 |
| Titel |
Using different remote sensing data to improve snow cover area
representation |
| VerfasserIn |
Rafael Pimentel, María José Pérez-Palazón, Javier Herrero, María José Polo |
| Konferenz |
EGU General Assembly 2016
|
| Medientyp |
Artikel
|
| Sprache |
en
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 18 (2016) |
| Datensatznummer |
250135359
|
| Publikation (Nr.) |
 EGU/EGU2016-16214.pdf |
|
|
|
|
|
| Zusammenfassung |
| The crucial role of an accurate estimation of the snow cover fraction distribution in mountain
hydrology increases in low and warm latitudes, where water scarcity makes the snowpack a
fundamental resource. In Mediterranean mountain regions, the snow has a seasonal
character; it is not permanent during all year or the whole cold season, which is also
variable year to year because of the highly variable and sometimes extreme climatic
conditions. These characteristics results in a very heterogeneous snow cover distribution,
usually in not easy to access areas that lack ground monitoring systems. Remote
sensing information constitutes one of the best ways to monitor the snow cover
evolution; however, this high variability sometimes conditions the suitability of
the available sources of information needed to best represent the snow processes
involved.
This study proposed the combination of three different remote sensing data sources to
improve the seasonal representation of the snow cover fraction (SCF) distribution in Sierra
Nevada Mountains in southern Spain, a representative example of snow areas in semiarid
regions: 1) terrestrial photography; 2) Landsat imagery (spectral mixture analysis); and 3)
MODIS products (MOD10.A1) to improve the estimation of the snow cover fraction
distribution over mountain areas at different scales. For this, three different study sites were
selected over the study area: 1) a detail-scale piece of area (900 m2) at the snow
monitoring point of Refugio Poqueira in the Guadalfeo River basin (South face of
Sierra Nevada Mountains); 2) a hillslope-scale area (2,5 km2) nearby the latter
but in the North face; and 3) a large-scale area (4585 km2) over the 3479 m.a.s.l.
altitude in Sierra Nevada. The analysis was performed during the hydrological year
2010-2011.
The results show that terrestrial photography, whose spatial and temporal resolution can
be adapted to the process under study, constitutes the best technique to monitor snow
dynamics at both the detail and hillslope scales. At such scale, terrestrial photography is able
to reproduce the complex interaction between microtopography and snow, and also captures
the quick changes in the snowpack evolution during the different melting periods that occur
during the snow season over these areas. Therefore, the occurrence of mixed pixels,
composed by snow and no-snow areas, mainly during such snowmelt periods and in border
areas , which usually results in overestimations of SCF by coarser resolution data sources,
was corrected using terrestrial photography. This overestimation was reduced in both cases
after the correction, with RMSE of 0.08 m2m−2 and 0.25 m2m−2 for Lansdsat
and MODIS, respectively, at the detail scale area, and RMSE of 0.09 m2m−2and
0.18 m2m−2for Lansdsat and MODIS, respectively, at the hillslope scale area. At
the large-scale analysis, as expected, MODIS overestimated the Landsat-obtained
snow cover area with a RMSE of 0.02 m2m−2. At this scale terrestrial photography
is not a feasible option yet due to the impossibility of covering a huge area with
an only image. Nevertheless, the monitoring of pilot areas at this scale using a
ground camera network constitutes a promising step to improve the snow cover
representation by means of data fusion algorithms with these three data sources. |
|
|
|
|
|
|