Assessment of land degradation using machine‐learning techniques: A case of declining rangelands

Abstract

Increased use and increasing demands pose serious threats to rangelands. In this study, we document a pronounced downward trend in rangeland quality in the Alborz Mountains in Firozkuh County, Iran using analysis of three machine‐learning models (MLMs). A total of 1,147 transects were established to evaluate the rangeland quality trends from field data collected over a 7‐year period. Twelve independent conditional factors were analyzed for their relationships to range quality through three MLMs—Random Forest (RF), classification and regression tree (CART), and support vector machine (SVM). Based on assessments of the trained and validated models, RF, with a ROC‐AUC = 0.96, was determined to be the most robust. The results show that about 20% of the rangeland in the study area is in a critically degraded condition. Distances from roads and livestock density are the two factors most strongly linked to degradation. These results, in combination with field observations, indicate that the rangelands of the study area face two major challenges (overgrazing and early grazing) that require new strategies to mitigate and prevent damages. This study may provide important guidance for evaluating rangeland conditions in other regions of the world.

Impact of ineffective measures on the increasing land degradation in a monsoon‐dominated region of India: Issues and policy implications

“…The badland topography, through the formation and development of the gully formation, is widely found and not capable of agricultural production as well as vegetation. Water‐induced soil erosion reduces soil fertility and vegetation cover and poses a threat to the ecosystem and infrastructure (Yousefi et al, 2021). In the field of agricultural and socioeconomic conditions, surface erosion through the formation of gullies is a major step of land degradation in India (Pani & Carling, 2013; Yadav & Bhushan, 2002).…”

A new conceptual framework for spatial predictive modelling of land degradation in a semiarid area

“…Furthermore, its computations may be mode complicated compared with other modelling approaches. Yousefi et al (2021) have used this model in the evaluation of rangelands degradation utilizing machine‐learning algorithms.…”

Assessment of Ensemble Models for Groundwater Potential Modeling and Prediction in a Karst Watershed

“…This value expresses a model's ability to correctly distinguish positive and negative observations in the validation data. High sensitivity indicates a high number of true predictions (true positives), and high specificity indicates a low number of false positives [10,45]. False-and true-positive rates are shown on an X and Y chart.…”

“…This suggests that a linear combination of two or more variables should be considered. When there is multicollinearity among the factors used in a model, the coefficients in the resulting model are invalid because the effect of each explanatory variables on the response variable simultaneously includes the effects of the other variables in the model [37,45,46]. The variance of regression coefficient estimators, therefore, is increased and model's prediction reflects a larger potential for error.…”