Atmosphere/surface interactions in the ECMWF model at high latitudes
The surface boundary condition is an essential aspect of an atmospheric model as it controls the surface fluxes of momentum, heat and moisture into the atmosphere. The ocean represents a relatively simple boundary condition with a slowly varying sea surface temperature which can be kept constant to a reasonable degree of approximation in short and medium range forecasts. Although there are uncertainties in air/sea transfer over the ocean, current formulations are probably accurate within 20% (see e.g. Beljaars 1997), which is very good compared to the situation over land.
Land surfaces and sea ice are much more complex because they are highly interactive and they respond to the atmospheric and solar forcing at very short time scales e.g. resulting in a strong diurnal cycle. Furthermore, typical land surfaces tend to be inhomogeneous, which has strong impact on momentum, heat and moisture transfer. This paper focuses on the thermal aspects of the atmosphere/surface coupling over land, snow and sea ice with emphasis on high latitudes. The schemes that control the surface/atmosphere interaction in the ECMWF model have evolved enormously over the last 15 years. The land surface scheme has gradually been changed from a 2-layer model with a climatological deep soil boundary condition to a fully prognostic 4-layer model with 6 surface tiles and data assimilation for soil moisture and temperature (Viterbo and Beljaars 1995; Van den Hurk et al. 2000; Douville et al. 2000). The sea ice model was changed from a single slab model to a 4-layer sea ice temperature model and separate tiles for sea ice and open water.
In this paper a few model changes are presented which were particularly relevant to high latitude atmosphere/surface coupling. Emphasis is put on the thermal coupling including sea ice and snow. The purpose is to illustrate the relevance of the various physical aspects. The starting point is the 4-layer land surface model introduced by Viterbo and Beljaars (1995) in August 1993. It was developed and tested in single column mode using various data sets. The scheme had rather good summer performance, but fully coupled to the atmosphere it turned out that the amplitude of the diurnal cycle of temperature was too large and that the winter temperatures over continental areas were too low. These problems could be reduced by introducing the process of soil moisture freezing and by re-tuning the stable boundary layer formulation. These two aspects of the thermal coupling between atmosphere and surface will be discussed in section 2. The BOREAS experiment was a major effort to study various aspects of the atmosphere / land surface interaction in boreal forest. One of the immediate messages from the field experiment was that snow impact on albedo was far less in forest areas than in open terrain. It resulted in a preliminary fix for snow albedo in forest areas with very positive impact on temperature biases in spring. Results will be shown in section 3.
The snow albedo, the handling of partial snow cover and terrain heterogeneity could be handled in a more consistent way with a tiled version of the land surface scheme called TESSEL. It was particularly beneficial in areas with snow, because the selected tile structure explicitly distinguishes between exposed snow and snow under high vegetation. These aspects are discussed in section 3. The tile structure also allowed for partial ice cover over the ocean and a more responsive sea ice temperature model, which is discussed in section 4.
The resulting model was used for the 45 year long re-analysis (known as ERA-40; Uppala et al. 2005) covering the period from 1957 to 2002. Section 5 discusses systematic data assimilation increments of soil moisture, soil temperature and snow water equivalent. The observations are mainly from SYNOP messages which provide indirect information on soil quantities through the model. Study of increments is interesting, because it provides information on model deficiencies. Further evaluation of the realism of the atmosphere to land surface coupling is given in section 6 making use of the BERMS data.
Beljaars, A., G. Balsamo, A. Betts and P. Viterbo (2006): Atmosphere/surface interactions in the ECMWF model at high latitudes, ECMWF Seminar on Polar Meteorology, 4-8 September 2006. http://www.ecmwf.int/publications/library/do/references/list/200609