Estimating natural soil drainage classes in the Wisconsin till plain of the Midwestern U.S.A. based on lidar derived terrain indices: Evaluating prediction accuracy of multinomial logistic regression and machine learning algorithms

• Published in: Geoderma Regional Vol. 35; p. e00728
• Main Authors: Rahmani, Shams R., Libohova, Zamir, Ackerson, Jason P., Schulze, Darrell G.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.12.2023
• Subjects: Alfisols; Artificial neural network; C5.0 decision tree model; class; digital elevation models; Digital soil mapping (DSM); landscapes; lidar; lidar digital elevation model (DEM); Machine learning algorithms; Mollisols; Multinomial logistic regression; Natural soil drainage class; neural networks; prediction; Random forest; regression analysis; soil; SSURGO; Stratified random sampling; subsurface drainage; Terrain attributes; waste disposal; Wisconsin; Wisconsin; C5.0 decision tree model; Machine learning algorithms; Alfisols; Random forest; Mollisols; Terrain attributes; Stratified random sampling; Natural soil drainage class; Digital soil mapping (DSM); Multinomial logistic regression; Artificial neural network; SSURGO; lidar digital elevation model (DEM)
• ISSN: 2352-0094; 2352-0094
• DOI: 10.1016/j.geodrs.2023.e00728

Accurate spatial prediction of natural soil drainage condition is not only important for agriculture and hydrological modeling but also for installing subsurface drainage and onsite waste disposal systems. For this research, 154 sites were selected based on a stratified random sampling method. For each site, drainage class was identified based on visual examination of soil cores. A digital elevation model developed from lidar data was used to derive seven terrain indices. Terrain indices were used to predict drainage class using four prediction models: multinomial logistic regression and three machine learning algorithms (random forest, C5.0, and artificial neural network). Based on 30% random hold-back validation data, all digital soil mapping (DSM) models provided similar results. The artificial neural network provided relatively higher overall accuracy of 70%, a kappa coefficient of 0.59, and a Brier score of 0.34. The rest of the DSM models as well as SSURGO provided an accuracy that ranged between 64 and 66%, a kappa coefficient that ranged between 0.52 and 0.54, and a Brier score that ranged between 0.41 and 0.46. •Multinomial logistic regression and machine learning models provided similar predictions for natural soil drainage classes•In a low relief environment, the natural soil drainage classes can be effectively predicted with LiDAR DEM•Digital soil mapping models provided slightly better results than conventional soil mapping