Page History

This exercise builds on Exercise 1, but uses observation data in BUFR format instead of analysis fields.

Retrieving the observation data

Create a new Mars Retrieval icon and rename it to obs. Edit it and set the following parameters in order to retrieve BUFR observation data from 3 days ago:

Retrieve the data and examine it. Metview's BUFR Examiner displays the contents of the BUFR data (Figure 8). Each message contains many measurements. If you visualise the data, you will see a standard display of synoptic observations. Figure 9 shows this, using the shaded_coastlines icon from the solutions folder (this plot has also been zoomed to show a smaller area).

Extracting the 2 metre temperature

Create a new Observation Filter icon and rename it to filter_obs_t2m. With this icon we will extract just the 2m temperature into Metview's custom ASCII format for scattered geographical data - geopoints. Set these parameters:

Note that 012004 is the code for 'Dry bulb temperature at 2m'. If you examine this icon now, you will see the result: a table of geo-located temperature values. When you visualise the data, you will see that the actual values are plotted as text on the screen; we can do better than this. From the solutions folder, drop the coloured_markers icon into the Display Window. The shaded_coastlines icon may also help make the points easier to see (Figure 10).

Retrieving the forecast data

Create a new Mars Retrieval icon, rename it to t2m_forecast, and set these parameters :in order to retrieve the 48-hour forecast made 5 days ago for 2-metre temperature.

Computing the forecast-observation difference

This is just the same as in Exercise 1, using a Simple Formula icon; create a new one and rename it to fc_obs_diff. Drop t2m_forecast into the Parameter 1 box, and filter_obs_t2m into the Parameter 2 box. Visualise the result - you will see that the result of a field minus a scattered geopoints data set is another geopoints data set. For each geopoint location, the interpolated value from the field was extracted before performing the computation. From the solutions folder, drop both the diff_symb_hot and the diff_symb_cold icons together into the plot in order to get a more graphical representation of the result.

Overlaying data in the same plot

To plot the forecast field together with the observation differences, do the following. Visualise t2m_forecast and drop the shade icon into the plot. Now drop fc_obs_diff into the plot, followed by (or with) diff_symb_hot and diff_symb_cold. The observation differences don't stand out well against the strongly coloured field, so drop shade_light into the plot to obtain something like Figure 11.