基于二维光谱特征空间的土壤线自动提取算法

土壤线是在二维光谱空间中土壤纯像元光谱反射率按照大小排列而成的直线,它综合反映了具有不同水分条件的土壤含水状况,对于了解土壤和植被的理化性质和生态特征有着重要的意义。由于土壤线采用手工描绘效率低,因人而异,致使涉及土壤线的模型在业务化应用和推广中受到限制。为此,提出了一种土壤线自动提取算法,以及算法实现的流程和步骤,利用该算法可以基于遥感影像构建的二维光谱特征空间自动提取土壤线参数。通过时间序列和实际估算2种方法分析了算法的性能,发现在提取土壤线参数上具有较好的稳定性和精度,时间序列内提取土壤线斜率的平均偏差小于0.1,与实际估算的土壤线斜率偏差也很小。该算法已被应用于农田干旱监测模型PDI、MPDI和SPSI的土壤线自动提取运算中。结果表明,在保证模型精度的前提下,自动算法有效地提高了模型运算的效率,对农业干旱模型的业务化推广有实际的应用价值。

Soil line automatic identification algorithm based on two-dimensional feature space

Soil line is a straight line in two-dimensional spectral space,which is generated from a series of pure soil pixels ranged by reflectance.It is a comprehensive reflection of the soil’s characteristic in different water conditions,and helps to understand the physical,chemical and ecological properties of soil and vegetation.For a long time,the soil line has been manually identified with low efficiency and large variability,which limits the further application and promotion of soil line based models.Aimed to solve these problems,the paper proposed an soil line automatic identification algorithm and stated the details of its procedure.With this algorithm,the soil line and its fitting parameters can be automatically identified in the two-dimensional spectral space and calculated based on the reflectance of pixels in the remote sensing image.Through the validation of field experiments and time series data,the algorithm was proved to be a stable and accurate way to identify soil line.The algorithm was applied in drought monitoring model,such as PDI,MPDI and SPSI.The results show that the automatic algorithm can greatly raise the efficiency,improve the accuracy of soil line identification,and contribute to the widespread application of the drought models.