Global sensitivity of parabolic equation radar wave propagation simulation to sea state and atmospheric refractivity structure

Abstract

Inadequate representation of the environment is a limitation for prediction of radar system performance as well as for validation of propagation codes. To improve understanding of how different environmental effects/parameters compete and compare, this study examines the sensitivity of radar wave propagation to a suite of environmental parameters for low grazing angle near‐surface radar systems at 3–15 GHz at horizontal and vertical polarizations. A global sensitivity analysis method is used, which accounts for parameter interactions, and propagation is modeled using the parabolic equation method. Environmental parameters examined include eight sea state parameters and eight parameters characterizing the vertical structure and character of range‐independent refractivity profiles. The relative importance of parameters varies more with frequency than polarization, and parameter interactions are found to be significant. Atmospheric mixed layer parameters are found to be the most sensitive, particularly the thickness of the mixed layer. The most significant ocean surface parameter is swell period, although sea directionality is important at 3 GHz and sea surface roughness and salinity are important at 9 and 15 GHz. Because of the spatial variability of sensitivity throughout the domain, regional analysis is performed to determine the most important parameters in different regions of the domain (1000 m in altitude and 60 km in range). These regional sensitivity results, along with those for the whole domain, provide guidance on prioritization of environmental characterization in numerical weather prediction and inversion studies (e.g., refractivity from clutter studies), which are two common methods currently used to address environmental effects on propagation.

Impact of Radar Data Sampling on the Accuracy of Atmospheric Refractivity Inversions Over Marine Surfaces

“…These results are presented in Figure as the median percent error over the trials of each domain and sampling density examined. Notably, the curvature parameter is most accurately recovered with low data densities suggesting a high sensitivity of the propagation to c 0 , consistent with the findings in Lentini and Hackett (). Duct height errors are shown to remain below 20% in data density experiments 10 2 and above for all regions.…”

“…Errors in mixed layer slope are presumably higher than the other parameters because the regions were selected for their sensitivity to the other two parameters, which are more difficult to estimate from bulk environmental measurements. Additionally, the PL in general is less sensitive to the mixed layer slope than the other two parameters at low altitude (<200 m; Lentini & Hackett, ).…”

“…Given the previously demonstrated sensitivity of inversion solution accuracies to measurement strategies, it would be worthwhile to examine the influence of data density and source location on inversion accuracy. The global sensitivity study by Lentini and Hackett () provides some insight into optimal locations to sample RF data, where they examine locations of high sensitivity of PL to various environmental parameters; but this study does not demonstrate directly how sampling from different locations could impact the accuracy of inversions. Variations in PL patterns due to changes in refractivity can be localized, making targeting certain regions advantageous (Lentini & Hackett, ).…”

“…The global sensitivity study by Lentini and Hackett () provides some insight into optimal locations to sample RF data, where they examine locations of high sensitivity of PL to various environmental parameters; but this study does not demonstrate directly how sampling from different locations could impact the accuracy of inversions. Variations in PL patterns due to changes in refractivity can be localized, making targeting certain regions advantageous (Lentini & Hackett, ). In contrast, changes in PL patterns due to variations in sea state tend to be spread over larger areas due to the effects of multipath (Penton & Hackett, ).…”

An Information Entropy-Based Sensitivity Analysis of Radar Sensing of Rough Surface

“…Recently, global-based SA methods, such as extended Fourier amplitude sensitivity test (EFAST) [14][15][16][17][18][19][20][21] and Sobol's method [22][23][24], have begun to consider the probability distribution functions of parameters as prior knowledge in examining changes in the model response. However, these methods often suffer difficulty in illustrating SA variations for specific surface conditions or observational configurations, which is essential in establishing surface sensing algorithms.…”