Estimating evaporation duct heights from radar sea echo

Abstract

The evaporation duct is a downward refracting layer that results from the rapid decrease in humidity with respect to altitude occurring in the atmospheric surface layer above bodies of water. The evaporation duct affects radar detection ranges at frequencies of approximately 1 GHz and above. Models based on Monin‐Obukhov similarity theory are usually used to calculate evaporation duct refractivity profiles from bulk measurements of air temperature, humidity, wind speed, and the sea surface temperature. Modeling results by Pappert et al. [1992] indicated that the falloff of radar sea echo as a function of range was an increasing function of the evaporation duct height. On the basis of those results, the authors proposed inferring the evaporation duct height by a slope fit to modeled clutter power, a nonlinear least squares inversion procedure. Data for testing the inversion procedure were obtained using the S band Space Range Radar at Wallops Island, Virginia. Evaporation duct heights were inferred from the radar data on the basis of the assumption of a range‐independent evaporation duct height and sea clutter radar cross section (σ°). Validation data consist of buoy and boat in situ bulk measurements. The result of comparing the radar‐inferred evaporation duct heights and those calculated from bulk measurements indicates that the radar‐inferred duct heights are strongly correlated with those from the in situ measurements, but there is some uncertainty as to whether they are biased or unbiased. That uncertainty arises from the assumed dependence of σ° on the grazing angle ψ. That dependence is currently a matter of debate in the open literature, with the lower and upper ends of modeling results being σ° ∝ ψ0; and σ° ∝ ψ4, respectively. We show results for both dependencies and note that the σ° ∝ ψ0; provides the best agreement with our measurements. It should be noted that inferring the evaporation duct height from radar sea echo is a problem that stresses the modeling of low‐grazing‐angle backscatter.