 |
| Titel |
Developing robust spatial interpolation techniques for temperature and precipitation in a data-sparse alpine catchment |
| VerfasserIn |
Andreas Jobst, Daniel Kingston, Nicolas Cullen |
| Konferenz |
EGU General Assembly 2015
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 17 (2015) |
| Datensatznummer |
250101461
|
| Publikation (Nr.) |
 EGU/EGU2015-608.pdf |
|
|
|
|
|
| Zusammenfassung |
| Providing adequate input data are available, distributed and physically-based hydrological
models should constitute the most detailed and realistic possible representation of catchment
hydrology. However, the combination of sparse monitoring networks and the high
spatio-temporal variability of climate in alpine environments makes such models challenging
to implement. Here, a fully distributed hydrological model (WaSIM) is implemented for the
Clutha river, New Zealand, at a spatial resolution of 1 km2. The Clutha catchment
(21680 km2) is the largest in New Zealand and is situated in the lower half of the
South Island, extending eastwards from the Southern Alps. The interaction of the
predominant westerly winds with the steep orography of the Southern Alps leads to a large
precipitation gradient decreasing sharply from annual totals above 10 m near the
main divide to less than 0.5 m inland. In the upper catchment, large amounts of
precipitation are stored as seasonal snow, which significantly influences the annual
discharge regime. As such, a correct spatial representation of precipitation totals and
high elevation temperature is fundamental to the realistic simulation of river flow.
However, there are no long term precipitation sites in the headwaters, and only two
(relatively short) high elevation temperature records. Furthermore, the majority of
long-term temperature records are located in inter-montane valleys that are prone to
strong winter lapse rate inversions. Consequently, standard interpolation techniques
or fixed lapse rates do not provide suitably realistic temperature or precipitation
fields that are fundamental to accurately simulate the spatial variation in catchment
hydrology.
In order to overcome these issues of data availability, a variety of geostatistical
techniques have been investigated as the basis for generating realistic climate fields. The
development of the precipitation field was based on a trivariate spline and a 30-year rainfall
normal surface. Development of the temperature field was more complex, in part
owing to different lapse rates and the range of physical processes controlling daily
maximum and minimum temperature (Tmax and Tmin, respectively). Based on short
term high elevation records Tmax lapse rates were found to vary between coastal
and inland sites. A linear regression model based on distributed monthly relative
humidity grids was used to approximate the spatial variation of lapse rates across the
catchment. Spatial variation in Tmin is shown to be captured by a simple inversion
model combining variable monthly inversion strengths in the lower atmosphere with
negative lapse rates above the inversion layer. Validation of the Tmax and Tmin fields
with independent weather stations demonstrates that these techniques provide a
substantial improvement over previous spatial interpolations of temperature for
the upper Clutha. Finally, the newly generated spatial fields were used as input
to the hydrological model, with discharge from multiple points across the upper
catchment used to assess the resultant model skill in terms of simulating daily and
monthly discharge. In a last step, the modelled seasonal snow field was validated,
both spatially and temporally, using remotely sensed (MODIS) snow cover data. |
|
|
|
|
|
|