2B.5

**Remote sensing measurements of wind and turbulence profiles for short-term wind power forecasting**

PAPER WITHDRAWN

**Rod Frehlich**, University of Colorado, Boulder, CO

Improvements in short-term wind forecasts can be realized by better upstream measurements of wind and turbulence profiles. This is particularly important for improving the forecasts of wind power generation. The most attractive technology for these measurements is the eye-safe scanning coherent Doppler lidar because the very narrow laser beam permits observations at large distances over height intervals that resolve the relevant boundary layer processes, especially for challenging conditions such as the nocturnal jet. A narrow beam Doppler radar would also satisfy some of the measurement requirements when sufficient signal power is available. Both of these remote sensing techniques can improve short-term forecasting because of the improved spatial averaging which results in more accurate radial velocity measurements compared with fixed observations such as towers and surface met stations.

Improved forecasts also require a definition of observation error that is consistent with the numerics of the numerical weather prediction model. A perfect measurement or "truth" is defined as the spatial average of the atmospheric velocity field based on the spatial filter of the numerical weather prediction model. This produces two components to the observation error: an instrument error and an observation sampling error that describes the mismatch between the observation sampling volume and the spatial filter of the model. Rigorous calculations of both the instrument error and the observation sampling error depend on the scanning geometry, the system operating parameters, and the spatial statistics of the wind field, i.e., the local turbulence. The scanning Doppler lidar also provides the most reliable estimates of the spatial correlation statistics of the wind field after applying corrections for the spatial filtering by the lidar pulse and corrections for the velocity estimation error. The accuracy of the turbulence estimates is also improved compared with single point measurements such as instrumented towers, sodars, profilers because of the larger sampling volume with increases the number of independent samples of the turbulent processes. Results are presented for the predicted instrument error of a scanning Doppler lidar for typical atmospheric turbulence conditions determined from lidar-derived estimates of the spatial statistics of the velocity. The improvements in observation sampling error and critical unresolved issues for improving the short-term forecasts of wind power are also discussed.

Session 2B, Observations and Modeling Related to Renewable Energy Applications II

**Tuesday, 3 August 2010, 1:30 PM-3:00 PM**, Torrey's Peak III & IV** Previous paper Next paper
**