Home

This wiki describes how to use Wxgen. Wxgen creates weather time-series that are:

1. gridded
2. arbitrarily long
3. with multiple weather variables.

The program offers many different options for creating these time-series and allows a comparison of various combinations of options.

The generator uses a database of large scale weather sequences from the ECMWF model. The generator can create gridded output on the same scale as the database (0.25° by 0.25°). Below is an example of 10 simulations of 1000 day length for temperature and precipitation for a single point (Oslo): images/timeseries3.gif

Download test database

To use Wxgen, you will need to download the ECMWF-based database. Contact Thomas Nipen to get a hold of this. For more information about this dataset, check this wiki page.

This database has the following variables: 2m temperature (K), precipitation (mm)

1. sea level pressure (Pa)
2. cloud area fraction (0-1)
3. zonal and meridional wind speed (m/s). These are all daily averages, except precipitation, which is a daily total.

In addition to this database, we also provide an aggregated data, useful for demonstrating the verification part of Wxgen. This database has averaged values for the whole of southern Norway. To run the examples in the wiki, you should therefore have two files:

.. code-block:: bash

$ ls db.nc db_agg.nc

Basic usage

Wxgen is a command-line program with three modes:

1. simulation
2. truth
3. verification.

The simulation mode creates one or more scenarios of a desired length, and is run as follows:

.. code-block:: bash

wxgen sim [options]

The output from this is a gridded NetCDF file that could be used for hydrological modelling.

The truth mode assembles a scenario that best represents the truth, based on the database. It connects the day 0 forecasts from the database (using ensemble member 0). When a day 0 forecast is not available, it will use the day 1 forecast from the previous initialization (and so forth until it finds one).

.. code-block:: bash

wxgen truth [options]

Finally, the scenarios can be evaluated and plotted using the verification mode as follows:

.. code-block:: bash

wxgen verif [options]

Memory usage

Creating scenarios can require a large amount of memory. However, the -mem option can be used to limit the amount of memory used. For example -mem 5 will force wxgen to only load about 5 GB of data from the database at once. This will in general mean that it will take longer to produce scenarios, since data may have to be read multiple times.

About this tutorial

The first part (creating scenarios) goes through the command-line options for creating scenarios. This part shows verification examples of why each option is beneficial. The third part is designed to be a reference for the different verification options.