| Changes in soil moisture(SM) content and soil organic matter(SOM)content are important information that precision agriculture needs to obtain from the soil.Grasping the changes in SM content and SOM content is the basis for observing the development of soil and the law of spatial and temporal changes,and can provide reliable data for soil research and protection.With the ability to obtain abundant spectral information,the rapidly developing hyperspectral remote sensing technology can obtain fine spectral information of ground features.This technology will gradually replace traditional chemical analysis methods and become one of the main methods for monitoring soil component content changes.However,most of the current inversion models for SM content and SOM content are statistical models,lacking a complete physical foundation,and the inversion accuracy and the scope of application of the model are limited.Therefore,through the construction of the semi-empirical model for SM content inversion and the semi-empirical model for SOM content inversion,the scientific problem of establishing the corresponding relationship between soil characteristic parameters and spectral information can be solved,so as to realize the large-scale and high-precision monitoring of SOM content and SM content.In addition,it has been discovered in practical applications that parameters such as spectral range,spectral resolution,and signal-to-noise ratio(SNR)directly affect the inversion accuracy,and the lack of research on the relationship between remote sensing parameters and inversion accuracy restricts the expansion and in-depth data application field.Regarding application requirements as the starting point and end point of the development of hyperspectral remote sensing instruments,this research is based on the constructed SOM and SM inversion model and the inverse form of the model.Through application demand investigation and research,the sensitive wavelengths,the best spectral resolution and the applicable range of SNR for SOM and SM are selected,so that the development of remote sensing instruments is more targeted and effectiveness.This dissertation focuses on the inversion of SOM and SM content and the study of spectral remote sensing parameters.The main research work is as follows:Aiming at the problem that the inversion accuracy of SM content needs to be improved,this study built a SM content inversion model based on Kubelka-Munk(KM)theory,and deeply explored the relationship between SM content and reflectance.By studying the relationship between SM content and the diffuse reflectance R?,which is often regarded as an unknown parameter or constant in the KM model,the SM content information is introduced into KM model,and then a semi-empirical model for SM content inversion is constructed based on KM theory.The validity of the model is verified by the experimental data of four sorts of soils.The results showed that the root mean square error of prediction(RMSEP)is less than 1.7%,the coefficient of determination(R~2)is more than 85%,and the ratios of the performance to deviation(RPD)is more than 2.5 in the wavelength range of 470-2400 nm.Therefore,the SM content inversion model constructed in this paper has high prediction accuracy,and can be well applied to the prediction of water content in different sorts of soils.The construction of the semi-empirical model for SM inversion is of great significance to grasp the temporal and spatial variation of SM in real time,and provides a solid theoretical support and powerful tool for the subsequent research on the selection of sensitive wavelength,the best spectral resolution and the applicable range of SNR for SM.Considering that most of the current inversion models of SOM content are statistical models,lacking a complete physical foundation,and the inversion accuracy and the scope of application of the model are limited,this study constructs a semi-empirical model for inversion of SOM content,which provides a new method for estimating SOM content.Firstly,the relationship between reflectance and transform reflectance is derived based on KM theory,and then the SOM content is introduced into the model using the absorption coefficient and scattering coefficient related to the SOM content,thereby constructing the SOM inversion model.The validity and reliability of the model are verified by the validation set data.The results show that in the wavelength range of 450-2400nm,the model has high estimation accuracy,especially in the wavelength range of 552-950nm,the model has the highest accuracy(RMSEP:0.18%,R~2:89.9%,RPD:3.2).Compared with the current statistical models that are mostly used for the estimation of SOM content,this model has stronger theoretical basis and has stronger applicability.It can overcome the limitations of applying hyperspectral remote sensing data to estimate the SOM content in a large area.The construction of the semi-empirical model for SOM content inversion provides a solid theoretical foundation and powerful tool for the subsequent research on the selection of sensitive wavelength,the best spectral resolution and the applicable range of SNR for SOM.In order to make the instrument development more pertinent and effective,the spectral remote sensing parameters have been studied:(1)Considering that the traditional wavelength selection method requires a large amount of actual measurement data,and its performance is directly related to the quality and representativeness of the"training data",a method based on the sensitivity analysis of the soil radiation transmission model-successive projection algorithm(SA-SPA)is proposed to select the sensitive wavelengths.In order to verify the effectiveness of this method,the experimental data set was used to estimate the soil component content at the selected wavelengths,and the method was compared with the traditional wavelength selection methods.The results show that the sensitive wavelength interval of SOM is 450-1020nm,and the sensitive wavelength points of SOM are 468nm,476nm,496nm,599nm,775nm and 900nm.The retrieval accuracy of SOM content at the selected six wavelengths is high(RMSEP<0.234%).Especially at 599nm,RMSEP is 0.176%,R~2 is 90.4%.The sensitive wavelength intervals of SM content are 853-1125nm and 1900-1988 nm,and the sensitive wavelength points of SM content are1010nm,1068nm,1915nm,1946nm and 1988nm,and the retrieval accuracy of SM content is high at the selected five wavelengths(RMSEP<1.21%).Especially at1915nm,the accuracy of soil moisture estimation is the highest,RMSEP is 1.04%.(2)This study resampled the spectral data in the selected sensitive wavelength range to obtain data with different spectral resolutions,and explored the influence of spectral resolution on the inversion accuracy of the model constructed in this research,so as to determine the best spectral resolution of soil nutrients in the corresponding sensitive wavelength range.The research results show that in the sensitive wavelength interval of SOM,the inversion accuracy is higher when the spectral resolution is less than 17nm;at the sensitive wavelength points of SOM,the inversion accuracy is highest when the spectral resolution is 13nm;in the sensitive wavelength interval of SM,the inversion accuracy is higher when the spectral resolution is less than 29nm;at the sensitive wavelength points of SM,the inversion accuracy is the highest when the spectral resolution is 34nm.(3)By adding Gaussian distributed random noise to the noise-free spectral data simulated by the soil radiation model,the spectral data collected under different SNRs can be simulated.The influence of different SNRs on the inversion accuracy of the constructed model is studied,and the range of SNRs that can be used for soil physical and chemical composition measurement is obtained.The research results show that:in the sensitive wavelength interval of SOM,when the spectral resolution is 17nm,the applicable SNR range is 140-340;at the sensitive wavelength points of SOM,when the spectral resolution is 13nm,the applicable SNR range is 150-360;in the sensitive wavelength interval of SM,when the spectral resolution is 29nm,the applicable SNR range is 150-340;at the sensitive wavelength points of SM,when the spectral resolution is 34nm,the applicable SNR range is 150-350.The research results will provide solid theoretical basis and technical support for the quantitative inversion of SOM and SM content based on spectral remote sensing,and the development of special rapid measurement spectrometers. |
PDF Full Text Request