09.11.2014 Views

Low (web) Quality - BALTEX

Low (web) Quality - BALTEX

Low (web) Quality - BALTEX

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

61<br />

Sensitivity studies of model setup in the alpine region using MM5 and<br />

RegCM<br />

Irene Schicker, Imran Nadeem, Herbert Formayer<br />

Institute of Meteorology, University of Natural Resources and Applied Life Sciences, Peter-Jordan-Straße 82, A-1190<br />

Vienna, Austria, irene.schicker@boku.ac.at<br />

1. Introduction<br />

Regional climate modeling in the alpine area is a<br />

challenging task. Currently, our institute is participating in<br />

two projects, CECILIA (http://cecilia-eu.org) and<br />

Reclip:century (Basic Data Set of Regional Climate<br />

Scenarios), to study regional climate change in the Alpine<br />

region using two different meteorological models, MM5<br />

(Grell et al., 1994) and RegCM (Pal et al., 2007). The<br />

present study is a part of these projects and focuses on<br />

selection of domain, nesting versus direct runs, comparison<br />

of different physical parameterization schemes and lateral<br />

boundary conditions. The horizontal resolution of 10 km is<br />

used to capture the effects of the complex topographical and<br />

land use features of the region.<br />

The case study is carried for the year 1999. This year is<br />

particularly interesting because of flooding in Danube<br />

Catchment in May followed by the storm in December.<br />

2. Domain and Model setup – MM5<br />

As the innermost domain should cover both the Alpine ridge<br />

and the eastern parts of Austria and comparability with the<br />

Greater Alpine Region (GAR) (Auer et al., 2007) should<br />

also be given, no changes on the horizontal domain size<br />

have been made. For the outermost domain, four different<br />

grid box setups have been tested. Results of two of them,<br />

domain ML and domain L, are shown here (see Fig. 1). Due<br />

to computational limitations 30 vertical half σ levels have<br />

been used.<br />

Zängl options for alpine modeling (z-diffusion, orographic<br />

shadowing) implemented in MM5V3.7 have been used in all<br />

the test runs. One test run used additional improvements in<br />

the NOAH LSM scheme implemented by G. Zängl and his<br />

group (Mauser and Strasser, 2007). Table 1 gives an<br />

overview of the different setups and physical<br />

parameterizations used.<br />

Table 1. Setups of the different sensitivity runs.<br />

Cumulus schemes used are BM for the outermost domain<br />

and Grell for the innermost domain.<br />

ML1 ML2 L1 L2<br />

PBL ETA/MRF ETA ETA/MRF ETA<br />

radiation RRTM RRTM RRTM RRTM<br />

Land<br />

5 layer<br />

5 layer<br />

NOAH<br />

NOAH<br />

use<br />

soil<br />

soil<br />

cumulus<br />

scheme<br />

BM/Grell BM/Grell BM/Grell BM/Grell<br />

explicit<br />

moisture<br />

Reisner 2 Reisner 2 Reisner 2 Reisner 2<br />

3. Domain and Model setup – RegCM<br />

The innermost domain used by RegCM3 simulations closely<br />

resembles that of MM5. The boundary conditions used for<br />

various simulations were ECMWF Interim Re-Analysis<br />

(ERA-Interim, 0.75° and 1.5° grid spacings, 6-h intervals),<br />

the ECMWF 40 Years Re-Analysis (ERA40, 1° and 2.5°<br />

grid spacings, 6-h interval) and finally the 2.5°, 6-h<br />

NCEP/DOE AMIP-II Reanalysis (Reanalysis-2). Sea<br />

Surface Temperature for the simulated periods was<br />

obtained from a UK Met Office Global Ocean Surface<br />

Temperature (GISST), a set of SST data in monthly 1°<br />

area grids. Table 2 summarizes different domain settings<br />

and Nesting strategies for RegCM3 simulations.<br />

Table 2. Setup of RegCM sensitivity runs.<br />

Reanalysis Resolution Nesting Physics<br />

2 / 3 nests 1 / 2 nest<br />

ERA 40<br />

30 km <br />

10 km<br />

1 way direct<br />

ERA 30 km<br />

Interim 10 km<br />

1 way direct<br />

NCEP/DOE<br />

AMIP-II<br />

90 km <br />

30 km<br />

10 km<br />

1 way<br />

30 km<br />

10 km<br />

Domain setup of the RegCM runs is shown in Figure 1.<br />

Grell<br />

Conv.,PBL<br />

(Holtslag),<br />

BATS1e<br />

Radiation:<br />

NCAR<br />

CCM3<br />

SUBEX<br />

Figure 1. MM5 and RegCM domains. In magenta MM5<br />

domain 2, which has been used in all three domain<br />

settings. In blue RegCM domain 2 and 3 respectively. In<br />

green MM5 domain1 L, in red MM5 domain ML. RegCM<br />

domain 1 is shown in cyan.<br />

4. Results<br />

Very first results of the May 1999 case study simulated<br />

with MM5 show that the general patterns of the<br />

precipitation are very well captured if using grid nudging<br />

options to avoid drifting of the model. As grid nudging is<br />

not advisable when performing climate simulations, also<br />

some runs without grid nudging have been performed.<br />

First results of the May 1999 flooding event of two<br />

different L1 runs, without grid nudging, are shown in<br />

Figures 3 and 4. In Figure 2 the 72 h precipitation sum,<br />

obtained from the gridded observation data set of Frei and<br />

Schär (1998) is shown for comparison.<br />

The RegCM3 simulation driven with ERA40 and ERA-<br />

Interim shows that direct downscaling to 10km produces<br />

better results than Nested Run 30km10km. When<br />

recently released ERA-Interim Reanalysis was used as<br />

lateral and boundary conditions, the simulated

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!