windsurfatlanta.org weather: Windmodels

Southeast Radar (Image: Courtesy of NOAA)

Atlanta Windrose Plots. Historical Plot of Wind Direction and Strength by Month
Back to theABC Weather Page
Back to theABC Home Page

Weather Forecasts on the Internet

by Bert Eskridge, our mole at NOAA

Every twelve hours the National Weather Service gathers data from a world-wide system and runs several numerical models which predict a variety of atmospheric variables. The first model run is a global model which is on a coarse scale (about 5 degrees by 5 degrees and at 30 levels). The output of this model is use to give initial conditions and boundary conditions for the Nested Grid Model (NGM). The NGM is a high resolution model used to predict several atmospheric variables (temperature, humidity, heights of pressure levels, etc.) for the continental US. This is one of the models of interest to windsurfers.

It is difficult to use the NGM predicted fields directly to make weather forecasts. To predict what most people consider weather, the National Weather Service has used several years of NGM predicted atmospheric variables and hourly surface weather observations to developed statistical models to predict wind speed and direction, cloud cover, temperature, and precipitation at each individual weather station. These models are called Model Output Statistics or MOS models.

MOS forecast are available on the internet at the following two sites: (The Purdue site is very busy and you may get a quicker response from the second site.)

At the Purdue address, you type in the location of choice, in the example below KAVL, Asheville NC, choose MOS, and then get data. The output for Asheville is shown below. Time in these forecasts is given in Universal Time Coordinate or UTC. To get local time subtract four hours from the UTC time during daylight savings time and five hours for standard time.

In item I. below the predicted wind speed (knots) and direction is for a one minute average at the forecast time. Item II. below also says NGM MOS data. However, this is incorrect. It is the direct output of the NGM. The wind speed and direction is a layer average from the surface to about 1000 ft. The wind predicts from the NGM have been found to be too high. They are a good estimation of peak or gusts for the forecast time. The other variables are defined below.

POP12 Probability of receiving at least 0.01 in of precipitation.
COV cloud cover, CLR: clear to 20%, SCT: 20 to 60%, BRK: 60 to 80%, and OVC: is 80 to 100%.
CIL cloud height in 100+s of feet
DEW dewpoint (a measure of humidity)
PSL surface pressure
PREC six hour predicted precipitation
THK ignore
LI lifted index. Large negative numbers indicate strong thunderstorms.
VERT vertical wind speed. Negative values indicate clear skies, positive values clouds.

Item III. is the MOS forecast from the Medium Range Forecast model (MRF). This forecast is not as accurate as the NGM MOS for the first 48 hr. Under CLOUDS, WINDS, POP12 are two columns. The first is a prediction of the average value for the 00 to 12 UTC period and the second column is a prediction for the 12 to 24 UTC period. A predicted average wind speed of 12 knots for a 12 hr period means very good winds.

I.   NGM MOS data for AVL Initial time: 12Z  8 MAY 97

HOUR   TIME         TEMP   DEW       DIR  SPD   CIL     COV    VIS    WX     MX/MN    POP12
   6     8/18Z     67      50       190   10  250     SCT    10 
  12     9/ 0Z     63      54       240    9   50     OVC    10    TRW+ 
  18     9/ 6Z     58      54       330    9    3     OVC     4    TRW- 
  24     9/12Z     57      53       340   10   20     OVC     4     R-       54         65
  30     9/18Z     69      47       330   18   20     OVC    10    TRW- 
  36    10/ 0Z     59      42       340   17  250     SCT    10    TRW-      72         51
  42    10/ 6Z     48      37       340   15          CLR    10 
  48    10/12Z     44      34       340   18          CLR    10              41           0
  54    10/18Z     60      30       350   20          CLR 
  60    11/ 0Z     56      29       340   15          CLR                    64           0

II.    NGM MOS  (actually NGM output) data for AVL Initial time: 12Z  8 MAY 97

HOUR   TIME      PSL    PREC      DIR   SPD     THK   LI  VERT
   0     8/12Z   1021             180   14     6080    5    1.1
   6     8/18Z   1017   0.00      200   12     6140    2    0.7
  12     9/ 0Z   1012   0.12      210   23     6170   -1    7.2
  18     9/ 6Z   1013   0.22      290   18     6150   -1   -0.6
  24     9/12Z   1013   0.00      290   12     6110   -2    0.4
  30     9/18Z   1012   0.07      270   12     6100   -1    0.4
  36    10/ 0Z   1011   0.00      300   21     6060    9   -2.5
  42    10/ 6Z   1016   0.00      320   23     6010   13   -3.5
  48    10/12Z   1015   0.00      320   20     5970   15   -3.6
Total:                  0.41

III.  MRF MOS data for AVL Initial time:  0Z  8 MAY 97

 DAY     DATE    MAX  NRM           MIN NRM        CLOUDS     WINDS        POP12
  0     8 MAY     67   74 ( -7)      50               76        6          48
  1     9 MAY     65   74 ( -9)      52  50 ( +2)     96  75    6  12      88  55
  2     10 MAY    59   74 (-15)      41  50 ( -9)     37  32   10  11      13   2
  3     11 MAY    63   74 (-11)      37  50 (-13)     28  29    6   8       6   4
  4     12 MAY    67   74 ( -7)      41  50 ( -9)     34  45    3   7       5  15
  5     13 MAY    69   74 ( -5)      49  50 ( -1)     50  60    5   7      23  29
  6     14 MAY    70   74 ( -4)      52  50 ( +2)     67  64    3   7      36  30
  7     15 MAY    73   74 ( -1)      52  50 ( +2)     60  59    4   7      34  28