, 2011a and Woźniak et al., 2011b. The results of this work together with the operational system’s configuration are presented in this paper. Data assimilation is an analysis that combines time-distributed observations and a dynamic model. This kind of analysis gives much better results than simpler methods like the spatial interpolation of observations. According to the way in which the updating is
done in time, data assimilation can be divided into variational and sequential data assimilation. In the first approach, past observations Ganetespib until the present time are used simultaneously to correct the initial conditions of the model. In sequential assimilation, observed data are used as soon as they appear in order to correct the model state. There are many different methods of introducing the observed data into the model, from the Cressman scheme, through Optimal Interpolation, 3-D and 4-D variational methods, to different modifications
of the Kalman Filter. As the 3D CEMBS operational system uses the Cressman scheme, other methods will not be presented in greater detail in this paper. The Cressman method is a simple and computationally fast assimilation scheme, which makes it a good choice for a data assimilation system used to create forecasts in operational mode. It is also very accurate in comparison to Fluorouracil price its low complexity. Its main disadvantage is that it may produce unrealistic extrema in the grid values
near the edges of the spatial domain. It can be also unstable if the model grid density is higher than the observation grid density. However, in the case of satellite data this is not an issue, as the spatial resolution of the satellite data used is higher than the model grid resolution. The Cressman method Amino acid comes down to few simple steps that are performed as follows. Firstly, the background state xb is set equal to the previous forecast performed by the model. Then the satellite data used for the assimilation are stored in the matrix denoted by y. Data suspected of being invalid because of clouds, the presence of ice or any other reason are masked out. The result of the analysis xa is then calculated according to the following equation: xa(j)=xb(j)+∑i=1nw(i,j)y(i)−xb(i)∑i=1nw(i,j)+E2,where i and j represent the satellite and model data grid-points respectively, and di,j is the distance between points i and j. The main parameters of the Cressman method that need to be chosen are the influence radius R and the shape of the weight function w, which determine how the satellite data influence the model. One of the disadvantages of this method is that the influence radius has to be determined by trial and error; this makes parameterization of this method laborious. After many trials with different sets of the parameters, the one that gave the best results was chosen. The radius R of the influence was set to 20 grid-points.