Effects of water tension and surface roughness on soil hyperspectral reflectance

• Published in: Geoderma 2021-03, Vol.385, p.114888, Article 114888
• Main Authors: Koch, Markus, Schodlok, Martin C., Guggenberger, Georg, Stadler, Susanne
• Format: Article
• Language: English
• Subjects: Hyperspectral; Imaging; Method; Soil moisture; Soil properties; Standardized; Water tension
• ISSN: 0016-7061; 1872-6259
• DOI: 10.1016/j.geoderma.2020.114888

•Soil surface roughness (SSR) changes reflectance intensity but not spectral features.•Prediction accuracy by PLSR is independent of SSRs.•Water tension (WT) changes reflectance intensity and absorption features.•Surface changes due to drying can be avoided by measuring WTs.•Good correlation between WT and reflectance/relative band depth is shown. Hyperspectral imaging techniques have been successfully employed to characterize soil chemical and physical parameters. However, such approaches can lack utility due to noise sources such as soil moisture (SM) and soil surface roughness (SSR). Moreover, studies dealing with SSR and SM lack comparability and standardized methods. To quantify these noise sources, reflectance patterns of different SM and SSR were measured in the visible-near (VNIR, 350–1000 nm), short-wave infra-red (SWIR, 1000–2500 nm) and longwave infrared (LWIR, 7500–13000 nm). For SSR analysis, three defined particle sizes were analyzed to represent different roughness states. For SM analysis, water tension (WT) was used to achieve a homogenous water pressure in the sample. At different WT stages changes in SSR were observed macro- and microscopically. We show that SSR and WT both regulate reflectance, in varying amounts. For both SSR and WT texture plays an important role. Different SSRs change the reflectance intensity. Hereby, quantitative analysis of soil components is not affected. For different WT, the surface roughness did not change visibly, but a formation of a microscopically fine-grained surface layer was observed. High WT (~pF 4.2) have shown to change reflectance and spectral feature geometry considerably. A correlation between WT and reflectance/relative band depth becomes apparent. WT thus influences characteristic spectral signals of soil constituents. For retrieving soil properties by remote sensing techniques, SSR, relatively speaking, is of minor importance. In this context, the WT of investigated samples is recommended to be addressed to avoid its impact on the estimation of soil properties.