 |
Accuracy of Wind Map Estimates
The accuracy of the MesoMap system has been tested extensively against reliable surface wind measurements in several parts of the world. The following two tables provide the results of two such comparisons at grid scales of 400 m and 1 km, respectively, in two types of terrain—moderate (rolling) and complex (mountainous). The region is the northeastern United States.
| Moderate Terrain - 400 m Grid (New York) |
|
| Site |
True Elev (m) |
Model Elev (m) |
Projected/
Measured
Speed (m/s) |
Simulated
Speed (m/s) |
Bias (m/s) |
Adjusted
Bias (m/s) |
|
| 1 |
18 |
26 |
6.8 |
7.0 |
0.2 |
0.1 |
| 2 |
34 |
30 |
7.9 |
8.3 |
0.4 |
0.4 |
| 3 |
101 |
90 |
6.7 |
6.5 |
-0.2 |
0.0 |
| 4 |
543 |
492 |
7.3 |
6.8 |
-0.5 |
0.1 |
| 5 |
631 |
610 |
7.3 |
7.5 |
0.2 |
0.5 |
| 6 |
564 |
548 |
7.1 |
7.1 |
0.0 |
0.2 |
|
| AVERAGE |
7.2 |
7.2 |
0.0 (0%) |
0.2 (+2%) |
| STD DEVIATION |
|
|
0.3 (4%) |
0.2 (2%) |
| Complex Terrain - 1 km Grid (NH/VT) |
|
| Site |
True Elev (m) |
Model Elev (m) |
Projected/
Measured
Speed (m/s) |
Simulated
Speed (m/s) |
Bias (m/s) |
Adjusted
Bias (m/s) |
|
| Berlin |
85 |
534 |
5.7 |
5.7 |
0.0 |
0.2 |
| Colebrook |
630 |
584 |
5.5 |
5.5 |
0.0 |
0.2 |
| Mt. Sunapee |
830 |
746 |
10.3 |
9.4 |
-0.9 |
-0.4 |
| Walker Mtn. |
480 |
352 |
5.7 |
4.9 |
-0.8 |
0.3 |
| Grandpa's Knob |
620 |
420 |
7.6 |
5.3 |
-2.3 |
0.3 |
| Mt. Mansfield |
1280 |
1040 |
12.1 |
8.4 |
-3.7 |
0.0 |
|
| AVERAGE |
7.3 |
6.0 |
-1.3 (-18%) |
0.1 (1%) |
| STD DEVIATION |
|
|
1.5 (20%) |
0.3 (4%) |
The "projected/measured" mean speeds have been independently extrapolated from the measurement height (40 m) to the assumed wind turbine hub height (65 m) and adjusted to the long-term norm using correlations with nearby airport stations. In moderate terrain (first table), the accuracy of the model is very high, with a mean error of virtually zero and standard deviation of just 4%. In complex terrain (second table), the model tends to underestimate winds on mountaintops where the site elevation exceeds that of the corresponding grid cell in the model data base. This is an example of an error caused by a sub-grid-scale effect whereby the average elevation of the grid cell is less than the elevation where the wind measurement was taken. The "adjusted bias" attempts to isolate and eliminate the effect of sub-grid-scale elevation differences. The adjustment is calculated by fitting the bias to the elevation discrepancy (model elevation minus true elevation). The r2 "goodness of fit" of these regressions is 58% in moderate terrain and 96% in complex terrain, indicating that sub-grid-scale variations play an important role in explaining the discrepancies. The graph below shows the relationship in complex terrain.
Once the sub-grid scale terrain variations are accounted for, the MesoMap system is in very close agreement with the observations in both types of terrain. The average adjusted bias over all stations is 1-2% of the projected/measured speed, while the standard deviation of the bias is 2-4%. These discrepancies are within the measurement error of the meteorological equipment. It is recommended that a plus or minus 5% prediction accuracy be generally assumed for the Wind Map results at most sites.
|
|
