Horizontal wind at 700 hPa over Sumatra (0°N, 100°E ) analyzed by National Center for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis was compared with wind observed by a VHF wind-profiling Doppler radar at Kototabang (0.20°S, 100.32°E) during 2001 2007. Radiosonde wind data at Padang (0.88°S, 100.35°E) were reported through Global Telecommunication System (GTS) for assimilation into NCEP/NCAR reanalysis. To examine if radiosonde observations at Padang improve wind quality of NCEP/NCAR reanalysis over Sumatra, dataset of radar wind, which was compared with NCEP/ NCAR reanalysis, was divided into the group when radiosonde observations were reported though GTS (Group A) and when they were not reported (Group B). For each group, correlation coefficient and regression line were computed. In Group A, both zonal and meridional winds showed better correlation coefficients (0.90 and 0.77) than in Group B (0.79 and 0.63). Amplitude of NCEP/NCAR-reanalysis zonal and meridional winds showed a better agreement in Group A (82 and 96% of radar wind) than in Group B (66 and 80% of radar wind). These results suggest that wind observations in the Indonesian Maritime Continent contribute to improve wind quality of NCEP/NCAR reanalysis there.
IntroductionThe Indonesian Maritime Continent (hereafter IMC) is one of the regions where deep cumulus convection occurs most frequently (e.g., Yamanaka et al. 2008). Because cumulus activities over the Indian Ocean, IMC, and the western Pacific are modulated by synoptic-scale disturbances such as intra-seasonal variation (ISV; see Zhang 2005), many studies investigated synoptic-scale wind disturbances over them using global objective reanalyses (e.g., Matthews 2000). Previous studies have shown that assimilating observed winds to objective reanalysis contributes to improve wind quality of objective reanalysis in the tropical region. Using a windprofiling Doppler radar at Christmas Island in the central Pacific (2°N, 157°W), Gage et al. (1988) showed that monthly-mean standard deviation (bias) between observed and reanalysis winds reduced from 3 5 m s 1 (1 3 m s 1 ) to 1 2 m s 1 (less than 0.5 m s 1 ) by assimilating observed winds into objective analyses. Using radar wind data observed at 4 tropical stations, Schafer et al. (2003) showed that agreement between NCEP/NCARreanalysis and radar winds was closest at the station where radar wind data were assimilated in NCEP/ NCAR reanalysis (Christmas Island) compared with other 3 stations where radar wind data were not assimilated. The poorest agreement occurred at Piura, which is farthest from the operational radiosonde site (1234 km). Over IMC, upper-air observations assimilated into objective reanalysis are sparse. Though IMC has longitudinal extension of~4000 km, only 12 radiosonde observation sites exist over Indonesia (Yamanaka et al. 2008;Okamoto et al. 2003). Further, observations were not always carried out (later explained in Table 1). Therefore quality of wind in objective reanalysis ove...