Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
215<br />
Assessing the performance of seasonal snow in the NARCCAP RCMs – An<br />
analyses for the Upper Colorado River Basin<br />
Nadine Salzmann (1,2,3) and Linda Mearns (3)<br />
(1) University of Zurich, Department of Geography, Switzerland, nadine.salzmann@geo.uzh.ch<br />
(2) University of Fribourg, Department of Geosciences, Switzerland<br />
(3) National Center for Atmospheric Research, Institute for the Study of Society and Environment (NCAR/ISSE), Boulder,<br />
CO, USA<br />
1. Introduction<br />
Seasonal snow is a major component of the hydrological<br />
cycle and thus of the climate system Cohen and Rind (1991).<br />
The dynamics of the snow cover depends on the atmospheric<br />
circulation, which in turn is thermodynamically influenced<br />
by the snow pack.<br />
In mountain regions, the dynamics of the seasonal snow<br />
cover is a key factor for hydrological run off and consequent<br />
availability of water for human consumption, irrigation and<br />
power generation e.g. Barnett et al. (2005). Due to the<br />
proximity to the melting point, snow is particular sensitive<br />
to changes in the atmospheric conditions with significant<br />
effects on the timing and magnitude of hydrological run off.<br />
Among the most promising tools for assessing future<br />
changes in the hydrological run off in mountain regions is<br />
Regional Climate Models (RCMs). They are able to simulate<br />
and provide climate input variables on a regional scale and<br />
for heterogeneous landscapes such as mountain topography<br />
Leung and Qian (2003). However, so far only a few studies<br />
analyzed the performance of mountain snow in RCMs.<br />
There are many reasons for that, including foremost the<br />
general difficulties associated with process-modeling in<br />
complex topography and the lack of ample snow<br />
observations in most mountain regions worldwide for<br />
validation purposes.<br />
Despite these limitations, here, we aim at assessing the<br />
performance of the seasonal snow in the RCM simulations<br />
calculated for the international RCM program NARCCAP<br />
(North American Climate Change Program,<br />
www.narccap.ucar.edu) Mearns et al. (2005).<br />
2. Study site and data<br />
The following analyses are conducted for the region of the<br />
Upper Colorado River Basin (UCRB) in the U.S. Rocky<br />
Mountains. Here, the seasonal snow pack of the highest<br />
peaks contributes about 70 % of the Colorado River’s total<br />
annual run off Christensen et al. (2004). Within the<br />
perimeter of the UCRB, 45 SNOTEL observation stations<br />
exist, most of them measuring Snow Water Equivalent<br />
(SWE), 2m Temperature (2mT) and precipitation (Precip),<br />
since 1981 and before. In addition to these point<br />
measurements, a second source of ‘observation’ is used in<br />
this study; the gridded NARR data (North American<br />
Regional Re-analyses), with a horizontal resolution of 32<br />
km. SNOTEL and NARR are compared in the following<br />
with results from NARCCAP simulations. Only NCEPdriven<br />
NARCCAP runs are used in a first step. The<br />
comparison includes the following RCMs, listed with the<br />
respective Landsurface Scheme (LSS) used in each RCM. It<br />
is in the LSS, where the processes concerning the snow<br />
dynamics are defined.<br />
ECPC (Experimental Climate Prediction Center) from<br />
Scripps Institution of Oceanography, La Jolla, CA, USA.<br />
LSS: NOAH<br />
MRCC (Modèle Régional Canadien du Climat) from<br />
Ouranos Consortium, Montreal (Quebec), Canada.<br />
LSS: CLASS<br />
RegCM3 (REGional Climate Model) from UC Santa<br />
Cruz, ITCP, USA.<br />
LSS: BATS<br />
WRF (Weather Research and Forecasting Model) from<br />
Pacific Northwest National Lab, Washington, USA.<br />
LSS: NOAH<br />
MM5 (Mesoscale Meteorologic Model) from Iowa State<br />
University, Iowa, USA.<br />
LSS: NOAH<br />
All RCMs are run with a grid spacing of 50 km and cover<br />
the UCRB by 10 grid boxes.<br />
3. Analyses and some results<br />
‘SWE’ and the two related variables ‘2mT’ and ‘Precip’<br />
were compared and analyzed in various spatial and<br />
temporal terms.<br />
It was found that NARCCAP RCMs generally simulate<br />
higher air temperatures than NARR and particularly than<br />
SNOTEL. However, the departures are relatively constant<br />
within different elevation levels, indicating that<br />
theoretically a lapse rate correction could be applied for<br />
adjusting the NARCCAP results towards the SNOTEL<br />
measurements. Though, based on the relatively small<br />
elevation difference between the SNOTEL and<br />
NARCCAP topography, the lapse-rate factor only explains<br />
a minor part of the deviation.<br />
Regarding precipitation the NARCCAP models are in<br />
general too dry compared to SNOTEL.<br />
A large-scale factor that influences the climate and<br />
particularly precipitation on the Colorado Plateau is the<br />
ENSO. Major El Niño events as e.g. in 1982-83, 1992-93<br />
and 1994-95 produce during winter exceptionally wet<br />
weather in the region. It is thus of interest, if these events<br />
are reflected in the datasets. However, no clear evidence<br />
is found in any of the datasets used here, pointing towards<br />
ENSO events.<br />
Finally, SWE, shows significant departures between the<br />
different RCMs, and also compared to SNOTEL and<br />
NARR. While the annual cycle is captured relatively well,<br />
the amount of snow as well as the duration and the time of<br />
maximum vertical snow depth are significantly<br />
underestimated by the NARCCAP RCMs and NARR