To improve the measurements of wind made from NSF/NCAR aircraft, a Kalman filter is developed and applied to archived data files from research projects. The filter is an error-state Kalman filter, and the emphasis is on improving the measurements of pitch and heading because they are usually the dominating source of uncertainty in measured wind. The NSF/NCAR Gulfstream V research aircraft is emphasized in the development, but the filter as developed can be applied to data from other present and past NCAR aircraft as well. So that the resulting filter can be applied in cases where the primary measurements from the inertial reference system were not recorded in the data file, a method is developed for retrieving those measurements by differentiating the recorded variables representing attitude angles and aircraft-velocity components. In addition, some new algorithms are introduced for estimating the rate of climb of the aircraft from the measured accelerations and for estimating the angle of attack from pressures measured at ports on the radome. In addition, simplified methods for estimating the errors in pitch and heading without the full complexity and processing requirements of the Kalman filter are documented. The result is that standard uncertainties in pitch and heading are reduced to about 0.01, so that they no longer dominate the uncertainty in the wind measurements. Some examples illustrate the effects of the Kalman filter on measured wind and on variance spectra. The processing technique is incorporated into an R script that can add the improved variables to a standard netCDF data archive so that they can be made available for community use. Documents that are accessible via links in this technical document provide information on the workflow that generated the document, the details of the processing algorithms, and instructions for use and modification of the processing script.
