Skip to main content
ARS Home » Plains Area » Temple, Texas » Grassland Soil and Water Research Laboratory » Research » Publications at this Location » Publication #387137
Research Project: Contributions of Climate, Soils, Species Diversity, and Management to Sustainable Crop, Grassland, and Livestock Production Systems

Location: Grassland Soil and Water Research Laboratory

Title: National soil organic carbon map of agricultural lands in Nepal

Author
item LAMICHHANE, SUSHIL - University Of New England
item Adhikari, Kabindra
item KUMAR, LALIT - Eastcoast Geospatial Consultants

Submitted to: Geoderma Regional
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/22/2022
Publication Date: 8/2/2022
Citation: Lamichhane, S., Adhikari, K., Kumar, L. 2022. National soil organic carbon map of agricultural lands in Nepal. Geoderma Regional. 30. Article e00568. https://doi.org/10.1016/j.geodrs.2022.e00568.
DOI: https://doi.org/10.1016/j.geodrs.2022.e00568

Interpretive Summary: Spatial heterogeneity of soil organic carbon (SOC) distribution can be modelled with a number of machine learning techniques and environmental variables used as SOC predictors. This study applied four machine learning algorithms, namely Random Forest (RF), Cubist, Extreme Gradient Boosting (XGB) and Support Vector Machine to map topsoil (0-20 cm) SOC from agricultural lands in Nepal. The RF model performed the best among all tested models, closely followed by the Cubist and then the XGB model. This study provides a baseline national map of topsoil SOC content from the agricultural land areas across Nepal.

Technical Abstract: Reliable and accurate soil organic carbon (SOC) maps at high resolutions are needed to monitor and improve SOC status in croplands and for agro-environmental applications. Topsoil (0-20 cm) SOC content in croplands was predicted and mapped across Nepal using state-of-the-art soil mapping techniques. Altogether 25,312 recent SOC observations were used to build and evaluate prediction models derived from four machine learning algorithms, namely Random Forest (RF), Cubist, Extreme Gradient Boosting (XGB) and Support Vector Machine. Twenty two environmental variables were selected as SOC predictors based on their correlation with measured SOC contents and non-collinearity with other covariates. The predictive performance of these models was compared using calibration (80% observations) and validation (20% observations) datasets. The performance of the models was also compared against a global SOC dataset compiled by International Soil Reference and Information Centre (ISRIC). The best model among the four algorithms was used to map and quantify the spatial distribution of SOC contents and standard deviation of all models as an assessment of prediction uncertainty. The RF model performed the best among all tested models, closely followed by the Cubist and then the XGB model. The predictive performance of all of these models was better than the global SOC prediction from ISRIC. This study provides a baseline map for the topsoil SOC contents from the croplands in Nepal and also provides a reference for similar SOC mapping studies.