Shortly after fruitful discussions at the MAP Meeting 2002 in Park City (Utah), the reanalysis of the MAP-SOP was launched on the VPP5000 at ECMWF. Using the ECMWF 12-hour 4D-Var global assimilation system (IFS), the MAP reanalysis was completed 80 days later.
Since autumn 1999, when the MAP field experiment took place, there have been substantial modifications in the IFS. The main model changes between the IFS operational in 1999 (Cycle 21R2) and the current version (Cycle 24R3) used for the MAP reanalysis are briefly summarized. During the SOP on 12 Oct 1999, the number of vertical levels increased from 50 to 60, a new orography and associated subgrid orographic fields were introduced, and changes occurred in the cloud and convection schemes. In 2000 other major changes involved an increase of the horizontal resolution to approximately 40 km (T511 spherical-harmonic representation), a revised treatment of the land surface scheme and a new parametrization of the long wave radiation. In the assimilation system, the 6-hour window 4D-Var was extended to 12 hours and the inner-loop resolution increased from T63 to T159. Recently, a new shortwave radiation transfer model and a new bias correction for satellite observations were included in the IFS.
One of the aims of the MAP reanalysis was to use as many extra MAP-SOP observations as possible. Special attention was paid to the European windprofiler data which were hardly used or monitored in 1999. After monitoring the data for the first two weeks of the MAP-SOP it became obvious that data of some profilers had to be excluded to avoid a degradation of the analysis quality. Consequently, data of 4 stations (Dunkeswell, Camborne, Clermont and L’Aquila) of the 16 European windprofilers which reported data during the MAP-SOP were blacklisted. Apart from this a priori exclusion all other extra MAP-SOP observations were available to the assimilation system.
The deliverables of the MAP-SOP reanalysis project are:
• A formatted archive of the additional MAP observations comprising data of European windprofilers, high-resolution radiosondes, surface stations, research aircraft data as well as dropsonde data. These data were formatted in BUFR code and archived in 6- hourly intervals throughout the MAP-SOP.
Figure 1. Time series of daily precipitation averaged over the Po catchment (extending from 7 E to 12 E and 45 N to 46.5 N) in the Southern Alpine Region.
• Analysed and forecast fields are stored in MARS. The basic analysed variables include not only the conventional meteorological wind, temperature and humidity fields, but also model products currently available in the ERA-40 reanalysis. The parameters of the MAP reanalysis are archived with a horizontal resolution of T511 for upper air fields, and a reduced Gaussian Grid with approximately uniform 40km spacing for surface und other grid-point fields. Upper air data were saved at each of the 60 “full” model levels and at 23 pressure levels. Additionally, a subset of upper air parameters is archived on fifteen isentropic surfaces as well as on the PV = +/- 2 surface. Tables of all analysed surface and upper air fields are given in Keil and Cardinali (2002). Global fields of two forecasts have been stored as well, i.e. forecasts initialized at 00 UTC +30 h and at 12 UTC +240 h. In order to give a flavour of the MAP reanalysis, daily precipitation totals (i.e. 24 hour accumulations from 06 UTC onwards) for the Southern Alpine Region, an area comparable with the catchment of the river Po (extending from 7 E to 12 E and 45 N to 46.5 N) are presented here. Figure shows the time series of daily precipitation averaged over the Po catchment area (66’000km2) for the high-resolution precipitation analysis, the o-suite and the MAP reanalysis, respectively. Precipitation observations are taken from high resolution (25km) analyses of Alpine rain-gauge observations embracing roughly 5000 rain-gauges. Peak values are found for IOP2b (20 Sep 1999; day 13 in Fig. ), when 27 individual stations reported rainfall amounts exceeding 200 mm in 48 hours. Generally, timing of the events is well represented in both suites. The operational forecast performs well, capturing the majority of the events recording more than 10mm/day area averaged precipitation. The MAP reanalysis shows an even better agreement with observations, e.g. on days 10 and 40 the spurious rainfall predicted by the operational forecast is omitted. However, the amount of rainfall is slightly overestimated. At the recent meeting of the MSC in Bad Tölz, Germany, it was decided to transfer all BUFRized MAP-SOP observations and a subset of 3 hourly main analysed fields on model levels for the European region to the MAP Data Centre (MDC) in Zurich. Additionally, analyses for the same period (first 30 days of the MAP-SOP) of a control analysis using the same IFS system without all extra MAP-SOP data are being transferred as well. This shall allow detailed investigation of the MAP reanalysis data using, e.g., mesoscale models driven with the data.
Reference
Keil, C. and C. Cardinali, 2002: On the ECMWF Re-analysis of the MAP-SOP. Proc. of 10th Conference on Mountain Meteorology, AMS, 307-310.
