspective of spatial dimension and spectral dimension.On one hand,this study are expected to reduce the blindness of remote sensing data selection and provide an new ideas with the efficient monitor of heavy metal stress,on the other hand,the study in this paper are helpful for the future design of low and practical multi-spectral and hyperspectral sensors,which provide indicators and theoretical reference.The main contents and conclusions are as follows:(1)For the analysis of spatial scale selection,firstly,we selected WRT(Weight of Root)as the diagnostic parameter of heavy metal stress based on the previous study and experimental analysis.The assimilation parameters LAI(Leaf Area Index)was scaled up by the pixel aggregation method,meanwhile,the improved WOFOST(World Food Studies)model was introduced to simulate WRT in different scale through the coupling of remote sensing data.Additionally,the assessment efficiency of different spatial scales was addressed by using statistical analyses.Finally,a qualitative ratio analysis was conducted to identify the optimal characteristic scale for the purpose of realizing better observations.Results indicated that the critical threshold for investigating the rice WRT in monitoring studies of heavy metal stress was larger than 64 m but smaller than 256 m.(2)For the analysis of spectral scale selection,firstly,450-900 nm of ASD spectral band was selected as the sensitive band of heavy metal stress.The wavelet coefficients were obtained using the Daubechies 5 wavelet function,which decompose the original rice spectral curves in the study area into 8-layer.And the one-dimension spectral signal was rebuilt using the wavelet approximation coefficients of each layer,which could provide fractal calculation for multi-scale spectral signals with different resolution.The wavelet fractal dimension was calculated by using the box dimension method.Furthermore,the wavelet fractal dimension and wavelet detail coefficient entropy was determined to identify the turning point of rice spectral characteristic scale.Result indicated that the wavelet approximation coefficients can reflect the whole trend of spectral curve.The variance of wavelet detail coefficient entropy and wavelet fractal value with different stress levels is larger at 0-4 scale.With the increase of decomposition scale,the difference between them is small,but the trend of different stress level is consistent.That is,the rice spectral resolution of less than 32 nm can better distinguish the different rice heavy metal pollution levels.This conclusion was further proved by the rice hyperspectral index corresponding to each scale.