Outburst Susceptibility Estimation Of Glacier Lakes In Nielamu, Tibet, Based On Presort Method

Glacier lake outburst disaster, or Glacier lake outburst debris flow(GLODF), is aunique form of disasters in glacier-rich high-altitude areas. With fast emerging and highdebris peak, it is especially short in its duration, and high in its currency amount anddestruction power, and large in influential area. It can be commonly seen in Tibet, China.Nielamu lies in the middle area of the Himalaya. With amount of outburst-danger glacierlakes in the region and large-scale outburst history in record, Nielamu is an emphasizedregion on glacier lake investigation, and disaster precaution and reduction. Upon regionalfiles and site surveys, the former research achievements have been summarized so as toconclude that bucket effect exists in glacier lake outburst problems referring to glacier lakeoutburst mechanism and different outburst modes. Thus glacier lake presort method isdeveloped, and used in the outburst susceptibility estimation of9glacier lakes with outburstdanger in Nielamu Region, Tibet.According to the3outburst modes: billow emerging, moraine dam piping, side kenneleroding, summarized by the group of “Survey of Significant Geo-Catastrophes in theHimalayas”, bucket effect of glacier lake outburst problems is claimed. Suggesting that thelakes’ glacier on the trailing edges and side slopes on the banks are included in BillowEmerging Co-system, whose low stability claims for the possibilities of overflowingbillows; while the piping in moraine dams and the erosion from side kennels result in theoutbursts due to the damage of moraine dam, as the outburst possibilities of those twomodes are controlled by the Moraine Dam System. The two systems are independent toeach other, thus the less stable one represents the outburst susceptibility. Based on the bucket effect of glacier lake outburst, presort method of glacier lake outburstsusceptibility is invented, with the specific procedure:1, establish the initial model with therelevant factors selected;2, calculating the respective weighted factor value sums of the2systems (Billow Emerging Co-system and Moraine Dam System) for both training sampleand evaluation sample, thus the binary choice is proceeded;3, establish the evaluationsub-models and obtain the results from sub-models as the final result.30surveyed glacier lakes from “Survey of Significant Geo-Catastrophes in the Himalayas”with expert evaluation results have been used as the training sample for the presortevaluation of9glacier lakes in Nielamu in which exist outburst possibility. Whenestablishing the initial model, method of rough set has been used, producing a BillowEmerging Co-system sample with training sample of17and evaluation sample of3, as wellas a Moraine Dam System sample with training sample of13and evaluation sample of6.As the result of sub-model evaluation using Artificial Neural Network, the outburstsusceptibility indexes of9glacier lakes are obtained. Among them,5coincide with theformer outburst susceptibility classification. The test suggests the model as well as the other4susceptibility indexes is reliable.