Analysis of short-term trends in snow cover variability in the European Alps - based on operational sub-pixel snow mapping with NOAA AVHRR data

In alpine regions such as the
European Alps, snow is a predominant environmental factor. High
accurate snow monitoring in the Alpine Region is of great
importance as temporal and spatial variations in snow coverage
have far-reaching consequences on the natural and the
socio-economic systems. It is required for various purposes such
as meteorological modelling, climate studies, snow mapping,
estimation of stored water equivalent or snowmelt runoff
prediction. In contrast to conventional in situ snow observations, remote
sensing data regularly provide spatial snow cover information which can
be used for climate induced studies on snow cover variability.
The main objectives of this study are
to assess the accuracy of chronological sequences derived from
fractional snow cover maps as well as to
detect and analyze temporal and spatial variability patterns
within the Alpine Region based on different statistical applications.

NOAA AVHRR has been employed for over 20 years and
consequently offers a unique data archive for long-term studies.
An additional advantage is the high temporal resolution of NOAA
AVHRR, whereas its medium spatial resolution (1.1 km at nadir)
means a challenge in rugged terrain. The used data set includes daily scenes from
the platforms NOAA-16 (2001 - 2002) and NOAA-17 (2002 - 2007).
The pre-processing includes calibration, georeferencing, atmospheric
correction and orthorectification. For mapping
snow, the widely used linear spectral unmixture algorithm is
implemented to estimate snow cover at subpixel scale. Principal
component analysis, including the reflective part of AVHRR channel
3, is used to quantify fractions of 'snow' and 'no snow' within a pixel.
Substantially, this algorithm improves the possibility to detect
differences concerning snow distribution over complex topography
for operational and near-realtime applications.

Time series of 7 years (2001 - 2007) are used to derive spatial
and temporal snow cover dynamics. The input data is
evaluated implicating difficulties in merging the data sets from
different sensors. Statistical performance
within the developed framework over the NOAA AVHRR data suggests
an accurate caption of snow related variables. High spatial
variability at lower elevation zones and the most persistent snow
cover located at high elevations in the central Alps are found.
The duration for snow persistence varies in different elevation
ranges and generally becomes longer with increases in the terrain
elevation. A slight decreasing trend in overall snow cover area is
found during the investigation period.
Additionally, the resulting snow cover curve is compared to ground
based temperature data and to particular meteorological events for
further verification. Temperature is found to explain about 90\%
of the snow cover extent. These high correlations support the
assumption of a reliable accuracy of the snow cover derived from
fractional snow cover maps and can be implemented in the
estimation of the dimensions of future scenarios in a warmer
climate.

The results presented in this study are based on a short-term analysis and
give an overview of possible methods to assess climatic change
impact on snow dynamics in the recent years. Further studies aim
to homogenize data derived from different sensors in order to
compile a longer time series. The understanding of past, ongoing and future snow cover dynamics under the presumptions of the anticipated climatic change is
crucial and can provide potential information to support the
adaptation process of i.e. tourism industry, natural hazard management and water power industry.