GFS forecast compared to local measurements at the Paranal radiometer
The value of the Pearson Correlation Coefficient between PWV forecasts from the GFS model (at
0, 6, 12 UTC) and the Paranal radiometer measurements
from January to December 2012. In parenthesis is shown for comparison the correlation coefficient without prediction, i.e. assuming PWV equal to its value at the time of analysis (persistence).
|
The statistics of the GFS analyzes at Paranal pressure level (750mb) are compared to Armazones pressure level (710mb)
Forecast Step |
Forecast hour (UT) |
All months |
Summer (Oct-Mar) |
Winter (Apr-Sep) |
---|
T+00h (analysis) |
all 00 06 12 18 |
0.87 0.86 0.88 0.88 0.88 |
0.93 0.96 0.95 0.97 0.91 |
0.80 0.81 0.84 0.78 0.80 |
T+24h |
all 00 06 12 18 |
0.84 (0.76) 0.79 0.86 0.86 0.86 |
0.90 (0.83) 0.90 0.91 0.93 0.86 |
0.75 (0.56) 0.70 0.81 0.72 0.76 |
T+48h |
all 00 06 12 18 |
0.79 (0.65) 0.75 0.78 0.79 0.79 |
0.87 (0.73) 0.89 0.86 0.90 0.86 |
0.70 (0.30) 0.61 0.81 0.69 0.73 |
T+120h |
all 00 06 12 18 |
0.72 (0.44) 0.68 0.71 0.76 0.76 |
0.83 (0.51) 0.79 0.79 0.90 0.87 |
0.51 (0.17) 0.42 0.59 0.48 0.52 |
Ascii file description
The ascii file contains 6 columns of data that represent: [Year Month Day
and Hour of forecast initiation, Forecast step (hour), GFS forecast (mm H2O)]
|
|
|
ERASMUS Model analyses compared to local measurements by the Paranal radiometer
PWV analyses (=forecasts at step 0) from the ERASMUS model at
0, 6, 12 and 18 UTC are compared to the Paranal radiometer measurements when available
from January to December 2012. The best linear fit, as well as the value of the Pearson Correlation Coefficient
are given in the graph.
|
All Data
GFS analyses compared to local measurements by the Paranal radiometer
PWV analyses (=forecasts at step 0) from the GFS model at
0, 6, 12 and 18 UTC are compared to the Paranal radiometer measurements when available
from January to December 2012. Forecasts for the time giving the
best agreement are enlighted in red. The best linear fit, as well as the value of the Pearson Correlation Coefficient
are given in the graphs. The graphs are shown for 1)all data, 2)clear time only, 3)summer only and 4)winter only.
|
All Data & 12UT
Clear time only
Summer all & 06UT
Winter all & 12UT
|
GFS 24h forecasts compared to local measurements at the Paranal radiometer
Same as above but for PWV forecasts at step 24h from the GFS model.
In parenthesis is shown for comparison the correlation coefficient without prediction, i.e. assuming PWV equal to its value at the time of analysis (persistence).
The graphs are shown for 1)all data, 2)all data with kalman correction (trained on the 14 previous days), 3)clear time only, 4)summer only, 5)winter only.
|
All Data & 12UT
All data Kalman
Clear time only
Summer all & 06UT
Winter all & 12UT
GFS 48h forecasts compared to local measurements at the Paranal radiometer
Same as above but for PWV forecasts at step 48h from the GFS model.
In parenthesis is shown for comparison the correlation coefficient without prediction, i.e. assuming PWV equal to its value at the time of analysis (persistence).
|
All Data & 12UT
Clear time only
Summer all & 06UT
Winter all & 12UT
GFS 120h forecasts compared to local measurements at the Paranal radiometer
Same as above but for PWV forecasts at step 120h.
In parenthesis is shown for comparison the correlation coefficient without prediction, i.e. assuming PWV equal to its value at the time of analysis (persistence).
|
All Data & 12UT
Clear time only
Summer & 06UT
Winter & 12UT
|