融合多源遥感数据的夏玉米土壤水分反演方法对比研究

Comparative Study on Soil Moisture Retrieval Methods for Summer Maize Using Multi-source Remote Sensing Data Fusion

为了解决在夏玉米植株高度较高(>1.5 m)情况下,无人机遥感土壤水分反演过程中冠层与地表之间多次散射对微波后向散射的衰减问题,寻找合适的反演方法。通过融合运用无人机多光谱和热红外数据、Sentinel-1A SAR卫星数据,结合田间实测数据,对植被覆盖下的土壤水分反演与精度验证进行研究;采用温度植被干旱指数(TVDI)、水云模型(WCM)以及引入MIMICS模型参数的改进水云模型(Improved WCM)3种方法进行土壤水分反演。其中,TVDI方法拔节期反演精度R2为0.50(10 cm)和0.42(20 cm),乳熟期反演精度R2为0.49(10 cm)和0.46(20 cm);WCM方法拔节期反演精度R2为0.53(10 cm)和0.44(20 cm),乳熟期反演精度R2为0.18(10 cm)和0.02(20 cm);Improved WCM方法拔节期反演精度为0.76(10 cm)和0.69(20 cm),乳熟期反演精度为0.78(10 cm)和0.74(20 cm)。采用引入MIMICS模型参数的改进水云模型方法得到的夏玉米2个生育期的反演效果,明显优于水云模型方法和温度植被干旱指数方法;3种方法的2个生育期反演精度均为10 cm高于20 cm。因此,引入MIMICS模型参数的改进水云模型方法更适合于玉米植株较高情况下的10 cm土壤含水量反演。

To address the challenge of multiple scattering between the canopy and the surface during soil moisture retrieval using drone remote sensing under high corn plant height(>1.5 m),we searched for an appropriate retrieval method.In this study,we used fused unmanned aerial vehicle(UAV)multispectral and thermal infrared data,Sentinel-1A synthetic aperture radar(SAR)satellite data,and field-measured data to investigate the soil moisture retrieval and accuracy verification under vegetation coverage.We employed three methods:temperature vegetation drought index(TVDI),water cloud model(WCM),and an improved WCM method that introduces MIMICS model parameters.The TVDI method had a retrieval accuracy of R2=0.50(10 cm)and 0.42(20 cm)during the jointing stage period,and R2=0.49(10 cm)and 0.46(20 cm)during the milk-ripe period.The WCM method had a retrieval accuracy of R2=0.53(10 cm)and 0.44(20 cm)during the jointing stage period,and R2=0.18(10 cm)and 0.02(20 cm)during the milk-ripe period.The improved WCM method had a retrieval accuracy of R2=0.76(10 cm)and 0.69(20 cm)during the jointing stage period,and R2=0.78(10 cm)and 0.74(20 cm)during the milk-ripe period.The improved WCM method using MIMICS model parameters outperformed both the WCM method and the TVDI method in both growth stages.The retrieval accuracy of all three methods was higher at 10 cm than at 20 cm in both growth stages.Therefore,the improved WCM method with MIMICS model parameters is more suitable for soil moisture retrieval at a depth of 10 cm under high corn plant height conditions.