中文摘要：

在棉花大田水分试验的基础上,采用自主设计的不同土层取样方法,同步获取了棉花冠层高光谱数据和不同深度土壤的水分含量数据以及棉花冠层水分含量数据,分析了棉花冠层含水量与土壤含水量之间的关系、棉花冠层高光谱数据与土壤含水量之间的相关性,构建了基于棉花冠层高光谱数据的土壤水分含量反演模型。结果表明：不同土层的水分含量具有较大差异,棉花冠层对不同土层水分含量的响应程度不同,0~30 cm土层水分含量与棉花冠层含水量的相关性最强,决定系数达到0.58;棉花冠层反射率与土壤水含量在可见光波段呈负相关,近红外波段呈正相关;在所有以棉花冠层高光谱数据的不同变换形式构建的不同土层含水量的PLSR反演模型中,以反射率倒数对数所建的模型对0~30 cm土层和以反射率对数所建模型对0~10 cm土层含水量的预测RPD均达到2.0以上,具有较好的预测能力,其余模型的预测效果不理想。

英文摘要：

Water content and reflectance spe ctroscopy data of cotton canopies and soil water content data from different soil layers were acquired simultaneously by using the independent design method at different soil depths on the basis of cotton field moisture test. The relationships between cotton canopy content and reflectance spectroscopy data of cotton canopy and soil water content were analyzed, and the inversion model of soil water content was established based on the spectral reflectivity of cotton canopy. The results showed that soil water content varied obviously from different soil layers, and that the influence of cotton canopy on moisture was different at different soil layers. Meanwhile, the correlation between soil water content（0- 30 cm） and water content of cotton canopy was the best with the determination coefficient value of 0.58. The correlation between reflectance of cotton canopy and soil water content was negative in visible bands, but positive in infrared bands. Among all of partial least square regression（PLSR） models including several different methods based on spectral reflectivity of vegetation canopy to monitor soil moisture at different depths, the ratio of performance to derivation（RPD） of PLSR model for monitoring soil water content were both more than 2.0 at soil depth of 0- 30 cm based on logarithm of reciprocal of reflectance and at soil depth of 0- 10 cm based on logarithm of reflectance, which could relatively and accurately predict the corresponding soil water content. However, the other models were not satisfied with the need of predictive ability.