Reconstruction of high-resolution meteorological fields from airborne observations: a comparison of different techniques

Light aircrafts represent valuable measurement platforms for atmospheric research, for they are able to provide high temporal and spatial resolution observations of the atmosphere. In particular, they are very suitable tools for the observation of atmospheric boundary-layer (ABL) structures in complex terrain, which are typically characterized by a fine-scale, fully 3D variability. Indeed, thanks to their good maneuverability, they allow to perform both vertical profiles and horizontal surveys. Various techniques have been used in the literature to remap data taken along flight trajectories onto regularly spaced, high-resolution 3D grids. In the present contribution the application of a geostatistical interpolation technique called Residual Kriging (RK) is proposed for the mapping of airborne measurements of scalar quantities. In RK the dominant (vertical) drift component underlying the original data is first extracted from the original data to filter out local anomalies, then the residual field is separately interpolated through an Ordinary Kriging algorithm and finally added back to the drift. In RK the determination of interpolation weights relies on the estimate of the characteristic covariance function of the residuals through computation and modeling of their semivariogram function. RK implementation inherently accounts for the characteristic anisotropy of the target field, and it also allows for an estimate of the interpolation error. A dataset from flights of an equipped motorglider, exploring ABL structures in the valleys near the city of Trento (in the southeastern Italian Alps) on fair-weather summer days, is adopted as test-bed database. RK method is used to reconstruct 3D high-resolution fields of potential temperature and water vapor mixing ratio for specific vertical slices of the valley atmosphere, integrating also measurements from the nearest surface weather stations (when present). From RK-interpolated meteorological fields, fine-scale local features of the valley ABL developing in connection with the occurrence of thermally-driven slope and valley winds are detected. RK’s performance is tested against the performances of other mapping methods commonly adopted for this application, i.e. methods based on Inverse Distance weighting or on Delaunay triangulations. A comparative evaluation, carried out by means of cross-validation procedures and in terms of physical plausibility of results, identifies RK as best-performing technique.