TY - JOUR
T1 - Temperate forest soil pH accurately Quantified with image spectroscopy
AU - Abdullah, H.
AU - Skidmore, A.K.
AU - Siegenthaler, Andjin
AU - Adiningrat, Devara Prawira
AU - Duan, Yiwei
AU - Rousseau, Mélody
PY - 2024/4
Y1 - 2024/4
N2 - Forest canopies to some extent obscure passive reflectance of soil traits such as pH, as well as below-canopy vegetation, in the optical to middle infrared portions of the electromagnetic spectrum (approximately 400–2500 nm) which are typically used in airborne and spaceborne image spectrometers. In this study, we present, for the first time, an accurate estimation of soil pH across extensive areas using hyperspectral imaging data obtained from the DLR Earth Sensing Imaging Spectrometer (DESIS) satellite. Furthermore, we investigate the impact of predicted soil pH variation on the concentrations of micronutrients in both leaves and soil. Our modelling is based on a comprehensive in-situ field campaign conducted during the summers of 2020 and 2021. This campaign collected soil pH data for model calibration and validation from 197 plots located across three distinct temperate forest sites: Veluwezoom and Hoge Veluwe National Parks in the Netherlands, as well as the Bavarian Forest National Park in Germany. The soil pH for each test site was accurately predicted by means of a partial least squares regression (PLSR) model, root mean square error (RMSEcv) of 0.22 and the cross-validated coefficient of determination (R2CV) of 0.66. Our findings demonstrate that there are patches of extremely low soil pH possibly due to ongoing soil acidification processes. We saw a particularly significant decrease in soil pH (p ≤ 0.05) in the coniferous forests when compared to the deciduous forest. The acidification of forest soils had a profound impact on the variation of soil and leaf micronutrient content, particularly iron concentration. These results highlight the potential of image spectroscopy data from the DESIS satellite to monitor and estimate soil pH in forested areas over extensive areas given sufficient data. Our findings hold significant implications for soil pH monitoring programs, enabling forest managers to assess the impact of their management practices and gauge their effectiveness in maintaining soil and forest vitality.
AB - Forest canopies to some extent obscure passive reflectance of soil traits such as pH, as well as below-canopy vegetation, in the optical to middle infrared portions of the electromagnetic spectrum (approximately 400–2500 nm) which are typically used in airborne and spaceborne image spectrometers. In this study, we present, for the first time, an accurate estimation of soil pH across extensive areas using hyperspectral imaging data obtained from the DLR Earth Sensing Imaging Spectrometer (DESIS) satellite. Furthermore, we investigate the impact of predicted soil pH variation on the concentrations of micronutrients in both leaves and soil. Our modelling is based on a comprehensive in-situ field campaign conducted during the summers of 2020 and 2021. This campaign collected soil pH data for model calibration and validation from 197 plots located across three distinct temperate forest sites: Veluwezoom and Hoge Veluwe National Parks in the Netherlands, as well as the Bavarian Forest National Park in Germany. The soil pH for each test site was accurately predicted by means of a partial least squares regression (PLSR) model, root mean square error (RMSEcv) of 0.22 and the cross-validated coefficient of determination (R2CV) of 0.66. Our findings demonstrate that there are patches of extremely low soil pH possibly due to ongoing soil acidification processes. We saw a particularly significant decrease in soil pH (p ≤ 0.05) in the coniferous forests when compared to the deciduous forest. The acidification of forest soils had a profound impact on the variation of soil and leaf micronutrient content, particularly iron concentration. These results highlight the potential of image spectroscopy data from the DESIS satellite to monitor and estimate soil pH in forested areas over extensive areas given sufficient data. Our findings hold significant implications for soil pH monitoring programs, enabling forest managers to assess the impact of their management practices and gauge their effectiveness in maintaining soil and forest vitality.
KW - ITC-HYBRID
KW - UT-Hybrid-D
U2 - 10.1016/j.rsase.2024.101161
DO - 10.1016/j.rsase.2024.101161
M3 - Article
SN - 2352-9385
VL - 34
JO - Remote Sensing Applications: Society and Environment
JF - Remote Sensing Applications: Society and Environment
M1 - 101161
ER -