Active-layer thickness estimation from X-band SAR backscatter intensity

The active layer above the permafrost, which seasonally thaws during summer, is an important parameter for monitoring the state of permafrost. Its thickness is typically measured locally, but a range of methods which utilize information from satellite data exist. Mostly, the normalized difference ve...

Full description

Bibliographic Details
Published in:The Cryosphere
Main Authors: B. Widhalm, A. Bartsch, M. Leibman, A. Khomutov
Format: Article in Journal/Newspaper
Language:English
Published: Copernicus Publications 2017
Subjects:
geo
Online Access:https://doi.org/10.5194/tc-11-483-2017
http://www.the-cryosphere.net/11/483/2017/tc-11-483-2017.pdf
https://doaj.org/article/4fba0cbe3c8b4c4da6417ae4bd31e6ec
Description
Summary:The active layer above the permafrost, which seasonally thaws during summer, is an important parameter for monitoring the state of permafrost. Its thickness is typically measured locally, but a range of methods which utilize information from satellite data exist. Mostly, the normalized difference vegetation index (NDVI) obtained from optical satellite data is used as a proxy. The applicability has been demonstrated mostly for shallow depths of active-layer thickness (ALT) below approximately 70 cm. Some permafrost areas including central Yamal are, however, characterized by larger ALT. Surface properties including vegetation structure are also represented by microwave backscatter intensity. So far, the potential of such data for estimating ALT has not been explored. We therefore investigated the relationship between ALT and X-band synthetic aperture radar (SAR) backscatter of TerraSAR-X (averages for 10 × 10 m window) in order to examine the possibility of delineating ALT with continuous and larger spatial coverage in this area and compare it to the already-established method of using NDVI from Landsat (30 m). Our results show that the mutual dependency of ALT and TerraSAR-X backscatter on land cover types suggests a connection of both parameters. A range of 5 dB can be observed for an ALT range of 100 cm (40–140 cm), and an R2 of 0.66 has been determined over the calibration sites. An increase of ALT with increasing backscatter can be determined. The root mean square error (RMSE) over a comparably heterogeneous validation site with maximum ALT of > 150 cm is 20 cm. Deviations are larger for measurement locations with mixed vegetation types (especially partial coverage by cryptogam crust) with respect to the spatial resolution of the satellite data.