Estimating the diurnal blue-sky albedo of soils with given roughness using their laboratory reflectance spectra

• Published in: Journal of quantitative spectroscopy & radiative transfer Vol. 217; pp. 213 - 223
• Main Authors: Cierniewski, Jerzy, Ceglarek, Jakub, Kaźmierowski, Cezary
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01-09-2018
• Subjects: Bare soil; Blue-sky albedo; Hyperspectral reflectance; Roughness; Bare soil; Hyperspectral reflectance; Blue-sky albedo; Roughness
• ISSN: 0022-4073; 1879-1352
• DOI: 10.1016/j.jqsrt.2018.06.003

•A model for estimation of blue-sky albedo of soil with defined roughness is proposed.•The model uses laboratory reflectance spectra of related soil samples.•The model was calibrated on soil samples obtained in Poland, Israel and France.•Best accuracy is given by the reflectance data transformed to 2nd derivative.
Various soil properties derived from soil reflectance spectra can effectively provide reliable data for soil monitoring, digital soil mapping, precision agriculture, and environmental modelling. This paper focuses on the improvement of the equations published in the previous paper predicting the diurnal blue-sky albedo of bare soils with given roughness using their reflectance spectra measured in laboratory conditions. The improved equations based on a 33% increase in the number of analysed soil samples and primarily on the quality expansion of the newly tested soil population. The albedo of the soils was measured when they were air-dried under clear sky conditions by albedometers working in a spectral range of 335–2800 nm. The roughness of the cultivated soils was measured using stereo-photographs taken with a digital camera and described by two roughness indices. The laboratory reflectance spectra of the soils were obtained using a spectroradiometer with a Hi-Brite Muglight receptor in the wavelength range of 350–2500 nm. The raw soil reflectance data and their transformations (linearization, normalization and filtering) were used in the equations mentioned above. Transforming spectra to the 2nd derivative using the Savitzky–Golay method gave the strongest accuracy of the prediction expressed by the coefficient of determination equals 0.91 and root mean square error 0.03.