Glacier Retreat Recovered from Landsat Images by Object-Oriented Classification in Central Tianshan, 1977-2013

Abstract:

The Tianshan Mountains are known as the water tower of the arid Central Asia. The mountain glaciers are rapidly retreating under the context of global and regional warming. Many works have been done to analyze the retreating rate of glacier areas and volume against individual glaciers. However, it is challenging to obtain an accurate estimate of the entire glacier extent at the regional scale due to cloud issues and the similar spectral signature of snow and glacier. This study develops an object-oriented classification to map the glacier extent by using landsat images and analyzes the retreating rate of glacier area during the past half century. Our preliminary results show that our objected-oriented classification performs better than the regular classification methods, such as the supervised classification with maximum likelihood, band ratio method, snow index method, and so on. Summer snowfall events on the high elevations mountains lead to greatly over map the glacier extent. After carefully selecting the timing of the landsat images and adjusting the glacier extent overestimated due to snow cover, our mapped glacier extent is much smaller and illustrates more reasonable decadal variations than the world glacier inventory and other common land cover products. The glacier extent on the Central Tianshan Mountains illustrates a constant retreating trend, but its retreating rate is slowing down in the recent years.