The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Publications Copernicus
Articles | Volume XLII-5
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-5, 787–789, 2018
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-5, 787–789, 2018

  19 Nov 2018

19 Nov 2018


P. H. Pandit1, U. Khot2, S. D. Jawak3,4, and A. J. Luis4 P. H. Pandit et al.
  • 1Department of Natural Resources, TERI University, New Delhi - 110 070, India
  • 2Symbiosis Institute of Geoinformatics, Symbiosis International University, Pune, Maharashtra, 411005, India
  • 3Svalbard Integrated Arctic Earth Observing System (SIOS), SIOS Knowledge Centre, University Centre in Svalbard (UNIS), P.O. Box 156, N-9171, Longyearbyen, Svalbard, Norway
  • 4Earth System Science Organization-National Centre for Antarctic and Ocean Research, Ministry of Earth Sciences, Government of India, Headland Sada, Vasco-da-Gama, Goa 403804, India

Keywords: Glacier Velocity, Image Matching, Antarctic

Abstract. Glaciers all over the world are experiencing changes at varying stages due to changing climatic conditions. Minuscule changes in the glaciers in Antarctica can thus have major implications. The velocity of glaciers is important in several aspects of glaciology. A glacier’s movement is caused by different factors such as gravity, internal deformation present in the ice, pressure caused by accumulation of snow, basal sliding etc. The velocity of a glacier is an important factor governing mass balance and the stability of the glacier. A glacier which moves fast generally brings more ice towards the terminus than a slow moving glacier. Thus, the glacier velocity can determine its load carrying capacity and gives indication on the ‘health’ of the glacier. Measurement of the ice flow velocity can help model glacier dynamics and thus provide increasing insights on different glacier subtleties. However, field measurements of velocity are limited in spatial and temporal domains because these operations are manual, tedious and logistically expensive. Remote sensing is a tool to monitor and generate such data without the need for physical expeditions. This study uses optical satellite imagery to understand the mechanisms involved in the movement of a glacier. Optical image correlation method (COSI-Corr module) is chosen here as the promising method to derive displacement of a moving glacier. The principle involved in this technique is that two images acquired at different times are correlated to find the shift in the position of moving ice, which is then treated as displacement in the time interval. Employing this technique we estimated the velocity of Mellor glacier (73°30′S, 66°30′E), a tributary glacier of the Amery Ice Shelf, Antarctica, over a span of four years from 2014 to 2017. Correlation is performed using Landsat-8 panchromatic images of 15 m resolution. Optical images from Landsat 8 often have noise due to atmospheric conditions such as cloud cover, so we used only those images cloud with cloud cover less than 10%. The glacier is covered in 128 path frame and 112 by Landsat-8. The correlation frequency was calculated using the correlator engine. Window size taken here is 256 and step sizes is 64 for both x and y dimensions. Once the correlation is calculated for an image pair for a specific time-period, we obtain three different outputs. Two of them indicated displacement (one in x direction and another in y direction) and the remaining output provided signal to noise ratio. The band math tool using displacement outputs in ENVI software performed velocity calculations. This gives us a raster image showing velocity at each point or pixel. Some errors such as noise persist and their correction is performed in ArcGIS software. In order to get pure signals, we removed all the signals with a signal to noise ratio less than 0.9 and this was carried out using raster calculator tool. All the resultant velocity rasters were interpolated and bias was calculated between seasons of two consecutive years. Two maps were generated for each year, one for early summer i.e. from January to April and one from September to December using the resultant velocity raster. The mean values of velocities found for Mellor glacier from Jan-April 2014, 2015, 2016 and 2017 were 374.06 ma−1, 413.59 ma−1, 278.62 ma−1 and 406.66 ma−1, respectively. Velocities for September-December 2014, 2015, 2016 and 2017 were found to be 334.63 ma−1, 334.43 ma−1, 367.37 ma−1 and 381.31 ma−1, respectively. The biases are computed for all the seasons of four years and root mean square (RMSE) values are estimated. These RMSE values signify the season-wise variations in the velocities. RMSE values for season of Jan–April 2014–15, 2015–16 and 2016–17 were 75.92 ma−1, 147.82 ma−1, and 133.33 ma−1, respectively. Similarly, RMSE values for season of September-December 2014–15, 2015–16 and 2016–17 are 35.7 ma−1, 51.29 ma−1 and 35.84 ma−1 respectively. The results showed variations in velocities for different seasons. We plan to integrate this data with the discharge rates, to estimate mass balance and melting rates of the glacier to decipher mechanisms at work for the Mellor glacier.