About KNMI Colloquia and workshops Staff Publications Contact Sitemap Homepage Research KODAC: KNMI Operational Data Centre
Background on 3D case set up

Below we give a short summary of the methods used to derive the initial profiles and forcings for the 3D RICO precipitating cumulus case.

Large scale forcings

Unlike the BOMEX case, for which accurate large scale forcings were obtained from an extensive set of observations, we were unable to deduce reasonable forcings from RICO observations. As an alternative, we used a RACMO HindCast to get estimates of the large scale forcings. The RACMO (Regional Atmospheric Climate Model) HindCast is a high-resolution version of the ECMWF model initialized each 24 hours with the ECMWF analysis at 12 UTC. The simulation was performed for a small domain, consisting of 90 x 92 gridpoints with a resolution of 20km, in which the RICO research area (61.46W, 17.97N) is contained.
A total period of 2 months (December 2004 and January 2005) was simulated, and output was generated every 10 minutes on a 5 x 5 grid centered around the RICO research area. The output includes, among others, large scale tendencies due to advection (a combined vertical and horizontal advection), vertical profiles of horizontal winds, temperature, moisture and vertical velocity and the precipitation at the surface. A time serie of the latter shows that the relative amount of precipitation produced in RACMO coincides reasonably well with the observations by the SPolKa radar (shown in Introduction), which gives us certain confidence in the RACMO results for the undisturbed RICO period we are considering.

The large scale forcings we specify for the 3D models include:

  • The subsidence rate

  • The large scale temperature tendency due to horizontal advection

  • The large scale moisture tendency due to horizontal advection

  • The net radiative temperature tendency

We performed a budget analysis and used the RACMO tendencies to construct vertical profiles of these large scale forcings. By making a spatial and temporal average, we obtained good results for the vertical profiles of subsidence and advection. In constructing the forcings we aimed for simple profiles while staying close to the averaged RACMO profiles. Additionally the large scale forcings are balacing each other at the upper part of the (LES) domain while the vertically integrated total budgets for heat and moisture should be as closed as possible. The results from RACMO for the subsidence velocity and advection are shown in Figure 2 as mean profiles (black line) +/- half a standard deviation. In red the forcings as prescribed for this LES case are shown. The subsidence profile shows a downward velocity that is increasing with height, which will give us the expected warming and drying in the trade wind layer. The horizontal advection gives on average a cooling throughout the whole trade wind layer, and a drying in the lower layers and a moistening in the upper layers.

Figure 2. RACMO vertical profiles of (from left to right): (a) Subsidence velocity (m/s), (b) Temperature tendency due to advection (K/d) and (c) Moisture tendency due to advection ((g/kg)/d). In black the mean profile is shown, +/- half the standard deviation. In red the prescribed LES profile is shown.

The one remaining forcing, the net radiative tendency, was obtained separately by using two offline radiation schemes initialized with the above described profiles of temperature and humidity. By comparing and averaging the results over twenty-four hours, we obtained a profile that prescribes a cooling rate of 2 K/day close to the surface, which is slightly decreasing to about 1 K/day in the free atmosphere.

Figure 3. Calculated and prescribed net radiative tendency.

Combining all these forcings, we tried to close the budgets for heat and moisture. We considered an additional term for precipitation of 10 W/m2 as a sink for moisture and source for heat. This value is based on an average precipitation rate of 0.34 mm/day during the composite period obtained from the SPol radar observations. Estimates of tendencies due to turbulent fluxes were derived from LES. The budget analysis led to an almost closed budget for temperature, with radiative cooling and subsidence warming as counteracting forcings in the upper layers, and with advection terms that are in line with the thermal wind estimates. For the moisture budget we reduced the horizontal advective drying in the lowest 1.5 km slightly. Please see section 3.4 for definitions of the exact profiles we prescribe.

Initial thermodynamic profiles

For the initial profiles of potential temperature, specific humidity and the horizontal winds, we used the dropsondes measurements performed by the NCAR C130 aircraft (on all available flight days (6) within the period of 04/12/16-05/01/08), and radiosondes, launched every 6 or 12 hours from Spanish Point (Barbuda) during that same period. By using the average soundings during the period, we constructed initial profiles for LES. The average soundings are shown in Figure 3 below, along with the LES profiles.

Figure 4. Mean profiles of potential temperature (K), specific humidity (g/kg) and the zonal and meridional wind components (m/s) of all radiosondes released from Spanish Point in the period 04/12/16-05/01/08, here shown in black +/- a standard deviation. The dotted black line indicates the mean profile of saturation specific humidity (g/kg) during this period. The constructed initial profiles for LES are shown in red.

Parameterization of surface fluxes

During RICO, the research vessel Seward Johnson performed measurements from which surface sensible heat and latent heat fluxes are derived. Comparison with the surface fluxes obtained from RACMO showed comparable values. In the proposed simulation the surface fluxes are not fixed but parametrized. Due to the lack of ship measurements in the composite period we specify a mean sea surface temperature (SST) obtained from ECMWF analysis during the three week period.