Eta 40 km  - Accuracy Tests
 Updated: 8-23-98
 
     The fundamental basis for a good soaring forecast is a good morning sounding.  This process was developed at New Market, VA (8W2).  For this area Roanoke, VA (RNK), and Washington / Dulles, (IAD) are the only sources available for balloon soundings.  Both of these stations are some 90 miles away, and neither is in line with the prevailing wind flow.  Another factor is that many of us live some 2 hours driving time from the airport making it impractical to wait for data to become available.

     The Eta 40 km computer model has proven to be a great alternative to traditional sources.  This data provides an interpolated profile up to 60 hours in advance for any point designated by Longitude and Latitude coordinates.  A verification of this process was made by comparing actual 12Z balloon sounding data with 12 hour projections for RNK and IAD.  On the evening before a given day the projected sounding data was acquired for these two sites.  The morning sounding data was then compared.  Ten sets of balloon sounding data were algebraically subtracted from the corresponding computer model data at selected pressure levels.  The following table illustrates these differences in terms of mean and [standard deviation].
 
 
***** Summary of Data Subtractions: Dec. 7,8,9,13,14 (1997) for RNK and IAD *****

Pressure Level (mb)             Temp (C)          D.P. (C)              Wind (dir)        Wind (kts)

    500 (18,000 msl)                -0.3 [1.8]         +2.5 [4.5]               -3.8 [14.3]         +2.1 [10.3]
    700 (10,000 msl)               +0.4 [1.1]         +1.9 [10.0]           -13.8 [25.6]          -0.6 [8.5]
    850   (4,800 msl)                -1.0 [3.0]          -0.5 [3.6]             -10.5 [29.5]          -1.5 [8.8]
    925   (2,300 msl)                    0 [1.5]          -1.1 [2.3]            +14.4 [28.6]         +2.3 [5.5]
 
 
     Standard deviation reflects the distribution of values around the mean, or average, of the data.  The spread of these values as well as their concentration around the average is important in representing the accuracy of the measurements; or in this case how well projected data corresponded to actual measurements.  A low standard deviation reflects a mean (average) that is based on narrow deviations in relation to the balloon sounding data.

    The table above represents deviations in temperature, wind direction, and wind speed.  Wind direction deviations in the clockwise direction are represented as positive values.  This indicates that 12 hour projections of wind direction are rather general.  Wind speed is more accurate.  The temperature profile is the basis for anticipating thermal activity and, as demonstrated here, the projection is very accurate.

     One additional comment… The hour-by-hour soundings provided by GOES-8/9 (satellite remote sensing) indicate that temperature profiles above 3,000 feet (agl) remain rather constant through the course of normal daytime heating.  This finding contributes to our confidence in the standard method of deriving the thermal index profile. 

Projected Soundings Subtracted From Balloon Soundings:

           [ RNK: 8-12-98@12Z ]  | - |  [ Eta40km: 8-11-98@7:45pm ]

     Pressure Level          Temp.         D.P.          Wind (dir)          Wind (kts)

            500                       0              +2                 -2                      +1
            700                     - 1              +2                -10                      0
            850                     +2              +1                +32                    -1
            925                     +1              +1                -28                     +2

           [ IAD: 8-18-98@12Z ]  | - |  [ Eta40km: 8-17-98@7:20pm ]

           500                     -1               +3                +26                    -9
            700                      0              -10                +26                      0
            850                     +1              -4                  +2                    +2
            925                      0               -1                 +20                    +3 

           [ RNK: 8-18-98@12Z ]  | - |  [ Eta40km: 8-17-98@8:10pm ]

           500                      0               +5                 +20                   -4
            700                     +1               0                  +21                   -1
            850                     +2              -2                 +10                   -2
            925                      0               -1                 +19                     0 

           [ IAD: 8-21-98@12Z ]  | - |  [ Eta40km: 8-20-98@7:10pm ]

           500                     0              +7                  +12                   -4
             700                    -1              -3                  +12                   -6
             850                     0               -6                  -11                   -9
             925                    -1                0                  -16                   -4 

           [ RNK: 8-21-98@12Z ]  | - |  [ Eta40km: 8-20-98@8:15pm ]

             500                   +1               -6                  +21                  +1
             700                   +1              +14                 -24                  -2
             850                    -1               -1                  +11                 -11
             925                    -1               -2                   -80                  -2 

Balloon Sounding Data:
     http://weather.uwyo.edu/upperair/sounding.html

Eta40km Data:
     http://www.arl.noaa.gov/ready/profile.html

Upper Air Sounding Plots:

   http://weather.unisys.com/upper_air/skew/index.html