The results show that with assimilation of the UAS observations, marked improvement in RMSE and bias statistics occurs near the top of the marine atmospheric boundary layer and just above, using an independent set of radiosonde observations. Improvements to temperature and moisture profiles in the presence of strong inversions yields improved modified refractivity profiles and, thus, more accurate diagnosis of areas of positive and negative refractivity. These broad improvements to model forecasts are particularly strong in forecast lead times out to 6 hours, with positive influence extending to 12 hours and some evidence of continued influence at 24 hours.
Additional studies on the impact of UAS observations on forecasts of modified refractivity involved utilization of the COAMPS adjoint and tangent linear models. The adjoint and tangent linear models can predict regions of optimal perturbation where the sensitivity to initial conditions of the modified refractivity vertical gradient in a pre-defined space is greatest when modified refractivity is used as the response function. Results demonstrate the potential utility of targeted observations in mitigating model forecast error of the presence and characteristics of EM ducts. The adjoint model is also used to quantify the impact of UAS data assimilation on modified refractivity profile forecasts versus other assimilated observation datasets. Results indicate that a majority of model forecast error reduction at 12 and 24 hour forecast lead time can be attributed to the assimilation of UAS observations.