SNOW GLACIER HAZARD MAPPING AND MODELLING USING REMOTE SENSING DATA

Abstract

This study is being divided into six sub-objectives. First objective is for Glacier terminus position monitoring and modelling using remote sensing data. This study basically deals with the identification of glacier terminus position. The data used was of Landsat TM. MSS, ETM+ and OLI/TMRS sensors data. This study first converts the Landsat DN Images into Reflectance Images using formula from Landsat Handbook. After this the study computes NDVI and NDWI for the identification of glacier terminus. This Terminus position is identified for a period from 1963-2013. Toposheets and Landsat Images were used. Glacier Terminus is considered as the starting point of the river from it. The water pixels appear bright in NDWI and dark in NDVI. Further the Altitude of snout position was marked using car and also the retreat of glacier from 1963 onwards. Also the retreat rate for each year was being determined. Then the study performs the linear regression model for the altitude position of glacier terminus using cartosat DEM and the retreat of glacier terminus from 1963 onward was computed. Further the retreat rate was computed. The linear regression model gave a good value of coefficient of determination. Also a comparison was made on the prediction value of 2013, altitude position from the Artificial Neural Network and linear regression model. And it was observed that the ANN predicted with a better accuracy. Further analysis was made on the future affect of such a glacier retreat on cold water habitat, effect of it on irrigation of crop field and on the fisheries The second objective is Glacier Lake hazard modelling and mapping using remote sensing data. The objective of this study is to carry out the regression modelling of the glacier lake, which could be used to predict the formation of Glacial Lake Outburst Flood in future. The software used for this study were ERDAS Imagine 2013 and Arc GIS 9.3. Firstly we calculate the NDWI (Normalized difference water Index), which enables us to distinguish all the water bodies from other features. All the water pixels appear bright in this image. Hence all the lakes were easily identified. Study area is Alaknanda basin. Then we calculate the snow line altitude using contour of the area by Arc GIS 9.3 software. For this computation, SRTM DEM data was used. The glacier lakes were identified through this process. Next we calculate the area of glacier lake from year 1990 to year 2012, using Landsat Thematic mapper (TM) and Enhanced Thematic Mapper (ETM+) images. Further we perform the regression modelling on the observed data sets. It was observed that the pattern followed a linear path and the model predicted the formation GLOF (Glacial lake outburst flood) in the future and might affect the irrigation and cold water habitant. Also Artificial Neural Network was designed for which the input parameters used were slope value, aspect value and elevation value. ANN was trained for a period between 1990 and 2010. ANN and regression model predicted the area of glacial lake for the year 2012. Then error for both the predicted values was computed against the observed area of glacial lake for year 2012. ANN was found to be more accurate. * Third objective deals with the comparison of classification accuracies. The purpose here is to compare the classification accuracies of glacier change detection by using different techniques. Firstly, the study used sub-pixel based classification algorithm. For this, ERDAS Imagine 2013 software was used. It consists of the following sub-steps; (1) Pre-processing, (2) Environmental Corrections, (3) Manual Signature derivation and (4) Classification. Secondly, this study calculated indices from Landsat images and then supervised classification technique was applied. Thirdly, it classified the snow and non-snow area using object based algorithm. Software used for the object based classification is ecognition Professional 4.0. It was observed that the shadow effect is not removed in sub-pixel based classification, so its accuracy is less as compared to Object based classification. But the indices based method removed the shadow effect. Further on applying object based approach resulted in the best classification results. Fourth objective is to analyse the Impact of Snow Avalanche on Vegetation Area Using Remote Sensing Data. This study basically deals with the affect of avalanche on the vegetation. For this study the data used is Landsat TM (Thematic Mapper), OLI (Operational Land Imager) and TIRS (Thermal Infrared scanner) sensor data. The collected data is of summer months with acquisition dates 16 September 2009 and 11 September 2013. Cloud free data were collected .The study area is Alaknanda basin. For this study five avalanche paths were detected, shed 1, shed 2, shed 3, shed 4 and shed 5. Both NDVI (Normalized Difference Vegetation Index) and Tasselled Cap Transformation greenness method was applied in order to examine the amount of change in the vegetation due to occurrence of avalanche. The area was also classified into vegetation using the supervised based classification method, to determine the amount of change in the vegetation area. For computing the Tasselled Cap Transformation greenness, Raster calculator in ARC G1S 10.2 Software was used. Supervised based classification was done using ERDAS IMAGINE 2013 software. ITA Fifth objective is identification of th,. potential snow avalanche areas in Alaknanda basin, for this analysis the following parameter were computed like, elevation, slope, aspect, curvature and snow covered area. Further several constraints were applied on the parameters which holds the governing conditions for the avalanche to occur. For this implementation Matlab software was being used. Also the above parameters were computed on the ARC GIS software. Sixth objective is to monitor the glacier in Alaknanda basin using remote sensing data. This study basically monitors the great Himalayas between the year 1998 to 2013 using satellite data. The Landsat satellite data wera used to monitor variations in the area of glacier. Also in order to find the glacier outlfnes, techniques such as band ratio, NDSII and visual Interpretation were used. Further the snow covered Area (SCA) of the part of Alaknanda basin was computed both for the winter season and for the summer season. The analysis for the same was done between 1998 and 2013. It was observed that the amount of decrease in the snow covered area was more in winter season compared to summer season, which also shows the rate of retreat of glacier. This study also classifies the snow into two categories (I) Dry snow and (2) Wet snow. The pattern in the change in the area of these two categories were analysed both for the winter season and summer season.