Validation of Snow Extent Algorithms
Jiancheng Shi1
For climatic, hydrological, and snow hazard investigations, the snow-covered area is one of impor-
tant parameters. Our investigation will assist monitoring of snow processes over alpine watersheds
scale to regional and global scales by means of remote sensing technique. This paper will mainly
focus on validation of snow-covered area mapping algorithms that have been developed over past
few years for a variety of different sensors. The ground truth data was obtained from a VNIR color
photo that is taken simultaneously as AVIRIS and MAS image data on ER-2. This color photo covers
15 km 15 km and can be digitized with a pixel resolution from 1.5 to 4 m. We can access large
amounts of the ground truth data by using these very high spatial resolution digitized photo images
along with spectral information from AVIRIS and MAS image data. The image data of TM, ASTER,
AVHRR, and MODIS were simulated by taking the considerations of both the spectral response
functions and spatial point spread functions of these sensors. We performed the snow mapping algo-
rithms that have been developed recently for the sensors of TM and AVHRR by Rosenthal, and
MODIS by Hall. We validated the accuracy of the snow mapping algorithm under a variety of view-
ing and illuminations, land cover, atmospheric, and terrain conditions. Depending on the application,
the advantages and disadvantages will be discussed. Through this study, a database will be estab-
lished, thus providing excellent data for development, verification, and evaluation of snow mapping
algorithms for different satellite sensors.
1
ICESS, University of California, Santa Barbara, California 93106, USA
108