Validation with actual measurements is essential in any model-based mapping. Globally, wind potential maps are being prepared using the numerical downscaling techniques from atmospheric models. In general, the following three inputs are used for downscaling and mapping.
- Assimilated re-analysis data sets – which are gridded data, created through assimilating data collected from different sources at various time scale
- Generalized elevation data
- Generalized land use land cover data
Due to the nature of inputs and methodology of prediction, the results generated using atmospheric models is always interpreted with caution. Due to the accuracy level of inputs and uncertainties in the downscaling techniques, the modeled results may not predict localized wind variations accurately. Although, this issue cannot be fully resolved but the error level of the modeled map can be quantified with the help of actual measurements.
In this study, physical measurements from 406 stations were used (after theremoval of outliers) to validate the map result. The wind speeds at 120m was arrived at through measurement or derived with the help of power law index. Based on the validation study, the mean absolute percentage error of the map is found to be 5%. The scatter plot showing the comparison between actual measurements and model wind speeds is shown in figure below.
