Title: Understanding the relationship between geographical factors and dirt road density at the local level: a case study of Orkhon, Khongor, Darkhan, Sharyn Gol and Javkhlant soums

The Central Asian country of Mongolia is severely affected by desertification and land degradation. One of the main drivers of degradation is dirt road expansion, often informal and unplanned nature. A better understanding of the geographical factors that exacerbate dirt road expansion is helpful to develop management strategies to reduce dirt road-related land degradation. The aim of this study was to investigate the relationship between dirt roads and geographical factors at the local level using a case study area in the central northern province of Selenge. The study utilised geostatistical techniques to explore a total of 17 variables representing key natural and human characteristics. A total length of 2,998 km of dirt roads was determined in the case study area through satellite image analysis. Several dirt road clusters identified during the analysis indicated that dirt roads contribute significantly to land degradation in the study area. Ordinary Least Squares (OLS) and Geographically Weighted Regression (GWR) models showed some weaknesses in their capabilities. Multicollinearity was identified as one of the major challenges in such a local level analysis. The OLS model could explain up to 7% of the dirt road expansion in the study area, whereas the GWR model could explain 13%. The modelling results suggest that the effect of geographical factors on dirt road expansion varied throughout the study area. Rural settlement centres (soum and bag centre) and terrain slopes could influence an increase in dirt road density at the local level. Overall spatial analysis could identify some overall patterns in the data that may be valuable for identifying the most pressing geographic areas and their specific characteristics that require addressing through specific planning and management approaches.