Title: Transfer of Regional NWP Capabilities to Developing Countries Detlev Majewski, Deutscher Wetterdiens
1Transfer of Regional NWP Capabilities to
Developing CountriesDetlev Majewski,
Deutscher Wetterdienst, Germanye-mail
detlev.majewski_at_dwd.de
2Stepwise introduction of regional NWP
- Defined at the WMO Workshops in Pretoria (2002)
and Casablanca (2005) - Acquiring and using existing NWP grid point
products - Acquiring and running a local NWP model without
data analysis, but with verification - Running a local model, without data analysis,
fine tuning the system for the local region with
statistical adaptation - Running a local model with local analysis of data.
3Main components of a regional forecasting system
and corresponding data flow
4Key issues for NWP capacity building
- Defined at the WMO Workshops in Pretoria (2002)
and Casablanca (2005) - Hardware (computer, internet access)
- Software (operating system, compilers, regional
NWP model, graphics) - Data (initial and lateral boundary data files
local observations) - Manpower (computer skills (Unix, Fortran, Web),
meteorologist) - Training of NWP staff and forecasters
- Modernization of complete forecasting process
- Sustainability (involve universities, form
regional co-operations)
5GME data as lateral boundary data of HRM
The global model GME covers the Earth with an
icosahedral-hexagonal grid grid spacing 40 km
and 40 layers. 368642 grid points / layer Grid
cell area of GME 1384 km2 To cover a given HRM
domain, only a small part of GME grid points is
needed. If the HRM domain is 3000 x 3000 km2, the
number of GME grid points is given by 3000 x 3000
/ 1384 6500 only! Only these GME grid points
are being transferred to the National Weather
Service via the internet for running HRM.
HRM
6HRM at KMD, Kenya
Grid spacing 0.25 28 km 26E - 50E 12S -
12N 97x97 grid points, 40 layers Linux PC
server 1 processor Intel P4 1 GB RAM 2
scientists from 2001 until 2005. 3 scientists
since 2006. http//www.meteo.go.ke/nwp/
7Lessons learned, next steps
- Introduce NWP stepwise, taking the local
knowledge into account. - Build up a team of dedicated scientists,
encourage networking with the HRM community
worldwide, e.g. via http//groups.google.de/group/
hrm_help. - Aim for early showpieces to convince local
users. - Try to streamline the total forecasting process,
not only the NWP part involve the local
forecasters quite early, e.g. in the definition
of forecast products and graphics. - Start objective verification of the NWP model
against local observations as soon as possible. - Try to involve local universities by offering the
NWP model, data for case studies and computer
resources. - Overall goal Achieve a sustainable improvement
of the forecasting process at the service. - Regional NWP Training in Botswana 3 9 April 08
- Next Regional NWP Training Workshop at DWD 14 -
25 July 08
8Building blocks for regional NWP
- Topographical data of region at 1 km x 1 km
resolution - Regional NWP model (e.g. ALADIN, HRM,
MM5/WRF/ETA, RAMS) - Initial and lateral boundary data from global NWP
model (e.g. GFS, GME, JMA, ECMWF) - Reliable and reasonable fast data connection
between global centre and national weather
service running the regional model - Computer system (Linux PC cluster) to run
regional model - Graphical display of forecasts (e.g. GrADS,
DIANA Open Source Visualization of the Norwegian
Met. Service) Web pages - Verification package (e.g. ORMVERIF)
- Regional NWP training by advanced centres (e.g.
Météo France, DWD, KMA, NCEP, UKMO, SAWS, ACMAD,
WMO) - Training of computer skills (Linux, Fortran,
html) - Training of local forecasters in model
interpretation
9Any questions?
Questions?
HRM is used at Bosnia, Botswana, Brazil (INMET,
DHN), Bulgaria, China (Guangdong), India (SPL),
Israel, Italy, Jordan, Kenya, Libya, Malawi,
Mozambique, Nigeria, Oman, Pakistan,
Philippines, Romania, Senegal, Spain,
Tanzania, United Arab Emirates and Vietnam. http
//www.met.gov.om/hrm/
Graphics by B. Ritter (DWD)