A step to the local use of orography at higher resolution:
Introduction of the local orography data

(more details mehdi.elabed++at++meteo.ma or neva.pristov++at++rzs-hm.si)
The complete version is on delage: ~mrpe691/report/c923/report_e923.ps

Running Aladin model with high resolution (few kilometers) needs a good description of the climate parameters at high resolution.

The orography data set used now for creating the operational climate files is a global, 2'30" orography data prepared (by E. Legrand) from a global, 30" orography data (GLOBE or GTOPO30). This resolution is sufficient for running Aladin model at grid distance greater than approximately 5 km. But with finer resolution we need better description of the orography.

The question is how to include finer orography data which usually cover only a limited area and sometimes does not cover the whole domain of the model integration. The aim of this study is to try to improve the representation of model orography using a local, finer data set, which is not necessary covering the domain of model integration and is not a complete data set.

Possibilities to include local orography data

On the basis of the discussions several possibilities arose:

a) Preparing an Aladin Orography file
File with the orography data can be included in the step one of the e923 configuration using the key LNORO. In this case a new orography and a new land sea mask (if asked) are used. It is read from an Aladin file (name is neworog) and after checking the frame, it undergoes the biperiodisation and spectral fit if grid point orography is read. The advantage is the standard format of files (FA) but it is supposed that the fine orography covers the whole area (however, there is a way to avoid this).

b) Inside the code
This would be done like it is done now for the computation of the fields using vegetation data. This possibility seems to be complicated to implement since the local files have different format (not FA files). After several discussions we concluded that this possibility is improper since the e923 configuration is already quite complicated, having several steps.

c) Outside the code
In this case first a new orography data set with higher resolution would be created. From the local fine data and from the operational data an intermediate, new orography data set can be created. This possibility does not imply any changes to the Aladin code, concerning the format of the files and it can be adapted easily in each country. This would also give unified definition and methods when creating new data sets between Aladin consortium partners.

Finally we decided to create a new orography data set which can be used at the first step of the configuration e923. This is a safe method and can be adapted without any changes in the ALADIN model code.

Creation of the orography data set

The basic data for creating new set are local orography data in high resolution. Since only sub-grid mean orography height is usually available and than some information about topography variability, about slopes, and about number of peaks are required the resolution of orography data set can not be the same as the local orography data. We also have to consider which data should be used in the areas of no high resolution data.

Orography data set

Files needed for the first step of the configuration e923 can be named orography data set. The following files are included:
Oro_Mean Mean orography Hmean (mean of H) m
Oro_Max Maximum orography Hmax (max of H) m
Oro_Min Minimum orography Hmin(min of H)  m
Sigma Sub-grid standard deviation of Hmean m
Nb_Peaks  Number of sub-grid peaks 
Hmax-HxH-Hmin_ov4  mean of (Hmax-Hmean)(Hmean-Hmin)/4  m2 
Dh_over_Dx_Dh_over_Dy  mean of dHmean/dx*dHmean/dy  m2/km2 
Dh_over_Dx_square  mean of (dHmean/dx)**2  m2/km2 
Dh_over_Dy_square  mean of (dHmean/dy)**2  m2/km2 
Water_Percentage  Land/Sea mask  % of water 
Urbanisation  Fraction of urbanisation  % of city 
Oro_Max and Oro_Min are not needed directly for e923, they are used to get data in Hmax-HxH-Hmin_ov4.

Methodology

At the beginning three kinds of grids should be defined:

The idea is first to interpolate the available data (orography,...) from a local to a target grid. Over the areas with no data the values are set to missing value. Then also the fields from the basic grid are transferred onto the target grid. Finally the missing values from the first step are replaced with the values from the second one.

Interpolation from the local to the final grid
For each point of the final grid a box equal to the mesh of the target grid is defined. The value in the target grid point is computed from the values in points of local grid inside this box (see figure 1). In boxes where there is no local value a missing value is taken.


Figure 1: One example of a possible local (empty circles) and final grids (black circles) with the final grid box.

- Mean height of orography in the target grid point is the average of the values of the local grid points which are inside the box.
- Sigma is the value of standard deviation of the orography height inside the box.
- Derivatives are calculated in the local points and then averaged over the points in the box.
- Inside every box of the final grid the number of peaks is calculated according to points of the local grid.
- For calculation of Hmax-HxH-Hmin_ov4 three types of data are needed for each local grid point: the orography height (H), the minimum (Hmin) and the maximum (Hmax) sub-grid heights for the point. If all data are present than for each point of local grid value of (Hmax-H)*(H-Hmin)/4 is calculated and the value in the point of target grid is the average over the local points. But usually only the height is available and in this case the values are set to missing value.
- For calculation of Water_Percentage Hmax and Hmin are also very useful. It can be supposed that there is water if Hmin and Hmax are equal to zero. If those data are not available, values in local grid should be evaluated in different way, maybe from coast lines. If local data for proportion of water are available then they are averaged inside the boxes, otherwise they are put to missing value.
- If local data for Urbanisation are available then they are averaged inside the boxes, otherwise they are put to missing value.
Interpolation from the basic grid to the final grid
Interpolation from the basic data grid to the final grid is done by taking the value of the nearest point of the basic grid to the point of the final grid. This is done for all fields.

Merging the data into final grid
In this step all values of the final grid are filled, which is a merge between the two results obtained from interpolation from the basic grid to the final one and from the local to the final grid. If there is a missing value the values issued from the basic grid are selected, otherwise the value issued from the local data is taken.

Example

In Slovenia, data for orography height in 100 m resolution are available, but this is the case only inside the borders (the Western part of Slovenia can be seen on figure 2). We combine these data with the basic orography in resolution of 2'30" (see figure 3). For presentation of results one smaller domain located around the west border of Slovenia is chosen. The target, final resolution is 4 times better than the basic one. The result of merging the two orographies can be seen on the figure 4. In the area where local data are available these data are used (more detailed topography on the figure 4), while in the other parts the data from basic grid are used.
For selected domain there are 48x48 points in the basic grid, and 192x192 points in the target one. So the value of one point in basic grid is first prescribed to 16 points in the target grid. The value of the point in the target grid where local data are available is calculated as an average over the points inside the target box. When local grid points cover whole target box there are 96 local grid points inside the box.

Conclusions

To use the local orography data for the creation of ALADIN climatological files, a procedure was developed outside the ALADIN code. The orography data set needed for the first step of configuration e923 can be prepared in similar way as it is now but using also available local data. More about the program (how to use it, where it can be found) is written in the complete version.
There are still some tests to be done regarding creation of ALADIN climatological file at the basis of this data orography set. There are some problems to be expected. Slopes are much steeper when resolution of data is higher, so values in derivative fields can be very different. The number of sub-grid peaks depends on the amount of points inside the final box.
 


Figure 2: The orography of the Western part of Slovenia in the local grid (resolution 100 m, 950x1240 points).


Figure 3: The orography for selected domain in the basic grid (resolution 2'30", 48x48 points). The domain includes the Western part of Slovenia, the Northern part of Adriatic Sea, the Istrian peninsula the Eastern part of Italy, and a part of Southern Austria.


Figure 4: The orography for selected domain in the final grid (resolution 37.5", 192x192 points).


Home