Using wind fields from a high resolution atmospheric model for simulating snow dynamics in mountainous terrain

It is widely known that the snow cover has a major influence on the
hydrology of Alpine watersheds. Snow acts as temporal storage for
precipitation during the winter season. The stored water is later
released as snowmelt and represents an important component of water
supply for the downstream population of large mountain-foreland
river systems worldwide. Modelling the amount and position of the
snow water stored in the headwater catchments helps to quantify the
available water resources and to estimate the timing of their
release. The presented work investigates wind induced snow
transport processes which are considered to be crucial for the snow
distribution in Alpine catchments. In contradiction to the
importance that is attributed to this process, there are only a few
studies available which have quantified the transport intensities
on the catchment scale. This can be attributed to the fact that the
even today not much is known about the spatial characteristics of
wind fields which are the driving force for snow transport
processes. The presented thesis tries to overcome this lack of
information by using physically based wind fields predicted by an
atmospheric model (PSU_NCAR MM5 model) for the modelling of the
snow cover (simulated by SnowModel). All of the used models are
described in great detail in the literature, validated in many
different regions, and can be seen as applicable with regard to the
goal of this work. As snow transport processes are particularly
important on a comparatively small scale a numerical inclusion of
the responsible processes into regional models is inadequate.
Hence, while this study itself mainly uses smaller scale physically
based models, a parameterisation scheme is presented at the end of
this thesis that is able to incorporate its main findings into
larger scale models. All of the presented work was carried out at
the Berchtesgaden National Park. The site is highly appropriate
because of the extremely rough terrain and the good accessibility.
Furthermore, the instrumentation of the area is comparatively good
and the data sources (GIS, field campaign data) are excellent. The
thesis deals with the winter seasons (August - July) 2003/2004 and
2004/2005. For this period, data of 5 meteorological stations, 1
field campaign and two Landsat ETM+ images were available. As
mentioned before, physically based wind fields were used as input
for the snow transport modelling. An operational coupling between
atmospheric model and snow transport model was not pursued because
of the high computational costs of the atmospheric model. Thus, a
library of representative wind fields was produced in advance and
linked to the snow transport model via operational German weather
service Lokalmodell results. This becomes possible because of the
comparability of a MM5 model layer with one of the Lokalmodell
model layers. To link the wind field library to the snow model all
of the predicted MM5 wind fields were characterised by information
available from the Lokalmodell. This enable an easy detection of
the MM5 wind field which is closest to the real climatic wind
conditions at any Lokalmodell time step (1 hour). The produced MM5
wind fields have a spatial resolution of 200 meters. As an initial
check if the snow cover simulation of SnowModel in association with
the wind field library delivers adequate results with respect to
the snow distribution, model runs were first carried out at the
200m scale. An analysis of the results showed that the coupled
routine delivers acceptable results. It could be seen that with the
use of the MM5 wind fields, the snow cover becomes more anisotropic
and that transport processes over crests as well as sublimation
processes are predicted to become more intensive. Nevertheless, a
higher resolution was needed to quatify the effects and to validate
the results. In a subsequent step the MM5 wind fields were
downscaled to a 30m resolution. The downscaling procedure lead to a
better agreement between modelled and measured wind speeds. The
resulting 30m wind fields were used for high resolution model runs
which were validated on the basis of the field campaign and
remotely sensed data. A comparison with model runs using wind
fields interpolated from station data showed that the runs
performed with the MM5 wind fields deliver more consistent and
comprehensible results. Subsequently, the validity of the model is
discussed on the basis of selected results. High resolution model
results indicated that snow transport processes are effective at
high elevations but virtually negligible for regions below of 1800m
a.s.l.. Furthermore, it could be seen that the correct estimation
of snow transport from the surrounding areas to glaciers becomes
possible by using the MM5 wind fields. Very high modelled
sublimation rates at the mountains crests are discusses with
respect to their importance on the water balance. Furthermore, the
influence of preferential snow deposition and snow slides which
were not numerically predicted in this work were discusses.
Additionally, the applicability of atmospheric model results as
input for land-surface models could be confirmed. In a final step a
model scheme is presented that would make the generated information
available for regional scale models. This model parameterization
scheme which is based on the modelled 30m snow water equivalent
distribution within the test area was used for this area. The
scheme allows for a quick and simple description of the subscale
snow heterogeneity in regional scale models. This can lead to
considerable model improvements with respect to the description of
the energy and moisture fluxes to and from the surface. An accurate
description of these fluxes is essential for an accurate simulation
of the melt period and, therefore, for an acceptable calculation of
the runoff generation in larger scale models.