Premium
NUMERICAL WEATHER PREDICTION
Author(s) -
Richard Dumelow,
Andrew C. Lorenc
Publication year - 1956
Publication title -
weather
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.467
H-Index - 40
eISSN - 1477-8696
pISSN - 0043-1656
DOI - 10.1002/j.1477-8696.1956.tb00349.x
Subject(s) - citation , computer science , numerical weather prediction , information retrieval , meteorology , library science , geography
An Observing System Experiment is carried out using the Met Office threedimensional variational data assimilation scheme (3D-Var) and global forecast model. The experiment investigates the impact on forecasts of up to 6-days of denying whole observing systems (radiosonde, satellite, aircraft and surface). A total of 60 forecasts from two 30-day periods July 2001 and January 2002 are verified. The results show that overall all observing systems have a positive impact on mean forecast quality indicating that the Met Office 3D-Var system performs well. The experiment also gives an estimate of the relative benefit that the different components of the Global Observing System are having on Met Office NWP forecasts. Results indicate: • Satellite data have the largest positive impact of all observation types on forecasts in the southern hemisphere and have a significant positive impact in the tropics and northern hemisphere. • Satellite radiance data have a larger impact than data from Atmospheric Motion Vectors (AMV) which have a measurable impact in the extra-tropics only when satellite radiance data are not present. The benefit of AMV data can be most clearly seen in the tropics. • Surface data have a significant impact on surface pressure forecasts, but there may be some redundancy in the network for global Met Office NWP forecasts run at 90 km resolution. • Radiosonde data are important for northern hemisphere forecasts and also have benefit in the tropics and southern hemisphere. • Aircraft data play a relatively important role, particularly for short-range wind forecasts in the northern hemisphere and tropics. The results from this experiment are in broad agreement with those from a similar experiment, using four-dimensional variational assimilation, carried out by ECMWF. A further experiment investigates the impact of sub-components of the surface network (reports from land surface stations, ships and buoys). 30 forecasts from July 2001 are verified. The results from this experiment show that direct 'in-situ' observations of surface pressure have a large positive impact on NWP forecasts of mean sea level pressure. But removing whole sub-components of the surface network does not lead to a noticeable reduction in forecast quality.