access_ExtremeWxEventUseCase - ACCESS-NRI/accessdev-Trac-archive GitHub Wiki

PageOutline

Notes on "Investigating an Extreme Weather Event" CWSLab Use Case

WORK IN PROGRESS -- IN REVIEW

Michael Naughton [email protected], Wenming Lu [email protected]

The NeCTAR CWSLab project documents describe a use case for a researcher to investigate an extreme weather event by performing a high-resolution weather simulation experiment starting from an existing regional analysis and forecast product. This is WP1-U5 use case in the CWSLab [attachment:wiki:AccessNectarCwsl:WP1_UseCases.pdf WP1 Use Cases document], is included in reporting for Milestone 30 from the [attachment:wiki:AccessNectarCwsl:WP1_Plan_v1.0.pdf WP1 Project Plan].

The use case is broken down into the following sequence of steps. These wiki notes describe how these steps could be approached, using the ACCESS Model Experiment Library where appropriate.

Process to perform a high-resolution weather forecast simulation experiment nested (one-way) in a regional analysis and forecast

1. Researcher interested in particular extreme weather event

This is up to the researcher's curiousity.

2. Initial analysis using BOM Australian region analysis and forecast products

For Bureau researchers, weather charts are available internally on the Bureau's internal web and repositories. For external researchers, ACCESS NWP analysis and forecast fields are available at NCI in the rr4 project area of the RDSI repository (registration required). In both cases, both current and previous case periods are available. Using these archives, the researcher can examine cases and standard operational model performance in detail.

3. Decides to investigate simulation sensitivity to grid resolution

Based on this initial study, the researcher is interested to study high-resolution aspects of one or more severe weather events. One previous example has been to run 1.5km and higher resolution fire weather simulations for Victorian or WA fires, i.e. higher resolution than the APS0 (5km) or APS1 (4km) operational ACCESS-C systems. Another has been to run some Tasmanian rain events with a 4km domain matching existing WRF runs.

4. Selects relocatable model experiment as a starting point

Relocatable model congurations are provided in the ACCESS-C section of the ACCESS Model Experiment Library, at 4km resolution with the APS1 operational configuration, and 1.5km with NWP research settings.

Choosing one of these configurations, the first step is to install the experiment, run it as is without changing the domain or any other settings, and check the results against the reference results provided. These are steps 4.1 & 4.2.

4.1 Builds a grid with specific grid resolution

4.2 Runs and checks against archived output

5. Reruns Australian regional model from archived initial conditions and lateral boundary conditions to generate specific LBCs for high resolution model

We are assuming in this experiment that the case of interest is different to what has already been run in the operational ACCESS-C system, e.g. in its location or resolution or configuration, so the user needs to generate appropriate initial and boundary conditions. The step mentioned here of rerunning ACCESS-R to generate different boundary conditions should not be required, as we now have archive of daily ACCESS-R nesting "pi" files archived on raijin, viz. /g/data/rr4/samnmc/access-r/2014/04/pi_2014040100. All dates from Apr-2014 onwards are being routinely archived there, with selected earlier periods also available. Depending on space limitations, files may at some time be archived to tape, or removed. Other dates may also be made available on request.

5.1 Files obtained via catalogue w/o knowing details of file system or archive

This documentation is provided in lieu of a catalogue.

6. Runs high resolution model (version of ACCESS-C or 1.5 km relocatable)

The researcher now proceeds to run the the high-resolution model for his case of interest, following the instructions in the wiki page.

7. Archives and analyses model forecast results

The model outputs produced by the run can then be analysed by the researcher, using methodologies and tools of their choosing, depending on their scientific needs. Some of the plotting and analysis tools used in BoM research can also be provided on request for collaborating researchers to use.

Appendix: Sequence of steps for high resolution fire modelling studies (Robert Fawcett)

The following sequence of steps has been used in high-resolution fire weather modelling studies.

Process to perform a high-resolution weather forecast simulation experiment nested (one-way) in a regional analysis and forecast product.

  1. Researcher interested in particular extreme weather event

  2. Initial analysis using BOM operational Australian region analysis and forecast products

  3. Decides to investigate simulation sensitivity to grid resolution and/or domain choice

  4. Selects target domain and resolution.

  5. Does the user-specified target domain lie entirely within an operational ACCESS-CITY 4-km domain? 5.1. if โ€œyesโ€?

    5.1.1. Acquire ACCESS-CITY initial condition and its lateral boundary conditions (inherited from ACCESS-R or ACCESS-G)

    5.1.2. Rerun ACCESS-CITY 4-km forecast to generate LBCs for user-specified target domain (at 15-minute intervals or more frequently)

    5.1.3. Run ACCESS 1.5-km forecast on user-specified 1.5-km target domain, saving outputs at user-specified frequency (e.g., every 15 minutes, every 5 minutes, every 3 minutes)

    5.1.4. (optional) Extract LBC information from results of previous step (5.1.3.) Run ACCESS 0.5-km forecast on user-specified secondary target domain, saving outputs at user-specified frequency

    5.2. if โ€œnoโ€?

    5.2.1. Acquire ACCESS-R initial conditions (inherited from ACCESS-G).

    5.2.2. Rerun ACCESS-R 12-km forecast to generate LBCs for user-specified 4-km target domain

    5.2.3. Run ACCESS 4.0-km forecast on user-specified 4.0-km target domain

    5.2.4. Run ACCESS 1.5-km forecast on user-specified 1.5-km target domain

    5.2.5. (optional) Run ACCESS 0.5-km forecast on user-specified 0.5-km target domain

Back to ACCESS Model Experiment Library