Introduction
One of the major applications of weather prediction is in the management and maintenance of power transmission lines. The performance of the power distribution network and the utilization of the resources can be considerably improved based on accurate weather forecasts. As an example the amount of electronic current and hence the utilization can be increased in a specific part of the network if the weather is going to be cold and windy in that area.
Numerical Weather Prediction (NWP) models incorporate numerous fluid mechanics and thermodynamics aspects and elements of the atmosphere and earth in order to model and predict the weather behavior on the planet earth. These models provide various parameters of the upcoming weather such as temperature, wind, pressure (as in the figure), humidity and tens of other attributes in the horizon of hours and days.
Currently, as the figure shows the weather features are the only output of the NWP model and there is no measure of uncertainty or accuracy associated with the forecast outputs. This is due to the fact that the forecast are made based on pure physical formulas which are completely deterministic. However, based a variety of sources such as initial conditions, simplifying assumptions and spatial resolution of the model the output is going to be inaccurate and hence has some uncertainty with it.
Currently, as the figure shows the weather features are the only output of the NWP model and there is no measure of uncertainty or accuracy associated with the forecast outputs. This is due to the fact that the forecast are made based on pure physical formulas which are completely deterministic. However, based a variety of sources such as initial conditions, simplifying assumptions and spatial resolution of the model the output is going to be inaccurate and hence has some uncertainty with it.
Objectives
In this research work first we aim to evaluate the prediction accuracy of different outputs of the NWP model based on hind-casting. Second and more significantly we will analyze the uncertainty of the predictions and investigate its roots in the diverse weather features at the moment when the forecast was performed. Hence the results will show how the error margin of the prediction for different weather features is possibly explained by the weather situation such as pressure, wind, humidity or other aspects such as time, location or height of the prediction site.
Consequently, the results of this analysis will provide a clearer understanding of the prediction model and its accuracy. Thus, the power transmission optimization decisions are made not only based on the weather forecasts (e.g. temperature of the air around the power line) but also based on the uncertainty lying in the prediction itself. For example the forecast may be that the temperature is going to be -15 around a specific power line and there is very high uncertainty associated with this based on the pressure or the winds. In this case the power line managers may not risk increasing the amount of voltage of the lines. But if the forecast is -10 and there is very narrow error margin for this, the managers could confidently increase the load of the line. This will finally aid us to achieve a higher performance and reliability in the power grid.
Consequently, the results of this analysis will provide a clearer understanding of the prediction model and its accuracy. Thus, the power transmission optimization decisions are made not only based on the weather forecasts (e.g. temperature of the air around the power line) but also based on the uncertainty lying in the prediction itself. For example the forecast may be that the temperature is going to be -15 around a specific power line and there is very high uncertainty associated with this based on the pressure or the winds. In this case the power line managers may not risk increasing the amount of voltage of the lines. But if the forecast is -10 and there is very narrow error margin for this, the managers could confidently increase the load of the line. This will finally aid us to achieve a higher performance and reliability in the power grid.
