Another possible cause of an elevated inversion near the haze layer is where warm continental air is advected over a relatively cool and relatively deep marine layer in such a way that the marine layer remains undisturbed. A sharp elevated gradient might occur at the boundary between the two air masses.
But there is evidence that rather than remaining cool through the first 2000ft the air is warming, creating the low-level marine temperature inversion already discussed. We were advised by Tony Pallot that advection of still warmer air above this level was not a plausible mechanism on the day in question. Balloon ascent data suggest that there might be advected drier air from the continent above 2000ft n1The visibility above the 2000ft haze layer was said to be good, to 100nmi. Andy Young, an atmospheric scientist at San Diego State University, points out to us that this is an indicator of dry air. The optical cross-section of haze particles has a strong dependence on relative humidity. This is the main reason why a haze layer may often be diagnostic of an inversion, since there is a large RH jump associated with the temperature jump at the top of the layer (email to Martin Shough, 01.09.07)., but not significantly warmer. At this time of year the effect of surface temperature on air at this level, as opposed to the surface, is minimal (about +1° at most) n2Email from Tony Pallot to Martin Shough, 04.09.07..
This opinion is corroborated by the Meteo-France numerical simulation, which predicts a significant local advection inversion at low level over Breton coastal waters south of the Channel Islands, steepest through the first 165 ft (~50m), but no features of note above about 1000mbar (~630ft, 200m).
We attempted to further test these expert theoretical models against real-world observations by investigating the performance of the Jersey Airport Meteorological Department weather radar during the sighting time. Our reasoning was that an optical duct would potentially be a stronger duct at radar wavelengths, and a map of the ground clutter reflectivity could be regarded as a direct sampling of the microwave propagation environment in the sighting area.
We adopted several approaches. Firstly we wished to simulate the C-band weather radar picture by computer raytracing. By varying the effective ray curvature over a digital elevation model of the area we attempted to reproduce the ground clutter pattern and from this infer the radar refractivity at low elevation. On the basis of this we then proposed to use the well-known temperature/humidity refractivity relations to plot optical raypaths as a function of those in the centimetre region. In parallel the observed echo from the islands during the sighting period was compared with published historical observation data from the same radar and local expert advice was sought.
The best overall fit to the observed clutter pattern in the Channel Islands area was obtained with a ray curvature of +22" per km horizontal distance, a slightly super-refractive curvature but far short of radar trapping (22"/km corresponds to a refractive index gradient of approximately -32 N-units per 1000ft. Normal propagation in the standard atmosphere is taken to lie between 0 and -24 N-units per 1000ft; trapping occurs with gradients larger than -48 N-units per 1000ft). The height ASL of the Jersey C-band antenna would lie within the thickness of the expected Breton coastal optical duct, but it would be sited between the strong ducting region to the south and the region of weaker marine inversion in the vicinity of Guernsey and Alderney to the north. We would therefore expect that low elevation radar raypaths entering the duct could propagate with especially enhanced efficiency to the south, returning echo from coastal topography of Brittany with unusual strength up to about the 200m contour. This fits the radar evidence indicating that super-refractivity was most noticeable and most variable to the south, and declining over time, suggesting a weakening duct, whilst echo from the islands to the north shows less variability and neither Alderney nor the nearby northern coastal hills of the Cherbourg peninsula return any echo at all n3There is evidence of probable topographical masking to the NE of the radar, but historical data from 2004 (see Appendix D) prove that these areas do contribute significant clutter in super-refractive propagation conditions. .
So we found radar evidence qualitatively consistent with the general atmospheric situation described earlier, although reliable inferences about the strength of optical refraction were unfortunately not possible, mainly because of uncertainty about the vertical humidity profile, a variable on which radar refractivity is powerfully dependent. (A fuller account of these investigations is given in Appendix D).