Browsing by Author "Md Sarfaraz Asgher"

Now showing 1 - 3 of 3

Advanced Bivariate Geostatistical Modeling for High-Resolution Landslide Susceptibility Zonation for Effective Risk Management in the Northwestern Himalaya, India
(Springer Science and Business Media Deutschland GmbH, 2025) Imran Khan; Vikas Yadav; Ashutosh Kainthola; Harish Bahuguna; Debi Prasanna Kanungo; Ranjan Kumar Dahal; Shantanu Sarkar; Md Sarfaraz Asgher
Frequent landslides in the northwestern Himalaya India (NHI) region cause significant loss of life and property, making landslide susceptibility zonation (LSZ) crucial for identifying vulnerable areas. This study aims to develop LSZ maps for the NHI using four bivariate geostatistical models: Frequency Ratio (FR), Weight of Evidence (WoE), Information Value (IV), and Yule's Coefficient (Yc). A total of 38,697 landslides, covering 149.50 km2, were analyzed. The data was split into 70% for training and 30% for testing, ensuring robust model validation. Twelve causative factors were considered, including slope angle, slope aspect, slope curvature, relative relief, terrain roughness index, geomorphon, distance to drainage, land use land cover, lithology, distance to fault/thrust, earthquakes, and rainfall. The models identified high to very high susceptibility zones, covering 28.7%, 32.8%, 48.1%, and 48.2% of the region for the Yc, FR, WoE, and IV models, respectively. ROC analysis revealed that the FR model achieved the highest accuracy, with 0.845 (84.5%) for both validation and prediction. The IV model followed with ROC values of 0.833 (83.3%), while the Yc model performed similarly, with values of 0.831 (83.1%). The WoE model exhibited slightly lower accuracy, with ROC values of 0.830 (83.0%) for validation and 0.831 (83.1%) for prediction. Both the WoE and IV models covered over 98% of landslide areas in high and very high susceptibility zones, indicating a tendency to overestimate highly susceptible areas. The study suggests that the FR and Yc models are particularly effective for LSZ and risk assessment. These results provide valuable insights for hazard management, aiding researchers, planners, and policymakers in selecting appropriate models for LSZ and mitigating landslide risks in other vulnerable regions. © King Abdulaziz University and Springer Nature Switzerland AG 2024.
Advanced modeling of forest fire susceptibility and sensitivity analysis using hyperparameter-tuned deep learning techniques in the Rajouri district, Jammu and Kashmir
(Elsevier Ltd, 2025) Lucky Sharma; Mohd Rihan; Narendra Kumar Rana; Shiva Kant Dube; Md Sarfaraz Asgher
Forest resources are crucial for sustaining the global population, regulating climate services, and maintaining overall ecological balance. However, forest fires are causing a significant loss of forest cover worldwide. In this context, advanced deep learning techniques, which are novel to date, have been utilized to prepare forest fire susceptibility mapping. The present study aimed to predict forest fire susceptibility using three hyper-tuned techniques: deep neural network (DNN), elman neural network (ENN), and convolutional neural network (CNN). To identify the importance of influencing factors, sensitivity analysis was conducted using the DNN. The forest fire susceptibility map (FFSM) was categorized into five susceptibility zones: very high, high, moderate, low, and very low. Results indicated that the southern and southeastern parts of the study area are most prone to forest fires. The proportion of high susceptibility zone in the study area was found to be 34% for DNN, 37% for ENN, and 30% for CNN. Among all the models, DNN outperformed the others, achieving the highest accuracy of 0.8925, compared to ENN (0.8825) and CNN (0.87). Sensitivity analysis further revealed that evapotranspiration, temperature, land surface temperature (LST), distance to roads, aridity, and elevation were the most influential factors contributing to forest fires in the region. This study demonstrates an advanced and globally relevant approach to forest fire susceptibility analysis. The findings may be crucial for stakeholders and policymakers to make informed decisions regarding effective forest fire management and to protect vulnerable communities from unexpected losses. © 2025 COSPAR
Integrating FR, MFR and IV Models for Landslide Susceptibility Zonation Mapping in Joshimath Watershed, Uttarakhand, India
(Springer Science and Business Media Deutschland GmbH, 2025) Imran Khan; Ashutosh Kainthola; Harish Bahuguna; Vishnu Himanshu Ratnam Pandey; Md Sarfaraz Asgher; Ashish Bhardwaj; Deepali Gupta
Joshimath watershed in Uttarakhand, India, is known for experiencing creep, subsidence, and frequent minor and major landslides. Identifying zones prone to landslides through landslide susceptibility zonation (LSZ) is crucial for city planners to mitigate risks and reduce potential losses. This study employs three widely used and accurate statistical models: Frequency Ratio (FR), Modified Frequency Ratio (MFR), and Information Value (IV) to assess LSZ. A dataset of 271 landslides, derived from time-series satellite images, was utilized, with 70% (190 events) allocated for model training and 30% for validation. The analysis considered fifteen factors influencing landslide susceptibility, including slope, aspect, curvature, proximity to drainage, proximity to faults, proximity to roads, geomorphon, landform, altitude, lithology, and LULC data from both Google Earth and ESRI, RR, SPI, and TWI were evaluated, offering a comprehensive view of the various factors that may affect landslide occurrence. Based on ranking, the most influential factors are geomorphon, proximity to faults and drainage, proximity to roads, and aspect. In contrast, LULC (ESRI), RR, altitude, lithology, and slope demonstrate limited influence, while TWI, SPI, and curvature are the least influential factors. The susceptibility maps were classified into three categories. The FR model identified 59.5% of the area as low susceptibility, 32.1% as medium, and 8.3% as high, with 65.9% of landslides occurring in high-susceptibility zones. The MFR model classified 48.3% of the area as low susceptibility, 27.1% as medium, and 24.6% as high, with 78.4% of landslides located in high-susceptibility zones. The IV model classified 37.8% of the area as low susceptibility, 41.1% as medium, and 21.2% as high, with 77.4% of landslides occurring in high-susceptibility zones. ROC analysis validated the models’ predictive capabilities, with the FR model achieving the highest accuracy in both the Landslide Susceptibility Index (LSI) and LSZ at 83.1% and 84.5% AUC, respectively. The MFR and IV models also demonstrated commendable performance, providing valuable insights for landslide risk assessment. The findings emphasize the importance of model selection in LSZ mapping, highlighting the FR and MFR models as effective tools for risk management and land-use planning in landslide-prone areas. This study contributes to landslide susceptibility modeling and provides a framework for future research in geological hazard assessment. © The Author(s), under exclusive licence to Shiraz University 2024.