Curriculum Vitae
GIS Research Projects
Mapping a glacier in the Peruvian Andes through time using remotely sensed and survey data

Casual observations throughout recent decades have indicated that many montane glaciers are receding. These non-polar glaciers not only influence global sea levels and climatic conditions, but also provide important fresh water resources to human populations. To date, the behaviour of alpine glaciers has been studied by scientists carrying out field observations. Most commonly, networks of stakes are established and surveyed through time to map glacier movement. However, field trips are costly and dangerous to undertake. Satellite remote sensing offers the availability of images with resolution sufficient enough to discern moving ice margins. Coupled with GIS technologies, satellite remote sensing can offer scientists the ability to monitor glaciers at a fraction of the cost of direct field observation. Published research on the use of remotely sensed images and GIS for monitoring montane glaciers has relied on the vector model for visually comparing an ice margin from one time period to an unclassified image from another. This methodology allows only visual analysis of ice margin change. The methodology used here is map algebra. Using the raster model, mathematical operators are applied to reclassified images to calculate differences in ice extent in different years.

A SPOT multispectral image (from 1988) and a Landsat Thematic Mapper image (from 1985) are used here to assess the map algebra methodology for calculating change in the ice margin of the Quelccaya Ice Cap, Peru.

Quelccaya image
Quelccaya: SPOT image
The images were reclassified using ERDAS Imagine software and imported to Arc/INFO. Here, the map algebra methodology within the Grid module was used to find the difference in ice margin between the 1985 and 1988 images. It was found that the reclassification of the SPOT image was hindered by the sensor's limited spectral resolution. Reclassification of the SPOT image often confused moraines with debris-covered ice, overestimating the extent of the ice in valley areas in 1988. The ice margin was found to have receded more where the terminus was located on flat plateaux. The map algebra methodology was found to be an efficient way of calculating the difference in ice extent between the two time periods. This methodology is also capable of expansion by including other satellite images as they become available. However, this study also demonstrated the reliance of map algebra on accurate reclassification of satellite images. As such, the accuracy of the ice margin retreat measured here is better on plateaux than in valleys where the SPOT sensor had difficulty in distinguishing moraines and debris-covered ice.
Audience Link
Research scientists