Anshuman Bhardwaj

Applicability of Landsat 8 data for characterizing glacier facies and supraglacial debris

Abstract The present work evaluates the applicability of operational land imager (OLI) and thermal infrared sensor (TIRS) on-board Landsat 8 satellite. We demonstrate an algorithm for automated mapping of glacier facies and supraglacial debris using data collected in blue, near infrared (NIR), short wave infrared (SWIR) and thermal infrared (TIR) bands. The reflectance properties in visible and NIR regions of OLI for various glacier facies are in contrast with those in SWIR region. Based on the premise that different surface types (snow, ice and debris) of a glacier should show distinct thermal regimes, the ‘at-satellite brightness temperature’ obtained using TIRS was used as a base layer for developing the algorithm. This base layer was enhanced and modified using contrasting reflectance properties of OLI bands. In addition to facies and debris cover characterization, another interesting outcome of this algorithm was extraction of crevasses on the glacier surface which were distinctly visible in output and classified images. The validity of this algorithm was checked using field data along a transect of the glacier acquired during the satellite pass over the study area. With slight scene-dependent threshold adjustments, this work can be replicated for mapping glacier facies and supraglacial debris in any alpine valley glacier.

Modelling the hypsometric seasonal snow cover using meteorological parameters

This study established a decadal correlation between meteorological observations (temperature and snowfall) and satellite-derived seasonal snow cover for a glacier catchment. The study area was classified into 10 elevation zones. The time period for considering climatic variables was from the start of the significant fresh snowfall of the new season to the date of satellite image acquisition. The snowfall inputs from the five meteorological stations at different altitudes were interpolated for the entire catchment using a discretised thin-plate spline technique. A local temperature lapse rate for this specific time period was calculated. It was applied throughout the catchment for interpolating the temperature, which was further used to refine the interpolated snowfall. Such a hypsometric approach along with third-order polynomial curve fitting (R2 = 0.998) finally gave an equation for estimating percent snow-covered area for different elevation zones with a good accuracy and very low average RMSE (Root Mean Square Error) of 3.16 percent.

Demarcation of potential avalanche sites using remote sensing and ground observations: a case study of Gangotri glacier

This study demonstrates the effectiveness of remote sensing and analytical hierarchy process for avalanche hazard mapping. The layers incorporated in this study were of slope, aspect, profile curvature, ground cover and elevation. The accuracy of output was determined using the registered avalanche sites based on ground observations and field-based modelling techniques. 93.35% of avalanche-affected areas came under maximum and moderate hazard zones, thus proving the effectiveness of this technique for Gangotri glacier basin. A parallel study was done to observe the change in the results, if any, by using high-resolution DEM and Cartosat-1 imagery. Similar methodology was adopted and the outcome was having significant improvement over the previous result as 98.8% of the preregistered avalanche area falling within maximum and moderate hazard zones.

Automated detection and temporal monitoring of crevasses using remote sensing and their implications for glacier dynamics

Abstract Detailed studies on temporal changes of crevasses and their linkage with glacier dynamics are scarce in the Himalayan context. Observations of temporally changing surficial crevasse patterns and their orientations are suggestive of the processes that determine seasonal glacier flow characteristics. In the present study, on a Himalayan valley glacier, changing crevasse patterns and orientations were detected and mapped on Landsat 8 images in an automated procedure using the ratio of Thermal Infrared Sensor (TIRS) band 10 to Optical Land Imager (OLI) shortwave infrared (SWIR) band 6. The ratio was capable of mapping even crevasses falling under mountain shadows. Differential GPS observations suggested an average error of 3.65% and root-mean-square error of 6.32m in crevasse lengths. A year-round observation of these crevasses, coupled with field-based surface velocity measurements, provided some interesting interpretations of seasonal glacier dynamics.

Remote sensing of mountain snow using active microwave sensors: a review

This work provides an overview of various methods for estimating snow cover and properties in high mountains using remote sensing techniques involving microwaves. Satellite-based remote sensing with its characteristics such as synoptic view, repetitive coverage and uniformity over large areas has great potential for identifying the temporal snow cover. Many sensors have been used in the past with various algorithms and accuracies for this purpose. These methods have been improving with the use of Synthetic Aperture Radar sensors, working in different microwave frequencies, polarisation and acquisition modes. The limitations, advantages and drawbacks are illustrated while error sources and strategies on how to ease their impacts are also reviewed. An extensive list of references, with an emphasis on studies since 1990s, allows the reader to delve into specific topics.

Effect of sensor modelling methods on computation of 3-D coordinates from Cartosat-1 stereo data

The orbital and the rational polynomial coefficients (RPC) models are the two most commonly used models to compute a three-dimensional coordinates from an image stereo-pair. But it is still confusing that with the identical user provided inputs, which one of these two models provides more accurate digital elevation model (DEM), especially for mountainous terrain. This study aimed to find out the answer by evaluating the impact of used models on the vertical accuracy of DEM extracted from Cartosat-1 stereo data. We used high-accuracy photogrammetric DEM as the reference DEM. Apart from general variations in statistics, surprisingly in a few instances, both the DEMs provided contrasting results, thus proving the significance of this study. The computed root mean square errors and linear error at 90% (LE90) were lower in case of RPC DEM for various classes of slope, aspect and land cover, thus suggesting its better relative accuracy.

Changing climate and glacio-hydrology: A case study of Shaune Garang basin, Himachal Pradesh

The rise in temperature is already evident in Himalaya with rate of increase varying seasonally and spatially. Changes in precipitation are also evident with no clear trend. Several studies in different parts of Himalayas suggest that the glaciers are retreating in general with few exceptions as response to changes in temperature and precipitation. The stream flow in river basins in Indian Himalayan region (IHR) is already showing changes in studies undertaken in the last few decades. Use of glacio-hydrological models gives opportunity to estimate stream flow in glaciated river basins and understand the changes. The present study deals with estimation of discharge in Shaune Garang Basin, Himachal Pradesh using a glacio-hydrological model based on degree day factors. The model was used to estimate long term average of melt season discharge (1985-2007) in the basin. The modelled discharge shows good correlation with measured discharge for simulation period except for first year of comparison.

Monitoring the Status of Siachen Glacier Using Multi Temporal Remote Sensing Approach

The temporal monitoring of any glacier is important for observing the effects of changing climate. This study reports the decadal changes in Siachen glacier. Analysis was carried out on decadal basis by processing and analyzing Landsat images from 1978 to 2013. Images were co-registered within Root Mean Square Error (RMSE) limit of 0.5 pixel. An object based classification approach was adopted to perform temporal semi-automated areal change detection. The glacier inventory of 1978 showed around 74976 ha of glacier area which further decreased by around 1302 ha in 2013 with a shift of 1.5 km in the snout position.