基于时序Sentinel-1A数据的农田土壤水分变化检测分析

Soil Moisture Change Detection over Bare Agricultural Area by Means of Time-series Sentinel-1A SAR Data

土壤水分是作物生长的基本条件,同时也是作物长势监测、估产和旱情监测的重要参数。基于变化检测模型,利用不同时相Sentinel-1A数据估算农田土壤水分变化信息。首先利用积分方程模型(Integral Equation Model,IEM)模拟数据分析雷达后向散射系数变化与土壤水分变化之间的关系,在土壤表面粗糙度恒定的情况下,土壤湿度变化与雷达后向散射系数变化具有高度相关性,验证了变化检测模型用于估算土壤水分变化的合理性。在此基础上,利用河北邯郸研究区时序Sentinel-1ASAR数据和现场实测数据构建土壤水分变化检测模型,从而利用雷达后向散射系数变化估算土壤水分变化信息。结合现场实测数据得到,最小二乘和支持向量回归模型反演结果的均方根误差(RMSE)分别为5.1vol%和4.6vol%,决定系数(R2)分别为0.65和0.73。验证了时序Sentinel-1A数据用于监测农田土壤水分变化的实用性。

Soil moisture is the basic condition for crop growth,which is also a critical parameter for crop condition monitoring,yield estimation and drought monitoring.The objective of this study was to investigate the potential of time-series Sentinel-1ASAR data for soil moisture change detection.The change detection model was firstly evaluated based on the IEM simulation data.This approach relies on the assumption that the vegetation and surface roughness were constant during the study process.Therefore,the effect of vegetation and surface roughness on radar backscattering signal can be eliminated by the change detection model.Thus,the change of backscattering coefficient can characterize the soil moisture change.The simulation data indicated that the soil moisture change showed high correlation with the change of backscattering coefficient,which verified the rationality of change detection model for soil moisture change detection.Then the relationship between the backscattering coefficient change and soil moisture change was established based on the time-series Sentinel-1A data and field measured soil moisture.The root mean square error（RMSE）and determination coefficient（R^2）of the estimated results was（5.1vol%,0.65）and（4.6vol%,0.73）for least square and support vector regression model,respectively.The experimental results verified the practicability of time-series Sentinel-1ASAR data for soil moisture change detection over agricultural areas.