Predictive soil mapping in the Boreal Plains of Northern Alberta by using multi-temporal remote sensing data and terrain derivatives

Abstract

As Canada’s vast Boreal Plains are extensively managed, predictive soil mapping could be used as an effective tool to generate high-resolution soil information for the region to inform sustainable resource management. This study aimed to investigate the use of multi-temporal remote sensing data and terrain derivatives to map soil types in the region. A method of constraining sub-group and great-group soil type predictions based on the predictions at higher-order levels (great-group and order, respectively) was tested. Sentinel time series median values obtained using Google Earth Engine were tested in combination with first- and second-order digital elevation model derivatives for use as predictor variables in the predictive models. A recursive feature selection process was implemented to reduce the number of predictor variables used in model training. Soil classes were predicted at the order, great group, and subgroup levels and two approaches were tested. In the first approach, models were unconstrained based on previous predictions. In the second approach, models were constrained to predict only soil great group classes that occur within the predicted soil order for a given location and similarly predict only soil subgroup classes that occur within the predicted soil great group for a given location. Determined through independent validation testing, the most probable predicted soil maps had overall accuracies ranging from 42 to 68% and Kappa scores ranging from 0.33 to 0.48. Overall, the constrained models had the best performance of the approaches tested.

A Proposed Methodology for Determining the Economically Optimal Number of Sample Points for Carbon Stock Estimation in the Canadian Prairies

“…The optimal features were first selected using a recursive feature elimination process using only the training dataset [48]. The first step of this process involved testing for correlation amongst features.…”

Advances in soil survey and classification in Canada

“…In this collection, manuscripts apply various machine learning techniques to address classification (Heung et al 2022;Kiss et al 2023;Saurette 2022;Sorenson et al 2022Sorenson et al , 2023 and regression (Deragon et al 2022;Paul et al 2023;Saurette 2022;Zhang et al 2022) applications in DSM. When addressing classification applications of DSM, the methods of analysis fall within two broad categories: polygon disaggregation and point-based mapping approaches.…”

“…Kiss et al (2023) carried out a soil survey campaign to differentiate noncalcareous wetland, calcareous wetland, and upland units for the Prairie Pothole Region of Saskatchewan, while Saurette (2022) describe the prediction of soil textural classes for the ongoing Ottawa soil survey. In contrast, Sorenson et al (2022) take a unique approach by using an unconventional source for soil data, environmental impact assessment reports which reflect the importance, availability, and potential of private sector data in developing or improving DSMs. Maps of soil taxonomic units were produced for the Boreal Plains of Northern Alberta.…”

Evaluation of two miniaturized FT‐NIR spectrometers for rapid soil property analysis

“…• Scales 2 and 3 • Scales 3 and 4 • Scales 2, 3 and 4 • Scales 2, 3, 4 and 5 • Scales 3, 4 and 5 • Scales 4 and 5 Features were selected using a recursive feature selection process, as mentioned in Sorenson et al (2022). First, correlated features were removed based on a threshold value of 0.9 using the caret package find correlation function (Kuhn et al, 2016).…”