Utilising Multiple Grid Points in Numerical Weather Prediction System Forecasts
data from each NWP time step over the forecast horizon at a single grid point (or interpolation of grid points) that represents the location of interest. Due to uncertainty in forecasting rapid changes in wind power, see above, forecast users may benefit from multiple scenario forecasting techniques that better characterise uncertainty than a single forecast by producing chronologically consistent scenarios.
A common weather-forecasting approach used to provide multiple scenarios is an ensemble of NWP forecasts that accounts for at least some uncertainty in the NWP-modelling process. The NWP ensemble members may be based on slightly different initial conditions or use different physical assumptions about atmospheric behaviour. NWP ensembles are more computationally expensive than single NWP forecasts at the same spatial resolution. Consequently, spatial resolution is usually compromised in NWP ensemble forecasting (the ECMWF global ensemble model, for example, has a 32km resolution instead of 16km for their single deterministic model) and may cost more to purchase. This article presents a different method for producing multiple scenarios utilising multiple grid points in a single NWP forecast.
The Multiple Grid-point Method
As described above, the conventional method for using forecast information from a NWP output data set is to extract the data at a single NWP grid point for each time step. Due to possible misplacement error, this method may miss useful information at nearby grid points in the NWP data set at a particular time step. Additionally, the single grid-point forecast may miss further information due to temporal aliasing, since the NWP data set is only available at particular time steps, such as hourly.
Utilising data at multiple NWP grid points at each time step may provide more information on the possible behaviour of wind farm power output. However, this information cannot be used directly, as the wind-speed forecasts at each grid point are influenced by the way terrain local to that grid point is modelled in the NWP system. For example, wind speeds near the Earth’s surface over the ocean tend to be stronger than over land because the surface of the land is rougher than the ocean. Surface roughness effects are represented in the NWP model, so if a given
For all industry participants, an uncertain, large rapid change in wind power is likely to have significant but differing impacts on decision-making and its outcomes.
weather feature is over the ocean, higher near-surface wind speeds will be predicted than if the same weather feature was over the land. For this reason, the authors have developed a ‘terrain standardisation’ method to establish relationships between predicted wind speeds at neighbouring grid points. This method uses a NWP data set that covers a long period of time (one year or more) to establish wind-speed relationships.9
Figure 1: An Example of (A) Raw and (B) Standardised Wind Speed Field A
25
10 15 20 25
5 0
W 138 S 139 N E 140 Longitude (º) B
10 15 20 25
5 0
W 138 S 139 N E 140 Longitude (º) Sea 141 -39 142 143 -40 -38 Latitude (º) Land -36 -37 Sea 141 -39 142 143 -40 -38 Latitude (º) Land -36 -37
10 12 14 16 18 20 22
4 6 8
ms-1 25
10 12 14 16 18 20 22
4 6 8
ms-1
(A) Raw and (B) standardised wind speed fields for a European Centre for Medium-Range Weather Forecasts (ECMWF) forecast with projection time 15 hours ahead. Each plot (A) and (B) shows the wind speed fields in a 3D plot with the wind direction field shown on a 2D plot beneath it. The wind turbine symbol indicates the modelled location of the targeted wind farm, which is located on the south coast of Australia.
The standardised wind speed field can then be transformed (based on past observations) to wind power to produce a ‘site-equivalent wind-power forecast field’ using a transformation from wind speed to wind farm power output that is appropriate for the targeted farm. The wind-power values are site-equivalent because of the terrain standardisation, meaning that all wind-power values in the field are directly applicable to the targeted wind farm. Successive wind-power fields can be used to estimate the speed and direction of propagation of wind features using pattern- recognition techniques (algorithm not shown).
The result
is a horizontal grid of standardised wind speeds, where the wind farm of interest (or cluster of wind farms in close proximity) is located in the middle. A graphical example of the raw wind speeds and standardised wind speeds is shown in Figure 1 for a wind farm in Australia, using forecasts from the ECMWF global model.
MODERN ENERGY REVIEW – VOLUME 2 ISSUE 2
For the same example as in Figure 1, the wind-power forecast field is shown in Figure 2, with several wind turbine symbols to indicate alternative scenario forecasts illustrating potential misplacement errors. Displaying the site-equivalent wind-power forecast field, such as in Figure 2, provides a visual tool to characterise uncertainty in the NWP system forecast. For example, the white, uppermost wind turbine symbol, representing the actual location of a wind farm, is situated at a point of high wind power relative to the surrounding wind-power field. The four grey wind turbine symbols on the wind-power field represent possible misplacement errors of the wind feature forecast by the NWP in terms of the wind farm. They highlight that even a small misplacement error in any direction would result in a lower wind farm power. Thus it may be inferred that the predicted wind power value at the actual location of the wind farm has a low probability of being
45
Wind speed (ms-1
)
Wind speed (ms-1
)
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84 |
Page 85 |
Page 86 |
Page 87 |
Page 88 |
Page 89 |
Page 90 |
Page 91 |
Page 92 |
Page 93 |
Page 94 |
Page 95 |
Page 96 |
Page 97 |
Page 98 |
Page 99 |
Page 100 |
Page 101 |
Page 102 |
Page 103 |
Page 104 |
Page 105 |
Page 106 |
Page 107 |
Page 108 |
Page 109 |
Page 110 |
Page 111 |
Page 112 |
Page 113 |
Page 114 |
Page 115 |
Page 116 |
Page 117 |
Page 118 |
Page 119 |
Page 120 |
Page 121 |
Page 122 |
Page 123 |
Page 124