Fig. 1
Diagram of background wind vectors (BKG1 and BKG2) and the observational wind vectors (OBS1 and OBS2), used to present the differences of WRFDA background quality control procedures (BKG-QC) between asm_uv (standard assimilation method in WRFDA, assimilating wind observation in the forms of u and v components) and asm_sd (new assimilation method, assimilating wind observation in the forms of wind speed and direction).
Fig. 2
The experimental domain and the spatial and temporal distributions of MADIS surface (a) and satellite-derived Atmospheric Motion Vectors (AMV) observations (b–d) at 0000 UTC 1 August 2013 with the time window of ±3 h. The observation count (nObs) is shown by time window on top. The time window is denoted by colours in (a), and the vertical height is denoted by colours in (b–d).
Fig. 3
The background errors calculated by the National Meteorological Center (NMC) method for wind speed (a) and direction (b). The background errors are estimated by eq. (5). The degree number in (b) denotes the maximum background error of wind direction allowed for the statistics, and the percent number denotes how many grid points survive the statistics.
Fig. 4
The observation errors of wind speed (SPD) and direction (DIR) estimated by eq. (3) for MADIS surface wind (SFC) and satellite-derived Atmospheric Motion Vectors (AMV) observations.
Fig. 5
The distribution of surface wind direction (a–b) and speed (c–d) observations by observation minus background (OMB, black lines) and observation minus analysis (OMA, red lines) in EX_uv and EX_sd. σ and µ in panels denote the standard deviation and bias of OMB and OMA, respectively.
Fig. 6
The distributions of satellite-derived Atmospheric Motion Vectors (AMV) wind direction (a–b) and speed (c–d) observations by observation minus background (OMB, black lines) and observation minus analysis (OMA, red lines) in EX_uv and EX_sd. σ and µ in panels denote the standard deviation and bias of OMB and OMA, respectively.
Fig. 7
The time series of rmse and bias of wind speed (a–b) and direction (c–d) analyses in EX_uv and EX_sd fit to rawinsonde observations (RAOB). Near-surface RAOB (nlev ≤ 2) denotes RAOB available on two lowest height levels; upper-air RAOB (nlev > 2) denotes RAOB available above two lowest height levels. Black and red lines are statistics in EX_uv and EX_sd; solid and dot lines represent rmse and bias, respectively.
Fig. 8
The time series of rmse and bias of wind speed (a) and direction (b) analyses in EX_uv and EX_sd_uvobs fit to rawinsonde observations (RAOB). Black and red lines are the statistics in EX_uv and EX_sd_uvobs; solid and dot lines represent rmse and bias, respectively.
Fig. 9
The distribution of surface wind speed observations by observation minus analysis (OMA) in EX_sd and EX_sd_uvobs (a) and of surface wind direction observations by observation minus analysis (OMA) in EX_uv and EX_sd_uvobs (b).
Fig. 10
The biases of wind speed analyses in EX_uv, EX_sd and EX_sd_sfc fit to rawinsonde observations (RAOB).
Fig. 11
The rmse and bias of wind speed and direction forecasts in EX_uv, EX_sd fit to rawinsonde observations (RAOB) by forecast lead time.
Fig. 12
The vertical profiles of rmse and bias of 12-h wind speed and direction forecasts in EX_uv and EX_sd fit to rawinsonde observations (RAOB).
Fig. 13
The differences of 12-h wind speed and direction forecasts valid at 0000 UTC 3 August 2013 in EX_uv and EX_sd from rawinsonde observations (RAOB) at KSGF station in Springfield, Missouri.
Fig. 14
The Stage IV 24-h accumulated precipitation analysis (shaded) valid at 1200 UTC 3 August 2013 in (a); the analyses of wind (vector) at 850 hPa and specific humidity (red line) at 700 hPa at 1200 UTC 2 August 2013 and 24-h accumulated precipitation forecasts (shaded) valid at 1200 UTC 3 August 2013 in EX_uv (b) and EX_sd (c). The unit for precipitation is mm, for specific humidity is g kg−1, and for wind vector is ms−1. The black dot in (a) denotes the rawinsonde station KSGF in Springfield, Missouri. South Dakota, Kansas, Missouri, Illinois and Indiana are abbreviated to SD, KS, MO, IL and IN, respectively.