A data assimilation experiment and the global circulation during the FGGE special observing periods

Abstract

SUMMARYDuring the two Special Observing Periods of the First GARP Global Experiment (FGGE) a data assimilation experiment was conducted at the UK Meteorological Office. Throughout the period of the experiment observational data were assimilated dynamically into a numerical forecast model at the four main synoptic hours of each day.In the first part of the paper, the method of data assimilation is briefly described. Following that, a description is given of the zonally averaged atmospheric circulation, as diagnosed from our analyses. The results of this study are in generally good agreement with those found by previous workers. ~NTRODUCTIONDuring the Special Observing Periods (SOPS) of the First GARP Global Experiment (FGGE), the Meteorological Office conducted a data assimilation experiment. The purposes were to evaluate the performance of a method of dynamic data assimilation in a semioperational environment, and to produce a comprehensive set of global analyses for research purposes.The method of data assimilation will be briefly described, with more complete documentation available as a Meteorological Office internal report (Birch and Lyne 1980). It is essentially that due to Lorenc (1976) and is equivalent to the dynamical relaxation schemes of Hoke and Anthes (1976) and Davies and Turner (1977).The method of assimilating data directly into a numerical forecast model was used by Charney et al. (1969) and since by numerous other investigators (see Bengtsson 1975). Model variables are replaced by observed values at or near their time of validity. Refinements in methods of interpolation and data insertion have resulted in techniques which approximate closely the dynamical relaxation methods mentioned above. Miyakoda et al. (1978) have described such a system which, in one of its forms, is very similar to the method described here.The data were assimilated into the model as it was integrated over so-called assimilation periods, and between these periods the model was integrated normally. At each time step during an assimilation period the observed data (or rather their differences from values predicted by the model) were interpolated to model grid points, and these interpolated differences were then used to correct the model values. The variables acted upon were the basic variables of the model -temperature, horizontal wind, humidity mixing ratio and surface pressure.The SOPs ran from 5 January to 5 March and I May to 30 June 1979, though in each case the assimilation was started five days earlier to allow for initial adjustment. The assimilations were continued beyond the end of the second SOP, until 14 August 1979, to provide analyses for future investigations of the south-west monsoon. The initial conditions were interpolated from the Meteorological Office's operational analysis north of about 19"N, and merged with climatological fields in the tropics and Southern Hemisphere. The latter were zonally averaged south of 15"s. A brief account of the day to day running and performance of the assimilation scheme...

The impact of wave number selection and spin‐up time in spectral nudging

“…It is expected that the surrounding grid points will accommodate to this reference value and, hence, give a more realistic result, consistent at the same time with the physical equations of the model. One of nudging's first applications was in data assimilation (Anthes, 1974), and although nowadays different, more complex, techniques are being used for this purpose, nudging was once operational in weather forecast centres such as the Met Office both for global (Lyne et al, 1982) and regional (Bell, 1986) data assimilation. When a gridded analysis is available, nudging is generally applied throughout the model domain, constraining the solution by the reference fields at all points (Davies and Turner, 1977;Stauffer and Seaman, 1990).…”

The back and forth nudging algorithm applied to a shallow water model, comparison and hybridization with the 4D‐VAR

“…In spite of the differences, the term nudging is also sometimes used in the context of statistical interpolation. The standard nudging algorithm, which initially appeared in meteorology [6], is the first data assimilation method used in an operational way in oceanography [7][8][9]. Some recent studies have shown that it is possible to determine in a systematic way the optimal weighting coefficients of the recall force to the observations, and then nudging is equivalent to the Kalman filter, or in a variational form, to 4D-VAR [10,11].…”

Outlook for Improved Numerical Weather Prediction Using Satellite Data with a Special Emphasis on the Hydrological Variables,

“…The second group, the so-called dynamic relaxation methods, have been used extensivly by the UKMO and others (Hoke and Anthes, 1976;Davies and Turner, 1977;Fiorino and Warner, 1981;Lyne et al, 1982;Bell, 1983). In these methods a term forcing the model state toward the observations is added to the governing equations.…”