基于高光谱的河套灌区农田表层土壤质地反演研究

Surface Soil Texture of Field Inverted Using Hyper-spectrum in Hetao Irrigation

以河套灌区解放闸灌域农田表层0-10cm土壤为研究对象，研究风干土样的不同粒级含量与光谱特性关系，遴选出反映土壤粘粒、粉粒、砂粒含量各自敏感的光谱波段，分别建立土壤粘粒、粉粒、砂粒含量与敏感波段的一元线性回归模型和BP神经网络模型，模型验证结果表明：土壤粘粒、粉粒、砂粒含量的一元回归模型与BP神经网络模型精度基本一致，且都在85g以上。该研究结果可为应用高光谱遥感图像大范围识别河套灌区的土壤质地提供理论依据。

Study on the spectral characteristics of soil texture by remote sensing is the basis of recognition of the soil texture. Se- lecting surface soil of field 0-10 cm in Jiefangzha Irrigation Sub-district of Hetao Irrigation as an study object and measuring texture and spectrum of these air-dried soil, sensitive bands reflecting content of clay, silt and sand were found based on labora- tory hyper-spectrum data. Linear regression model and BP neural networks model were proposed to predict content of clay, silt, sand and the respective sensitive bands. The validation results show that prediction precision of clay, silt, sand linear regression model was basically consistent with that of BP neural networks model. The prediction precision of two models was more than 85%. The result may provide theoretical basis for identifying soil texture based on hyper-spectral image in Hetao Irrigation.