Exploring the benefits of higher resolution with RCA3.5

One benefit of regional climate models is that they offer increased resolution for simulating climate features of importance to a given region, at a relatively low computational cost. In particular, higher resolution provides more accurate surface forcing of regional climate processes, such as associated with topography, land-water contrasts and heterogeneous surface cover.

Forcing of this type is often important in determining local details in important climate variables such as precipitation, near-surface temperatures and wind fields. Furthermore, high model resolution also allows basic atmospheric processes to be more accurately represented, generally leading to more realistic, more intense, higher-order weather systems that, when time-integrated, make up the regional climate. In other words, higher model resolution generally improves the simulation of extreme events, such as wind storms and precipitation events.

Improved precipitation with high-resolution RCA3.5

To investigate potential benefits of increased model resolution we have run the RCA3.5 model for the period 1989-2008 forced by ERA-interim reanalysis boundary conditions, over a common pan-European domain, at resolutions of; 50km, 25km, 12.5km and 6.25km, respectively. A similar exercise has also been performed for a pan-Africa domain, with resolutions of 100km, 50km, 25km and 12.5km.

Here we present a few preliminary results, a more in-depth analysis is currently underway. Figure 1 shows the winter mean precipitation for 1989-2008, as simulated by the 4 RCA resolutions and represented by the Ensembles precipitation observations, mapped to two resolutions, 50km and 25km.

High-res winter mean precipitation
Figure 1. DJF winter mean precipitation 1989-2008, from 2 versions of the Ensembles-Observation data set (Haylock et al. 2008) at 50km and 25km (N.B. This data set is uncorrected for wind and evaporation effects). Also shown are 4 realizations of the same field from RCA3.5, run at 50, 25, 12.5 and 6.25km, all forced by ERA-interim boundary data on the same model domain. The figure shows the model domain minus the 12 outer grid points.
Enlarge Image

While there are no major differences in the large-scale patterns simulated by the 4 model versions, details in the spatial distribution of precipitation do appear to be improved with increasing resolution, note in particular, the higher intensity precipitation along the west coasts of the British Isles, Norway and north-east Spain. In figure 2, we show the same field, winter season mean precipitation, but zoom in on the Scandinavia region.

High-res winter mean precipitation Scandinavia
Figure 2. As in figure 1, but zoomed in over Scandinavia.
Enlarge Image

As resolution is increased there is a clear improvement in the placement of the model precipitation, with a simulated maximum of more than 350 mm/month on the west coast of Norway, as well as the appearance of west to east gradients in precipitation over Denmark and the west coast of Sweden.

At lower model resolution, these precipitation maxima are markedly reduced in magnitude and placed too far inland. To make a more in-depth analysis of the high-resolution models, likely requires higher-resolution observations, such as the 4km precipitation data set over Sweden and similar resolution observations over the Alpine region and British Isles. Such an analysis, concentrating on the intensity distribution of simulated precipitation, will begin soon.

High-resolution model runs over Africa

Figure 3 presents a similar assessment of the sensitivity of simulated precipitation to model resolution, concentrating on precipitation variability over Africa during the boreal summer.

Over the region from Chad to the Gulf of Guinea, precipitation in this season mainly arises from organized convection, embedded within synoptic-scale wave systems. These waves, generally form over the Darfur mountains and propagate westward within the latitude band 5°-15°N (Mekonnen et al. 2006). The typical wavelength and propagation speed of such African Easterly Waves (AEWs) results in a wave system passing over a given region roughly every 2-6 days. Hence, the intensity of precipitation associated with these waves can be well captured by time-filtering precipitation to highlight variability in the time window 2-6 days.

Results from such an exercise, using 3-hourly precipitation from RCA3.5 simulations and observed rainfall from the Tropical Rainfall Measurement Mission (TRMM), are shown in figure 3, where we plot the standard deviation of precipitation in the 2-6 day time-window. The standard deviation of precipitation is seen to progressively increase as the model resolution increases from 100km to 12.5km, with the 25km simulation most closely matching the TRMM observations, which are also available at 0.25° resolution.

Increased variability can be understood as a wider intensity distribution and hence more intense simulated precipitation at higher resolution. While difficult to evaluate due to lack of sufficiently high-resolution observations, we note the continued increase in precipitation variability as model resolution is increased beyond that of the TRMM data. Initial analysis also indicates the amplitude of the model diurnal cycle of precipitation also increases (improving) as resolution is increased. Further analysis to understand the causes of these improvements is underway.

precipitation standard deviation for the JAS summer season Africa
Figure 3. 2-6 day band-passed precipitation standard deviation for the JAS summer season, averaged for the period 1998-2008. Note all model simulations used the same domain, with the outer 12 points removed from the figures. Results for the 12km RCA3.5 model are highlighted only for the one latitude band of interest. The time-filtering of precipitation was not performed for other latitudes.
Enlarge Image

References

Haylock M.R., N.Hofstra, A.M.G. Klein Tank, E.J.Klok, P.D.Jones and M.New 2008. A European daily high-resolution gridded data set of surface temperature and precipitation for 1950-2006. J. Geophysical Res. 113, D20119, 12 PP., 2008 doi:10.1029/2008JD010201

Mekonnen, Ademe, Chris D. Thorncroft, Anantha R. Aiyyer, 2006: Analysis of Convection and Its Association with African Easterly Waves. J. Climate, 19, 5405–5421.
doi: 10.1175/JCLI3920.1