| || Knowledge Based System for Forecasting Snow Avalanches of Chowkibal Tangdhar Axis (J&K)
Author : Pant, L.M. ;Naresh, Praveen
Source : Defence Science Journal ; Vol:49(5) ; 1999 ; pp 381-391
Subject : 551 Geology;551.578.48 Avalanche
Keywords : Expert system ;Avalanche forecasting;Artificial intelligence;Knowledge based model
Abstract : "In this paper an attempt has been made to use artificial intelligence in avalanche forecasting and to develop a rule-based, expert system for predicting direct action avalanches of Chowkibal-Tangdhar axis (J&K). Using C-language integrated production system (CLIPS), procedural knowledge is represented in the form of rules. The condition attributes of the rule-based system are 28 variables selected from 1154 samples of snow met and snow profile data. The relative contribution of each variable on avalanche days and non-avalanche days and their influence on sitewise release of avalanche was studied to formulate 358 rules. These rules, which include 173 decision rules, were finally implemented and validated for running the model. Sixty-three samples of snow-met data and pit profile data attributing to avalanche days and 54 samples of non-avalanche days were run on the model. The results show that the knowledge-based model can predict avalanche days with 76 per cent efficiency. The misclassified results accounted for 28.2 per cent of 117 test samples. "