The variation in evaporative fraction and actual evaporation is examined for three sample days in the HAPEX-Sahel Intensive Observation Period (IOP), including data from all the vegetation types and sites. The trends in evaporative fraction over the IOP are also presented for eight sites. The high rate of evaporation from bare soil in the days following rainfall produces a variability in evaporation which makes differences between sites difficult to interpret on a day-to-day basis, but over the whole IOP it is shown that the millet uses a smaller proportion of the available energy for evaporation than the tiger bush or fallow savannah. The combined effect of differences in the total energy used and its partitioning into evaporation and sensible heat flux is demonstrated from the trends in cumulative total energy use and evaporation at the three southern sites, where it is shown that there is systematically less evaporation from the millet than from the savannah or tiger bush sites.

The PYREX program is a major field study of the dynamical influence of the Pyrenees mountains (on the border between France and Spain) on the atmospheric circulation. In October and November 1990 a large number of experimental means had been deployed in the area, including additional sounding systems, automated stations at high elevation sites, wind profilers, sodars, constant level balloons, and four research airplanes. The main focus was on the quantification of the retardation of the cross-mountain flow by the range, but a number of related mesoscale phenomena had been captured, e.g. the formation of lee waves, lee eddies, sheltering area's, and local surface winds. The data interpretation was supported by an extensive effort in meso-scale numerical modelling, and it is argued that only numerical models, conveniently qualified by observations, can document the momentum budget in a consistent way. Improvements in the field of mountain waves and mountain roughness representation in the large-scale atmospheric models are expected from the PYREX results, as well as improvements in the performance of mesoscale models for the operational forecasts. The present paper presents a broad introduction to the experiment. It reviews the experimental set-up and the available data sets. Finally, two more specific topics are discussed in depth, as examples of the available data and modelling strategy: the lee wave event of 15 October 1990 is presented with both experimental and model results. Some of the findings concerning the Tramontana wind are also discussed.

Data impact studies are presented for the French global model ARPEGE. The resolution of the model is T538 in spectral triangular truncation. It has a stretched coordinate with a stretching factor of 2.4, the maximum resolution becoming 2.4 times better over the pole of interest (Western Europe), and 2.4 times worse at the antipodes (South Pacific area). The maximum resolution is 15km at best over Western Europe. In the vertical, the model has 60 levels in hybrid coordinate, from the surface to 0.1hPa. The analysis is a four-dimensional variational analysis, with two minimisation loops at T107 and T224 (unstretched). The code is developed in collaboration with ECMWF. The analysis uses all types of observations: conventional observations, satellite winds, satellite radiances and GPS data (both ground-based and radio-occultation). These observations are also used in the limited-area model ALADIN, coupled with ARPEGE, together with extra observations as described in Fischer (same volume). In the past few years, several data impact studies were performed. In particular, we started assimilating Atmospheric Motion vectors (AMV) in BUFR format, using the Quality Indicator (QI) to choose the most accurate data; MODIS polar winds from the Aqua and Terra satellites; SSM/I radiances from F13 and 14; AIRS radiances (54 channels currently); Ground-based GPS data over Europe (in the context of the E-GVAP programme); scatterometer winds from QuikSCAT, ERS and ASCAT; ATOVS from MetOp; MSG Clear-Sky Radiances (CSR); GPS radio-occultation data from COSMIC, CHAMP and GRACE-A; IASI radiances (50 channels). Impact studies also dealt with improvement in our assimilation of data: increase in the AIREP density , Variational Bias Correction (VarBC) for radiances, new emissivity parametrisation over land for micro-wave data. A few of these improvements are presented in this paper, namely the assimilation of GPS radio-occultation, of IASI radiances, the impact of Variational Bias Correction for radiances and of an emissivity parametrisation for micro-wave radiances over land. The impact of data over specific areas is also investigated, usually linked with field experiments. We describe first results obtained over Africa in the context of AMMA and give information in the preparation for a future field experiment over Antarctica (Concordiasi).