Fig. 1.
Schematic diagram of ESRL OSSE components (bottom) versus real operational forecasting procedure (top).
Fig. 2.
Comparison of real data and OSSE O−F statistics for rawinsonde observations, each symbol indicates the global RMS O−F for a certain vertical level for 1d of the calibration period. Green symbols indicate O−A, red symbols indicate O−B, dark blue symbols indicate O−F for the 24hour forecast and cyan symbols indicate O−F for the 48hour forecast. Plus signs indicate the 1000hPa level, open circles the 850hPa level, closed circles the 500hPa level, open squares the 300hPa level, closed squares the 250hPa level, multiplication signs the 200hPa level, open diamonds the 150hPa level, open triangles the 100hPa level and closed triangles the 50 hPa level. (a, b) T126 OSSE prior to calibration, 5 July to 31 July 2005; (c, d) T126 OSSE after calibration, 5 July to 31 July 2005; (e, f) T382 OSSE after calibration, 5 August to 31 August 2005.
Fig. 3.
Comparison of real data and OSSE O−F statistics for aircraft observations, each symbol indicates the global RMS O−F for a certain vertical level for 1d of the calibration period. Green symbols indicate O−A, red symbols indicate O−B, dark blue symbols indicate O−F for the 24hour forecast and cyan symbols indicate O−F for the 48hour forecast. Plus signs indicate the 1000–700hPa range, open circles the 700–300hPa range and closed circles the 300–150hPa level range. (a, b) T126 OSSE prior to calibration, 5 July to 31 July 2005; (c, d) T126 OSSE after calibration, 5 July to 31 July 2005; (e, f) T382 OSSE after calibration, 5 August to 31 August 2005.
Fig. 4.
Comparison of real data and OSSE O−A and O−B statistics for AMSU-A and AIRS brightness temperatures, each symbol indicates the global RMS O−A or O−B for a particular channel at 0000 UTC for 1d of the calibration period. For AMSU-A, the channel colours are 1: black; 2: red; 3: green; 4: dark blue; 5: light blue; 6: magenta; 7: yellow; 8: orange; 9: purple; 10: yellow–green; 11: medium blue; 12: dark yellow; 13: aqua; 14: dark purple; 15: grey. For AIRS, these colours are rotated though the channel sequence, with every other channel plotted. O−A and O−B are plotted with the same symbol and colour for each channel, the O−A cluster has smaller RMS than the corresponding O−B cluster. (a, b) T126 OSSE, 5 July to 31 July 2005; (c, d) T382 OSSE, 5 August to 31 August 2005.
Fig. 5.
Global observation impact on the analysis field as a function of pressure level from data denial cases, from 15 July to 15 August 2005. Left, real data cases; right, OSSE cases with GMAO radiance observations. Top, temperature T (K); centre, specific humidity q (kg/kg); bottom, horizontal wind (m/s). RAOB impact, open circles; Aircraft impact, filled circles; AMSU-A impact, open squares; AIRS impact, filled squares; GOES radiance impact, crosses.
Fig. 6.
Analysis impact maps, 500hPa geopotential height, for real data denial cases from 15 July to 15 August 2005. Contour interval 2.5m.
Fig. 7.
As in Fig. 6, but for OSSE data denial cases from 15 July to 15 August 2005.
Fig. 8.
Global observation impact for OSSE data denial cases from 15 July to 15 August 2005, calculated using the Nature Run for verification. Top, temperature T (K); centre, specific humidity q (kg/kg); bottom, horizontal wind (m/s). RAOB impact, open circles; aircraft impact, filled circles; AMSU-A impact, open squares; AIRS impact, filled squares; GOES radiance impact, crosses.
Fig. 9.
Anomaly correlation for 500hPa geopotential height, calculated using analysis fields for verification, from 15 July to 15 August 2005. Individual forecasts: grey lines; mean of all forecasts: solid black line; one SD from mean: dashed black lines. (a, b) Real data control case; (c, d) OSSE control case with GMAO radiance; (e, f) OSSE control case with NCEP radiance. (a, c, e) 30°N–90°N; (b, d, f) 30°S–90°S.
Table 1. Mean, µ, and standard deviation, σ, of 120-hour forecast anomaly correlation for 500 hPa geopotential height, 15 July to 15 August 2005
[i] Northern Hemisphere (NH) areal mean from 30°N–90°N; Southern Hemisphere (SH) areal mean from 30°S–90°S.
Fig. 10.
Anomaly correlation for 500hPa geopotential height, calculated using Nature Run fields for verification, mean of forecasts from 15 July to 15 August 2005 as in Fig. 9. OSSE control case with GMAO radiance; (a) 30°N–90°N; (b) 30°S–90°S.
Fig. 11.
Mean anomaly correlation for 500hPa geopotential height for forecasts from 15 July to 15 August 2005. Solid black line, real data control; dashed line, ‘perfect’ OSSE control with GMAO radiance; dotted line with filled circles, error-added OSSE control with GMAO radiance; dotted line with open circles, error-added OSSE control with NCEP radiance observations. (a) 30°N–90°N, (b) 30°S–90°S.
Fig. 12.
Mean anomaly correlation for 500hPa geopotential height for forecasts from 1 August to 31 August 2005, as in Fig. 9 but for T382 cases. (a, b), real data control case; (c, d) OSSE control with GMAO radiance; (e, f) OSSE control with NCEP radiance. (a, c, e), 30°N–90°N; (b, d, f) 30°S–90°S.
Fig. 13.
Anomaly correlation impact of data denial cases from 15 July to 15 August 2005. (a, c) Real data cases; (b, d) OSSE cases. (a, b) 30°N–90°N; (c, d) 30°S–90°S. RAOB impact, open circles; aircraft impact, filled circles; AMSU-A impact, open squares; AIRS impact, filled squares; GOES radiance impact, crosses.
Fig. 14.
Global forecast impact for data denial cases from 15 July to 15 August 2005. (a, c) Real data cases; (b, d), OSSE cases. (a, c), 850hPa temperature impact, K; (b, d) 250hPa horizontal wind impact, m/s. RAOB impact, open circles; Aircraft impact, filled circles; AMSU-A impact, open squares; AIRS impact, filled squares; GOES radiance impact, crosses.
