Maximizing the value of short-term observational data using numerical weather prediction models
The mistake is often made of collecting measurement data over a short period (1-18 months), and then assuming the data represent mean conditions. In truth, measured on-site data are merely a random snapshot without any long-term historical context. Assuming that short-term, on-site observational data are representative of long-term mean conditions have significant financial consequences as short-term records can easily vary by +/-10% of the long-term mean.
Incorporating short-term, on-site observational data with NWP model output provides increased certainty when estimating long-term wind resource variability. NWP models can simulate past periods of time to create a complete climatology that places on-site observational data in the context of a long-term record. On-site data are integral to the process and are used to statistically correct raw model data and to generate uncertainty estimates. Unlike traditional Measure-Correlate-Predict (MCP) analysis, NWP models do not require off-site reference station data; thus, NWP-based methodologies are extremely valuable at locations where suitable off-site data for MCP analysis are unavailable.
Since the NWP methodology can be applied using only a few months of on-site data, developers can understand the long-term variability early on and confidently sort potential sites ensuring resources are spent on only truly viable projects. Furthermore, this methodology can be configured into a monthly-recurring, automated system further increasing the efficiency and decreasing costs. Blending the best of direct on-site measurements with advanced NWP techniques provides a reliable, efficient, and scalable methodology for understanding the long-term wind resource variability.