09.11.2014 Views

Low (web) Quality - BALTEX

Low (web) Quality - BALTEX

Low (web) Quality - BALTEX

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

99<br />

Effects of numerical methods on high resolution modelling<br />

Holger Göttel<br />

Max Planck Institute for Meteorology, Bundesstr. 53, 20146 Hamburg, Germany, holger.goettel@zmaw.de<br />

1. Introduction<br />

One of the remaining challenges in climate predictions as<br />

well as in weather forecasts is the correct modelling of<br />

heavy precipitation. The reason is the high temporal and<br />

spatial variability of cloud formation and precipitation.<br />

Models with a resolution of several kilometres (50 to 10 km<br />

in regional climate models) are not able to resolve small<br />

scale features like convective clouds explicitly. To account<br />

for these features they use convective parameterisations<br />

which are normally only based on few observations. It is<br />

unclear that these parameterisations are still valid under<br />

climate change conditions and for all current convection<br />

types due to the limits of observations.<br />

One solution is to increase the horizontal resolution to few<br />

kilometres where the model is able to resolve convective<br />

clouds explicitly. The increased computer power in the last<br />

decade allows an increase of resolution from 50 to 10<br />

kilometres. A further increase of resolution, which might be<br />

possible in the near future, has consequences for some<br />

simplifications in state of the art climate models. These<br />

approximations (e.g. the hydrostatic balance and the neglect<br />

of advection of precipitation) are not realistic at these scales.<br />

With the regional climate model REMO – a hydrostatic<br />

model which was extended with an evolutionary approach<br />

(Janjic et al., 2001) to a non-hydrostatic model – the limits<br />

of the Tiedtke scheme for convective systems in cold-airoutbreaks<br />

and the ability of the non-hydrostatic model to<br />

simulate such extreme events are investigated.<br />

2. REMO<br />

The regional climate model REMO (Jacob, 2001; Jacob et<br />

al., 2001; Jacob and Podzun, 1997) is a three-dimensional,<br />

hydrostatic atmospheric circulation model which solves the<br />

discretised primitive equations of atmospheric motion. Like<br />

most other RCMs, REMO has been developed starting from<br />

an existing numerical weather prediction (NWP) model: the<br />

Europa-Modell (EM) of the German Weather Service DWD<br />

(Majewski, 1991). Additionally, the physical<br />

parameterisation package of the general circulation model<br />

ECHAM4 (Roeckner et al., 1996) has been implemented,<br />

optionally replacing the original EM physics. In numerous<br />

studies, the latter combination (i.e., the EM dynamical core<br />

plus the ECHAM4 physical parameterisation scheme)<br />

proved its ability to realistically reproduce regional climatic<br />

features and is therefore used as the standard setup in recent<br />

applications, including the present study.<br />

This hydrostatic model was extended in this study to a nonhydrostatic<br />

model using the evolutionary approach proposed<br />

by Janjic et al. (2001). Additionally, a diagnostic advection<br />

scheme for precipitation is implemented which can be used<br />

online and offline (Göttel, 2009).<br />

3. Results<br />

Figure 1. AVHRR channel 5 retrieval from 17<br />

February 1997<br />

Figure 2. HOAPS estimated precipitation fluxes<br />

[mm/h] valid for 9:00 UTC 17 February 1997<br />

The AVHHR (Figure 1. ) show clouds in the postfrontal<br />

area due to a cold air outbreak over the open ocean. The<br />

cold air outbreak induces convection with intense<br />

precipitation. The precipitation estimates of HOAPS<br />

(Figure 2. ) reaches in the postfrontal area precipitation<br />

similar values like in the cold front of the low “Caroline”.<br />

This precipitation which is validated by ship observations<br />

doesn’t appear in the ECMWF model as well as in the<br />

regional model REMO (Figure 3. ). The missing<br />

precipitation in the models is caused by deficits in the<br />

cloud convection scheme. A re-calculation of this event<br />

with the non-hydrostatic version of REMO shows, that<br />

this model is able to reproduce postfrontal precipitation<br />

(Figure 4. ).

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

Saved successfully!

Ooh no, something went wrong!