Hyperspectral based approach to investigate topsoil characteristics of different taxonomic units of El-Fayoum depression

Abstract

Present paper is an initial attempt to study chemical and hyperspectral characteristics of topsoil of different soil taxonomic units in El-Fayoum depression. Archival data, field survey, soil profiles sampling and analyses, hyperspectral data assessment and spectral indices applications were integrated with statistical investigations to achieve the aim of the present study. Soil map of El-Fayoum shows the following main sub-great groups; Typic Torrifluvents, Typic Torrerts, Typic Quartizpsamments, Typic Calciorthids, Calcic Gypsiorthids and Typic Saliorthids. Typic Torriorthents recorded the highest levels of EC (260.5 dS/m), ESP (73.6), CEC (107.5 meq/ 100 g soil) and K+ (91.6 meg/L). However, the highest levels of Na (53.69 meq/L), Ca (28.61 meq/L) and Cl (86.01 meq/L) were associated with Typic Torripsamments. Typic Haplocalcids reported the maximum levels of SO4 (35.9 meq/L). The analyses of Tukey’s and one way ANOVA showed a great efficiency of SWIR2 to discriminate topsoil of all investigated taxonomic units of soil. On the other hand, topsoil of Typic Torriorthents, Typic Torripsamments, Typic Haplocalcids and Typic Quartizipsamments can be identified in all spectral bands (UV, VIS and IR) except the green band which showed limitation to define Typic Torripsamments and Typic Haplocalcids. The power of investigated hyperspectral bands to discriminate topsoil of various soil taxonomic units can be ordered as follows; SWIR2 > SWIR1 > NIR > Blue > Red > UV > Green. Based on linear regression analyses, innovative model for assessing gravel % was generated using SAVI retrieved from hyperspectral data, giving a promising accuracy (70 %). Further, a novel hyperspectral library for the topsoil of the investigated six sub-great groups of soil was developed for further applications of soil taxonomy on basis of hyperspectral data. Present findings are useful for encouraging the wide utilization of in-situ hyperspectral data sets for studying soil different variables.