| Mesan - an Operational Mesoscale Analysis
System SMHI has access to many different types of observations such as manual observations (called synop or metar), automatic station data, satellite and radar imagery. The best estimate of a meteorological parameter is given by combining all available observations of that variable in an analysis. The analysis is made on a grid where every value represents the mean for a grid square. In that process the quality and the representativety of each observation is taken into account. That means that an observation at a large distance from the square will have less influence on the value than an observation close to it. This is what is done in Mesan for different parameters such as
In figure 1 we can see an example of a Mesan product used by forecasters, and in figure 2 a product specially designed for aviation purposes.
Figure 1. Green colours indicate precipitation, turquoise snow, grey cloud cover, red temperature and yellow visibility. Black lines are isobars.
Figure 2. An analysis of significant cloud base in different coloured shadings, brown for cloud base below 300 metres, green 300-3000 and blue above 3000 metres. White areas have no significant cloud base. Dashed lines are isolines for visibility. Methods The analysis method used in Mesan is optimal interpolation. It takes into account the amount of information in each observation. How the information varies with e.g. distance is model by so-called structure functions. In a temperature analysis, observations over land contribute more to the knowledge of temperature of land than over sea. In figure 3 we can see the effect on the analysis, when this is taken into account via a structure function.
|
| Figure 3. The impact on a temperature analysis
when structure functions depend not only on horizontal distance (as in left panel) but
also on difference in fraction of land/water and in altitude (right). It makes it possible
to describe local variations in mountainous and coastal areas. Precipitation analysis is performed using a variable first guess. It makes it possible to increase spatial resolution in data-sparse areas. The description of the error varies as a function of the prevailing weather situation. It is based on a statistically established relation between wind, orography and variations in friction and latitude. With this relation approximately 50 percent of the observed climatological variance can be explained.
Figure 4. First guess error reflecting precipitation climate.
|
  |
| Figure 5. The effect of excluding (left) and including (right) the variable first guess error (figure 4) in a precipitation analysis using two fictive observations. It means that observations from a climatologicaly dry area will not lead to an underestimation of precipitation in rainy areas. |
| Updated 1999-04-19 |
