| Abstract Submission No. | ABS-3.1-0082 |
| Title of Abstract | Landslide susceptibility analysis along the National Highway (NH)-2 in the Wokha-Mokokchung region, Nagaland: an AI-enhanced geospatial approach |
| Authors | Dr. Maisnam Devika Devi* |
| Organisation | CSIR-North East Institute of Science and Technology (NEIST) |
| Address | csir-north east institute of science and technology (neist), jorhat jorhat, Assam, India Pincode: 785006 E-mail: maisnamdevika@gmail.com |
| Country | India |
| Presentation | Poster |
| Abstract | The Wokha-Mokokchung region along the National Highway (NH)-2 in Nagaland is highly prone to landslides due to its complex geology, rugged topography, and intense monsoonal rainfall. The area lies within Seismic Zone V and comes under the Disang Formation, comprising highly weathered shales, siltstones, and sandstones. Structural weaknesses, represented by extensive thrusts and faults, aggravate slope instability in this region. Combined with steep gradients and annual precipitation of ~2000¿4000 mm, these factors result in recurrent slope failures that disrupt regional connectivity and pose significant hazards. The study introduces an AI-enhanced geospatial framework that integrates machine learning with traditional parameters such as topography, hydrology, geology, environment, rainfall intensity, and anthropogenic influences. The Random Forest (RF) algorithm, known for its ability to process high-dimensional geospatial data, demonstrates superior predictive accuracy compared to conventional statistical methods. This study presents an AI (Random Forest)-enhanced geospatial framework for landslide susceptibility mapping along NH-2, which combines traditional conditioning parameters with the superior predictive performance and interpretability of this ensemble learning algorithm. The proposed framework generates high-resolution landslide susceptibility maps, which are rigorously validated through statistical measures including ROC¿AUC analysis, precision¿recall curves, and field verification during accessible seasons. Beyond methodological advancement, the study highlights the practical way of combining conventional geoscientific knowledge with modern AI techniques for systematic landslide risk assessment. The outcomes have direct relevance for hazard mitigation and regional planning in the northeastern Himalaya. |
| Keywords | Landslide, Wokha-Mokokchung, Random Forest |