Key points are not available for this paper at this time.
A new approach to estimate surface soil moisture from reflectance data in the solar spectral range (350-2500 nm) is presented, called the Normalized Difference Soil Moisture Index (NSMI). The motivation for this new index is to make use of spectral features that fulfill the criteria of robustness against covariates, physical comprehensibility and easy applicability in the field and from remote sensing platforms. Spectral measurements were taken in the laboratory from 121 prepared as well as 467 natural soil samples consisting of different sands and clayey substrates originating from a lignite mine reclamation site. While the preparation procedure performed on samples from the first group removed the covariates' influence on the reflectance spectra, the natural samples in the second group maintained the influencing factors like impurity, crusts, and organic matter. In a systematic way all wavelengths were combined in different spectral feature approaches and optimum bands or band combinations were found for linear correlation with soil moisture. For the natural samples, the NSMI achieved best results in this study with R 2 of 0.61 by combining reflectance values at 1800 and 2119 nm. This value increased to 0.71 when samples with significant xylite proportions had been removed. Analyses on the effect of single covariates showed that neither surface crusts nor substrate heterogeneity changed the correlation between soil moisture and the NSMI significantly. The NSMI can therefore be seen as a new index for quick assessment of surface or near-surface soil moisture directly in the field using spectral instruments.
Haubrock et al. (Wed,) studied this question.