Soil Moisture Estimation Algorithms Using NIR-Red-LST Feature Space

Soil moisture is one of the important variables in environmental research.It affects the exchange of water and energy fluxes between the surface and atmosphere,plays an important role in water resources management,environmental monitoring and global climate change research.Therefore,it is very important to accurately estimate the temporal and spatial variations of soil moisture for many environmental studies.With the development of remote sensing technology,it is possible to monitor soil water content rapidly,in large area and periodically,and soil moisture monitoring by remote sensing has gradually become one of the most commonly used methods for drought monitoring in large areas.In the acquisition method of soil moisture by remote sensing,the red and near infrared bands are sensitive to soil moisture and vegetation information,and most satellites can acquire these two bands of data.Therefore,NIR-Red spectral feature space method is often used to estimate soil water content.The lower edge of NIR-Red triangle feature space is considered as the soil line composed of bare soil pixels.However,due to the limited spatial resolution of remote sensing images,cloud cover and water,the lower edge of triangular space is not entirely composed of bare soil,which leads to inaccurate estimation of soil line and reduces the accuracy of soil water estimation.In addition,the estimation of soil water content based on NIR-Red feature space is only suitable for bare soil and low vegetation areas,In the high vegetation area where the soil is completely covered by vegetation,the reflectance received by satellite sensor is actually the reflectance of vegetation surface,which can no longer express the soil moisture information under vegetation cover.Therefore,the model can not retrieve the soil moisture content of underlying surface.This paper focuses on the above two apsects.Specifically:(1)A soil moisture estimation algorithm based on improved soil line is proposed.To solve the problem of inaccurate estimation of soil line based on remote sensing data,the objective function data items are constructed by using soil line equation,and the objective function regular terms are constructed by using water content estimation equation based on soil line.A cross-iteration framework is proposed,in which the estimation of soil line and water content are placed in the same framework.Soil line is corrected by error of water content estimation,water content is estimated by improved soil line,and solved by iterative cross-optimization.Finally,the accuracy of soil line estimation and water content estimation was improved.(2)A soil moisture estimation algorithm in three-dimensional NIR-Red-LST feature space is proposed.In order to estimate the water content in the high vegetation cover area,a new NIR-Red-LST three-dimensional feature space model was constructed by introducing the surface temperature.And from the point of view of energy balance,a new soil water index was proposed Based on the NIR-Red-LST space model.