![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Evaluation of snow dynamics modelling on a pixel scale using terrestrial
photography and Ensemble Transform Kalman Filtering |
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 |
250135657
|
Publikation (Nr.) |
EGU/EGU2016-16549.pdf |
|
|
|
Zusammenfassung |
Snow plays a crucial role in the hydrological regime in mountainous catchments, which
increases in semiarid regions, where the recurrence of drought period makes it necessary to
accurate the determination of the water availability from the snowpack. Physically based
approaches constitute one of the best ways to reproduce the snow dynamics over these highly
variable conditions. Moreover, they allow further understanding the processes involved, the
snowpack behaviour and evolution. However, in some cases the complexity of the modelled
process and the non-availability of all the required data for such models, avoid a correct
representation of certain aspects. In these cases, data assimilation techniques can help to
improve model performance and may also act as an indirect tool to understand the
represented processes.
This work assesses snow dynamics on a pixel scale (30x30m) in a Mediterranean site
(Sierra Nevada Mountain, southern Spain) combining physical snow modelling (WiMMed, a
physically based hydrological model developed for Mediterranean environments), ground
sensing information (terrestrial photography) and data assimilation techniques (Ensemble
Transform Kalman Filter), throughout a study period of two hydrological years: 2009-2010
and 2010-2011. Snow cover fraction and averaged snow depth were obtained from
the terrestrial photography images and used as observations in the assimilation
scheme. The model performance was evaluated using different combinations of the
variables assimilated: 1) only snow cover fraction, 2) only snow depth, 3) and both
variables.
The results show how the assimilation enhances the model performance. This
improvement is higher if the variable assimilated is snow depth, with RMSE= 0.14
m2m−2and RMSE=12.16 mm for snow cover and snow depth respectively. However, this
enhancement varies throughout the study period. During short snowmelt cycles, for example,
the assimilation of the snow cover fraction is the most efficient.
Nevertheless, certain periods can still be identified for which the assimilation does not
reproduce adequately the snow dynamics, and and further representation of the snow
processes in the model must be included to achieve such cases. Data assimilation not only
improves the model results but also helps to understand the limitations of the modelling. |
|
|
|
|
|