Jersey weather radar, investigation of propagation conditions

Home > Report on Channel Islands UAPs > Appendices | Traduction française

The clean nature of the beam-0 images was noted, specifically the absence of sea return. Some weather radars (e.g. the NEXRAD WSR-88D) operate with reduced sensitivity in certain modes, but inquiries confirmed that there is no other scan mode available in this case, the sensitivity being comparable in practice to the NEXRAD clear-air mode. Clutter removal is understood to be limited to the synthetic beam-S images. The fact that much of the beam-0 echo clearly corresponds to echo from island and coastal features n1 On each of the scans the echo appears to be displaced by several kilometres in relation to the geographical map overlay. The reason for this is unknown, but a similar displacement (though even larger) is noted by Rico-Ramirez, Cluckie & Shepherd (next note) in studies using Jersey weather radar images. No attempt has been made to correct this., the very features that would be on any permanent clutter map, seems to confirm this. So we think we are seeing all the data there are, without software manipulation, and that the images represent the actual clutter conditions of interest for our investigation n2 If we were seeing higher-elevation data that have been inserted to mitigate clutter in permanent clutter areas then we would expect to find a closely similar echo distribution on a higher elevation cut. But droplets or ice crystals in suspension aloft and yet closely matching the surface clutter distribution would be meteorologically un likely, and in fact none of the higher cuts shows the same pattern or intensity of echo..

A minimal amount of sea return on the beam-0 cut might be consistent with the absence of severe radar super-refractivity. Variable weather factors determine wave slope and orientation and thereby the intensity of sea return. We learned from Tony Pallot that sea return is rarely very significant on the Jersey radar, perhaps indicating typically small wave heights and swell amplitudes due to the sheltering effect of the Brittany peninsula and a short southerly wind fetch.

To check conditions on the day, records of the local sea state were sought. The Channel Islands Shipping Forecast issued by Jersey Met Office at noon, 23 April 2007 (Appendix C) gives the sea state as "smooth or slight" with no significant swell. This corresponds to World Meteorological Organisation code 2 or 3, defined as "smooth (wavelets) 0.1 to 0.5 m" or "slight 0.5 to 1.25 m" (codes 0 & 1 are "calm, glassy" and "calm, rippled").

Table 1. Measured wave parameters off Corbiere, Jersey
Time (Z)
Sig. wave (m) n3Significant wave height is the mean height of highest 1/3 of waves Period (s) Max. wave (m)
1300 0.67 6.30 1.00
0.70 6.70 1.28
1500 0.69 6.90 1.34

This forecast was confirmed by values from the Jersey Fisheries wave meter, a waverider buoy situated approximately 6 miles off Corbiere, SW Jersey. As shown in Table 1, the maximum wave height at 1400Z was measured at 1.28m with a period of 6.7 sec. This corresponds to a very shallow wave slope in the order of only one or two degrees and very little radar backscatter.

So absence of sea return is not probative evidence. However the absence of any return from the island of Alderney offers a possible independent test. We know that many parts of nearby coasts and islands are detectable above the radar horizon on beam-0, so two avenues were explored to determine the significance of the absence of echoes returned by Alderney. Evidence was sought regarding the past frequency of detection of echoes from Alderney, and the local radar horizon in standard propagation conditions n4 Propagation of radar waves is determined by the vertical structure of the atmosphere. The typical pressure, temperature and humidity gradients of the "standard" atmosphere result in a downward refraction. This has the effect of making the distance to the radar horizon greater than the normal optical horizon. When the vertical pressure, temperature and humidity gradients are non-standard, effects analogous to optical mirage can occur. Super-refraction (greater than normal downward bending) expands the radar horizon still further; sub-refraction (upward bending) contracts the radar horizon. was compared with the range and elevation AMSL of the island.

The typical frequency of detectability of Alderney was estimated from two sources: Anecdotal evidence from experienced users and a published study of accumulated rainfall measurements using the Jersey weather radar.

Tony Pallot was not surprised by the absence of echo from Alderney, suggesting that its low elevation was probably below the radar horizon. According to his recollection Alderney was almost never detected, even in strongly anticyclonic conditions when AP due to severe temperature inversion "occasionally" shows Guernsey and the French mainland. However the beam-0 scans on the subject date (and to a lesser extent higher elevation scans as well) clearly do show echo from Guernsey, and from the French mainland as well, even though the meteorological conditions appear not to suggest severe AP conditions on April 23 (Section 5). Indeed Tony Pallot's own opinion was that negligible radar AP was likely. So this merited further investigation.

The height AMSL of the Jersey weather radar was determined to be 84.2m (276 ft) at the antenna boresight. The horizon distance d to which a sea-level reflector would be detectable by centimetric radar in normal propagation conditions, neglecting topographical masking, is given approximately by

d(nmi) =1.23h(ft)

or d = 20nmi. Alderney is ~30nmi from Jersey. However at that distance any reflector higher than about 20m (65 ft) would be above the radar horizon. Alderney is 88m (289 ft) elevation AMSL at the airport runway, and the horizon distance for a reflector at this height in the same conditions is ~41nmi, significantly greater than the 30nmi range of Alderney. So one would expect echoes from Alderney except in sub-refractive propagation conditions, unless at very low beam angles there is physical masking of the radar in the direction of Alderney due to local topography.

Fig.1 Monthly accumulated echo on Jersey weather radar (beam-0, 0.5° elevation) during Feb-May 2004 (after Rico-Ramirez et al.) Fig.1 Monthly accumulated echo on Jersey weather radar (beam-0, 0.5° elevation) during Feb-May 2004 (after Rico-Ramirez et al.)

To clarify this matter, images of the accumulated echo from beam-0 scans of the same radar during a comparable seasonal period in 2004 were examined. In a four month study by Rico-Ramirez et al. of precipitation measurement accuracy using the Jersey weather radar, composite images were compiled of the aggregate echo detected in each of the four elevation scans during the months of February, March, April and May 2004 n5Miguel Angel Rico-Ramirez, Ian Cluckie & Geoff Shepherd, Jersey Radar Experiment; Interim Report', Water and Environmental Management Research Centre, Dept of Civil Engineering, University of Bristol, April 2005..

On these accumulated monthly scans (see Fig.1) a NE masking sector in the sea return is very evident. But this sea return is in a part of the beam radiating at small negative elevations just beyond the half-power (3dB-down) points of the 1.0° beam n6 The range of the outer edge of this arc of return is quite close to the theoretical sea level horizon distance of about 20nmi [38km] expected in typical propagation conditions, and so corresponds roughly to rays launched at 0° ## elevation. This coincides with the -3dB level for a boresight elevation of 0.5°. The inner edge is perhaps 0.9° #from the boresight. The topographic masking in this part of the beam is not necessarily indicative of raypaths at the main beam boresight elevation of +0.5°. And in fact we do see echo from Alderney on the beam-0 scans.

Fig.2 Frequency of radar refractive index gradients (N-units/km) in the Channel Islands area, calculated from radiosonde readings from March 2004 - September 2005. Note the increasing trend through the period (highlighted) of the Rico-Ramirez study. (After Gunashekar et al.) Fig.2 Frequency of radar refractive index gradients (N-units/km) in the Channel Islands area, calculated from radiosonde readings from March 2004 - September 2005. Note the increasing trend through the period (highlighted) of the Rico-Ramirez study. (After Gunashekar et al.)

The possibility that ground clutter has been subtracted and that this echo is very intense local precipitation echo seems small. (The above authors allude to a possible clutter map used by Jersey Met, but this seems to refer to the map used to insert higher-elevation data in the synthetic S beam product referred to above.) Many echoes on the beam-0 scan are clearly clutter from coasts and islands and are identified as such by Rico-Ramirez et al., so there is no evidence that a permanent clutter map has been subtracted or substituted in these images. The strong echo accumulated at Alderney, especially in the April and May beam-0 scans, appears therefore to be ground echo n7 The Jersey ground clutter pattern is further discussed by Rico-Ramirez, Cluckie, Shepherd & Pallot, "A High Resolution Radar Experiment on the Island of Jersey", Journal of Meteorological Applications 14: 117-129 (2007). DOI: 10.1002/met.13.

In support of this we can observe how the echo pattern changes. Generally, the intensity of the local arc of sea return progressively diminishes (relative to the general background) from February to May. This suggests decreasing average wave slope (linearly proportional to a decreasing average wind speed). At the same time distant land clutter and echo from surface vessels n8 Accumulated echo along Ferry routes from St Malo on the Brittany coast is evident, as also is a broader arc of strong echo running up the Channel roughly ESE-WNW to the north of Alderney, which is almost certainly a busy merchant sea lane on which hundreds of large container vessels pass every day. intensifies. During April the Alderney echo first becomes clear; in May it is strong. The same trend of intensification is apparent in the echo from Guernsey, suggesting an increasing seasonal frequency of super-refractivity (or decreasing sub-refractivity) causing an expansion of the radar horizon. On the other hand no parallel trend appears in the Jersey mean rainfall figures which for Feb-May 2004 were 37.1, 35.8, 75.4 and 29.9mm n9 Jersey Met Office climate records.

Finally, the frequency of radar ducting events in the Channel Islands area during the same 2004 period was estimated in a separate study by Gunashekar et al n10 S.D.Gunashekar, E.M. Warrington, D.R. Siddle and P. Valtr , Signal strength variations at 2 GHz for three sea paths in the British Channel Islands: detailed discussion and propagation modelling', Radio Sci., 42, RS4020.. A graph of refractive index gradients in N-units/km calculated from high-resolution radiosonde readings (Fig.14) shows a steeply increasing trend through the period March - June 2004, as inferred above, and this trend is paralleled by directly measured variations in 2GHz signal strength over the path between Jersey and Alderney, as well as by modified surface refractivity inversions between the heights of Guernsey and Alderney airports.

In short, the aggregated 2004 observations provide convincing evidence that the strength of ground clutter echo from Alderney was positively correlated with an increasing parabolic curvature of the radar ray paths in conditions of increasing average super-refractivity.

On 23 April 2007 we have evidence of two different inversion regimes, a severe one in the south near the Breton coast and another, much weaker, in the Channel Islands area. Considering the latter, we expect a small surface inversion of perhaps 2°- 3°C/kft in the area. This would by ruleof-thumb be expected to contribute a couple of N-units of refractive index n11 An N-unit is one part per million of refractive index, i.e., N = (n -1)106 = 350 where n = 1.00035. Around 350 N-units is a typical refractivity for centimetric radar at sea level, but it can vary from about 250 to 450 N-units. . And the semipermanent evaporation duct produces as a matter of course a super-refractive (for radio waves only) humidity lapse through the lowest few tens of metres. We might therefore expect to see radar evidence of a small degree of super-refractivity on low elevation cuts, but certainly not much.

So the fact that Alderney does not appear in the radar picture on 23 April can be interpreted as evidence that the latter does not indicate severe radar super-refractivity in the north Channel Islands area, consistent with expectation (from the horizon calculations), with the non-radar meteorological data (lack of evidence for significant temperature inversion in the North Channel Islands area) and with the professional opinions solicited.

But radar evidence of propagation conditions in the south over the French terrain on 23 April 2007 is more ambiguous. There is some variation in the ground clutter intensity during the 30 minutes covered by Figs. 15-17 in Section 5, and this can be interpreted in terms of varying refractivity. An observed diminution in the clutter pattern seems to indicate decreasing earthward bending of raypaths (i.e. a transition in the direction of less super-refractive or more subrefractive conditions). Comparing this clutter pattern with the 2004 images in Fig.1 might suggest that it is somewhat more intense than would be expected in conditions that give rise to the observed Channel Islands echo pattern, since the latter are monthly accumulations of echo that presumably include (unlike the April 23 images) at least some low-level precipitation over the Breton hills. But the comparison is very subjective.

Fig.3 Effects of seven radar refraction values from -7"/km to +40"/km simulated over a digital elevation model, compared with actual clutter patterns (lower right) observed by Jersey Weather Radar samedi 23 avril 2007 Fig.3 Effects of seven radar refraction values from -7

We were interested to see if it would be possible to make a more controllable and quantifiable test of the radio propagation conditions by comparing the observed echo pattern with that predicted by a computer simulation. Our hope also was that the result could be extrapolated to infer limits on possible refractivity at optical wavelengths. If so it might be possible to supplement meteorological evidence by means of indirect observational evidence, with a bearing on the possibility of optical mirage and related theories discussed in Section 6.d.

Fig.4. Refractivity of visible light (633 nm) as a function of RH for three temperatures at constant atmospheric pressure (Ciddor equation) showing negligible dependency. Fig.4. Refractivity of visible light (633 nm) as a function of RH for three temperatures at constant atmospheric pressure (Ciddor equation) showing negligible dependency.

With this in mind we proceeded to design a computer ray-tracing simulation of the Jersey weather radar coverage pattern for various refractivity values n12The simulation implements the basic radar equation for average received power with an attenuation correction factor for plane (area) targets (i.e. the Earth's surface), the solution being computed for each 90m surface element of the DEM being impacted by a ray. A backscattering algorithm checks where the ray impacts the surface of the DEM and applies the relevant (calm sea/land texture) backscattering profile. For the visualization of the clutter maps we used geocontours included in the Jersey weather radar BUFR decoding package (thanks to Miguel Angel Rico-Ramirez, Bristol U.), rescaled, recentred and rotated by 2° anticlockwise to fit the DEM, with clutter maps downgraded from the 90m resolution to match the 2 km resolution of the Jersey radar. The results (Fig.3) show the interplay between the antenna gain diagram (vertical polar profile, boresight elevation 0.5°), distance attenuation, and backscattering efficiency, for different refraction strengths.. The simulation was initially run over a digital elevation model (DEM) of the coverage area at 90m resolution, later coarsened to match the 2km pixel resolution of the Jersey weather radar images. The refraction values chosen were -7"/km, 0"/km, +7"/km, +15"/km, +22"/km, +33"/km and +40"/km (positive values indicating bending towards the earth). Initial results for a 0.5 degree ray tracing (Fig.3) suggest that the best fit is a refraction of +22"/km, reproducing the 23 April images rather well within the limits of what is necessarily a very approximate simulation in terms both of terrain reflectivity and propagation. This amount of ray refraction, 11"/km less than the earth curvature of 33"/km, is not far from the refraction assumed in the 4/3 earth' approximation used to model propagation in a standard atmosphere.

Clearly we should not expect to be able to capture complex and dynamic propagation conditions very effectively in such a crude simulation. But it is still reasonable to be a little surprised at the result. Given evidence of a significant advection inversion close to the Breton shore during the sighting period, with a marginally-ducting gradient of ~10°C/kft based on the Meteo-France numerical simulation and other circumstantial evidence, ought not the low-beam clutter pattern on mainland Brittany to show signs of stronger refractivity than a fairly bland 22"/km?

One possible explanation is that whereas optical refractivity is almost totally insensitive to moisture (Fig. 4), the same is not true of radar. In fact variation in humidity contributes far more to the radar refractive index than does variation of temperature.

A semi-permanent feature of the marine radio environment in almost all conditions is the well studied evaporation duct caused by a humidity lapse over the sea n13 Gunashekar et al., op. cit.. The effective duct height itself may only be a few tens of metres from the point of view of radio propagation. The question is whether the humidity lapse continues through the marine surface boundary layer (which we take to be in this case the region below the haze discontinuity at about 2000 ft associated with the continental dry air intrusion; see Section 5), or whether there might be an increase in moisture with height. Radar sub-refraction might occur if there a rising humidity gradient, and thus the radio effects of a small or even moderate temperature inversion could in principle be negated by a humidity "inversion".

Evidence from the Brest radiosonde shows no such gradient, rather there is a somewhat dry boundary layer overlain by the much drier layer already mentioned. The Trappes profile, far to the E and much further from the sea, does show some increase in humidity with height. However our understanding (Section 5) is that Brest, an essentially marine environment within the warm sector between fronts and in the same SSW low level airflow, is more representative for our purposes.

The noon Brest radiosonde ascent shows RH at 52% at the surface and <40% through the first 3000ft, which is quite dry, falling to an unusually dry 10% at about 2000ft, perhaps indicating that hygroscopic haze aerosols at this altitude - the same altitude as the reported Channel Islands haze layer - are drying the air. Such a haze is basically composed of airborne particulates whose optical cross-section is dependent on moisture but does not generally indicate RH at saturation or above.

Hygroscopic salt particles in a salt sea haze begin to swell at about 70% RH, but dusts and biological aerosols such as pollen will react to a lower RH than this. The latter is the type of "continental bad air" haze indicated ("not a salt haze") by observers, so a moderate optical thickness in this case indicates removal of moisture from air that is probably well below saturation in the first place. At the sighting time, Guernsey surface RH was recorded at 59% (17°T, 9°D), and Alderney surface RH at 77% (14°T, 10°D). The mean of these values is 68%, only a little below the 22-year historical April average for Guernsey n14 of 73%. The condensation level (cloud base) is close to the freezing level at 10,000ft.

So, tentatively, we would say there is no sign of humidity increasing unusually with height and therefore (though this is far from conclusive) no evidence of a subrefractive humidity gradient that might mask the effects of an expected super-refractive temperature gradient. It is therefore a test of our met model to see if it can explain the unexpected distribution of ground clutter in another way.

We think that this may be due to the shallowness of the coastal radio duct indicated by the Meteo-France ALADIN simulation combined with the Breton topography. In this model the temperature inversion gives way sharply to an overlying layer with a slightly excessive temperature lapse rate. So the duct traps rays launched at near 0° elevation and at very small negative angles (including some rays scattered at grazing incidence from the surface of the sea) and guides these over the geometrical horizon towards the land. Rays launched at small positive angles just too steep to couple into the duct either continue freely and do not refract earthward at all, or else intercept the top of the duct at a grazing angle and may be scattered by partial reflection back into the sky. Thus the duct acts to introduce a height cut-off in transmitter and receiver gain and tends to reject rays that impact the terrain at altitudes above the top of the duct.

The explanation of the observed clutter pattern could therefore be that the duct is enhancing echo strength returned from the terrain, but by the same token is restricting the area from which ground echo is receivable. The result is simultaneously to intensify and to contract the clutter pattern, favouring lower terrain, which in N Brittany means that clutter would tend to concentrate towards the coast and be minimised from higher ground inland. Thus a super-refractive surface duct which in theory expands the radar horizon produces, because it is capped below the maximum topography, an effect which resembles the contraction of the radar horizon due to subrefractive conditions.

In the ALADIN model the top of the inversion would be at about 200m ASL. From examining the Breton topography it is our impression that, if this explanation is correct, the effective top of the radio duct may have been somewhat below the 200m contour.

The above evidence and interpretation appears to confirm, or at least increases our confidence in, the meteorological picture developed in Section 5.

Home > Report on Channel Islands UAPs > Appendices | Traduction française