1. Bolin Centre
2. Data

Forecast of low clouds in the Arctic using Weather Research and Forecasting Model

Martin Hagman

This dataset contains output from the Weather Research and Forecasting Model (WRF), used by the Swedish Armed Forces. Their setup of WRF is known to have problems to forecast low clouds in stably stratified conditions in the Arctic Boundary Layer when the ground is covered by snow. The current dataset is generated to understand the cause for this.

Re-runs of WRF, using both a Single Column Model (SCM) as well as the full 3D model, have been conducted. 3D model data from January and February 2018 have been used to visualize the deficiencies. The SCM was set up in order to do numerous runs faster and to isolate physics from dynamics. This gives a better understanding of what goes wrong and different solutions can be evaluated. Cloud base output from both the 3D and SCM model was compared with observations from Sodankylä in northern Finland. When the SCM output was satisfactory, the same solution was evaluated also in the the full 3D-model.

Ouput from both the 3D model and the SCM is stored as NetCDF files. In the SCM, output is stored every 20 seconds, which also is the time step used. In the 3D model output is stored every 10 minutes or every hour, depending on how long the model runs are. The domain covers Denmark, Sweden, Norway and Finland as well as small parts of the Baltic states and the most northern parts of the European continent.

Download data

You can download all files automatically using the software wget and the following single command.

wget -i https://bolin.su.se/data/tmp/hagman-2020.txt

You may need to download the software wget.

The command uses wget to download all files specified in a file list. A list of all files is available at https://bolin.su.se/data/tmp/hagman-2020.txt

Because of the large data volume, downloading all files can take can take long time and requires sufficent disk space on your system.

To download a selected subset, you can edit the file list and run wget with your revised list. Alternatively, you can select files for download in the file list below.

Data files

Click on the categories below to view the files. Click on the download button to download the files shown in the list. Depending on your connection speed it may take considerable time to download the data since it is rather big. Thus, it is not recommended to use the web browser to download more than one day (24 files) per batch. Please use wget to download larger batches, as described above.

Citation

Martin Hagman (2020) Forecast of low clouds in the Arctic using Weather Research and Forecasting Model. Dataset version 1.0. Bolin Centre Database. https://doi.org/10.17043/hagman-2020

References

Data description

Model simulations data are stored as NetCDF files both from a Single Column Model (SCM) and from the full 3D model. Scripts used to visualize the data are written i Python and can be accessed in a git code repositiory.

3D model

A certain simulation is stored in a directory with the date and time when the model simulation was started. Additional information is added to the name of the folder to shortly give more knowledge about the simulation and, in some cases, manipulations done to the initial field. The initialization file has the name "wrfinput_d01". So, for example:

2018-02-18_00z_91_LEVELS

A simulation with 91 vertical levels started at 00z 2018-02-18. The name of every output file in the directory has the form, wrfout_d01_2018-01-18_00:00:00, which tells the time of the NetCDF file. This file contains all variables.

2018-02-18_00z_91_LEVELS_QC_QI_42h_T_to_Td_1e-5

A simulation with 91 levels started at 00z 2018-02-18, which is 42h long and where cloud water and cloud ice from IFS HRES are assimilated in an area around Sodankylä. Additionally, the temperature is lowered to the dew point at every level where the cloud water amount exceeds 1*10e-5 kg/kg air.

2018-02-18_00z_91_LEVELS_QC_QI_T_to_Td_0_Whole_Dom_MYJ_Icloud_2

A simulation with 91 levels started at 00z 2018-02-18, where cloud water and cloud ice from IFS HRES are assimilated in the whole domain. Additionally, the temperature is lowered to the dew point at every level where the cloud water amount exceeds 0 kg/kg air. The simulation is conducted with the MYJ Planetary Boundary Layer (PBL) scheme and cloud fraction parameterization set to 2 in the namelist.input file. Namelist.input is changed before a simulation is started and is stored in the same directory as the wrfout files.

2018-02-27_00z_91_LEVELS_Clouds_Where_Clouds_In_Sounding_YSU

A simulation with 91 levels started at 00z 2018-02-27, where cloud water is assimilated from where it appears in the vertical sounding observations from Sodankylä. The cloud water is assimilated in an area over the northern parts of Sweden and Finland in a way so that the liquid potential temperature is constant in the cloud. The PBL scheme used is YSU.

SCM model

This is, easily said, one column of data. Analyzing the data, one finds that it comprises a 23 or 22 matrixes, depending on variable. Every column of data has the same values, due to very high diffusion in the SCM, so it doesn't matter which column to look at. One difference compared to the 3D-model is that the whole model run is stored in one NetCDF file with all time steps into it. The name conventions are almost the same as for the directories in the 3D-model cases:

wrfout_d01_2018-02-18_00:00:00_QC_QI_T_to_Td_1e-5

A simulation started at 00z 2018-02-18, where cloud water and cloud ice from IFS HRES are assimilated. The temperature is lowered to the dew point at at every level where the cloud water exceeds 1e-5 kg/kg air.

wrfout_d01_2018-02-27_00:00:00_100T_LIQ_YSU

A simulation started at 00z 2018-02-27, where cloud water is assimilated at the vertical levels, where clouds appear in the observation vertical soundings from Sodankylä. The cloud water is assimilated in a way that keeps the liquid potential temperature constant in the cloud.

wrfout_d01_2018-02-27_00:00:00_025T_LIQ_YSU

Same as the recent case, but only 25 % of the total liquid potential adiabatic cloud water is initialized.

Comments

Swedish Armed Forces setup of WRF has problems to forecast low clouds in stably stratified conditions in the Arctic Boundary Layer when the ground is covered by snow. This dataset is generated to understand the cause for this. Re-runs, using both a Single Column Model (SCM) as well as the full 3D model, are conducted. First of all, stored 3D model data from January and February 2018 is used to visualize the deficiencies. Cloud base output from the model is compared with observations from Sodankylä in northern Finland. To be able to do numerous runs faster and to isolate physics from dynamics a SCM is set up. This gives a better understanding of what goes wrong and different solutions can be evaluated. The output from the SCM is also compared to observations from Sodankylä. When the SCM output is satisfactory the same solution is evaluated also in the the full 3D-model.

Ouput from both the 3D-model and the SCM is stored as NetCDF files. In the SCM, output is stored every 20 seconds, which also is the time step used. In the 3D model output is stored every 10 minutes or every hour, depending on how long the model runs are.

The effort to develop WRF began in the latter 1990's and was a collaborative partnership of the National Center for Atmospheric Research (NCAR), the National Oceanic and Atmospheric Administration (represented by the National Centers for Environmental Prediction (NCEP) and the Earth System Research Laboratory), the U.S. Air Force, the Naval Research Laboratory, the University of Oklahoma, and the Federal Aviation Administration (FAA). For researchers, WRF can produce simulations based on actual atmospheric conditions (i.e., from observations and analyses) or idealized conditions. WRF has a large worldwide community of registered users (a cumulative total of over 48,000 in over 160 countries), and NCAR provides regular workshops and tutorials on it.

The data set can be used to better understand the interaction between the air and the snow surface in cold Arctic conditions.

Model data from both the full 3D-model and the SCM is compared to observations from Sodankylä in northern Finland during January and February 2018. The SCM runs are idealized, i.e. they are started from an observation state of the atmosphere but is not forced with large scale dynamics such as subsidence or advection. The observation state comes from the IFS HRES analysis where the nearest grid point from Sodankylä is chosen.

A lot of parameters from the model are compared to observations, such as cloud fraction, specific humidity, relative humidity, 2m temperature, skin temperature, potential temperature, down welling radiation, outgoing radiation etc. Even more parameters are used in, for example, the conversion of pressure model levels into height in meters above the model terrain etc.

The data was generated by myself, on the Bi cluster, at the National Supercomputer Centre (NSC) at Linköping Universityu. It has a resolution of 3 km horisontally and has 46 and 91 levels vertically. The domain covers Denmark, Sweden, Norway and Finland as well as small parts of the Baltic states and the most northern parts of the European continent.

Web address (URL)

https://bolin.su.se/data/hagman-2020

DOI address

https://doi.org/10.17043/hagman-2020