Island wake impact on evaporation duct height and sea clutter in the lee of Kauai

Perturbed flow over and around an island can produce leeside vortices and a long wake region of reduced wind speed and altered thermodynamic structure that impacts the evaporation duct height field and directional wave spectra, both of which impact radar sea clutter returns. In this paper, predicted radar clutter is constructed by using evaporation duct height and wind fields from a mesoscale model along with appropriate sea clutter and electromagnetic propagation models. This predicted radar clutter is compared to shipboard observations of radar clutter taken off the leeward slide of Kauai, in December 1999.


INTRODUCTION
In December of 1999, shipboard observation of radar sea clutter were obtained in the lee of Kauai, HI aboard a U.S. Navy destroyer, the USS O'Kane. The O'Kane was equipped with Lockheed-Martin's TEP (Tactical Environmental Processor). TEP extracts NEXRAD-like weather information from the AN/SPY-1 radar. Substantial azimuthal variability in the radar sea clutter was observed and postulated to result from an island wake. To test this hypothesis, the Naval Research Laboratory's Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS) was run to: (a) examine COAMPS ability to forecast an island wake and its impact upon the wind, thermodynamic, and evaporation duct fields, and (b) investigate the feasibility of assimilating COAMPS forecast fields and radar observations to optimally infer structural refractivity features.
*The analysis presented in this paper was funded under the Remote Refractivity Sensing project that is funded by Code 321 of the Office of Naval Research (ONR). The experimental work was funded by ONR Code 322.

COAMPS FIELDS
The COAMPS model was run with an inner nest horizontal grid interval of 3 km by 3 km using a Louis surface flux parameterization [1]. The domain of the inner nest ran from 20.946°to 23.196°North and 158.46°t o 162.51°West, corresponding to a rectangular grid of 270 km by 450 km surrounding the island of Kauai. Fields obtained from the model included evaporation duct height (6), wind velocity (uW) and wind direction (6). The three contour fields are shown in Figure 1 through evaporation duct heights vary from O to 16 meters and the wind speeds vary from O to 10 m/s depending on the azimuth, The evaporation duct height is found to be higher in the high wind regions flanking the islancl and in the shear zones laterally bounding the wake, while duct heights in the wake itself are reduced from upstream values, thereby yielding pronounced inhomogeneity in the duct height field leeward of the island. The wind direction was out of the east, therefore causing the wind to be going over and around Kauai (Figure 3).

MODELED SEA CLUTTER POWER
The radar equation of Barton [2] can be manipulated to obtain the modeled radar clutter power Pcmo.e, in dBs, from the sea surface at range r to give cmo.e, (r) = A.(r) -2L(r) + a"(r) + oflset P where Ac is the area illuminated by the radar, L is the propagation loss obtained from an electromagnetic propagation model, and u0 is the normalized radar cross section. An offset is then added to PcmOd.l (r) which references the clutter power from range r. from the ob-served and modeled clutter power to give

offset = Pcoba (r.) -P.modeI (TO)
where Pcob.(ro )is the average observed clutter power at r~and Pcmode, (r~) is the average modeled clutter power at r. both taken over 360 degrees. The modeled clutter power was generated in the following manner. 1.
The range dependent evaporation duct heights (d), wind speeds (VW) and wind directions (0) In this equation, &f(z, 6) is the modified refractivity with respect to 6 and z (the altitude above the sea surface), while .zo is a roughness factor whose typical over-water value is 1.5 x 10-4 [3].
The range dependent refractivity profiles and wind speeds along with certain parameters of the SPY-1 radar (frequency, beamwidth, antenna height, etc.) are input into the Advanced Propagation Model (APM) [4] to generate values of propagation loss L and grazing angle~at a height of 1.0 m over a range of 1-200 km.
The Georgia Institute of Technology (GIT) sea clutter model is used to compute a". The GIT o" is a function of radar wavelength (A),~, 6, VW and average wave height (haV) [5]. Since the h.. formula in the GIT model is based on a fullydeveloped sea, -a poor assumption in the lee of Kauai -a constant wave height was assumed. The constant hav was calculated by finding the average wind speed on each azimuthal step of 1.5°. These wind speeds were than averaged over the entire 360°and this average wind speed was then input into the hav formula in the GIT model.

RESULTS
The observed thresholded clutter map taken aboard the USS O'Kane is shown in Figure 4. The modeled clutter map generated from the COAMPS fields is shown in Figure 5 with the red lines representing the range the observed clutter power extended compared to the modeled clutter ranges for each 10°sector. Qualitatively, the modeled map displays much of the same features as the observed clutter map. They both have clutter power extending further out in range in the northerly and southerly directions from center. This appears to coincide with the higher wind speed and evaporation duct heights that were seen in Figures 1 and 2. Looking toward the island of Kauai, (90°radial) and the island of Niihau, (250-260°radials) the clutter falls off much more rapidly which would indicate lower evaporation duct heights and wind speeds as figures 1 and 2 show. The RMS difference between the model-predicted and median-filtered, observed clutter is N 1.5 d13.

SUMMARY
This demonstrates an instance where the outputs from the COAMPS model can be mapped into the space of the radar observations with (at least) visually appealing results. Mapping from the space of the model to the space of the observations (or vise versa) is a necessary step in fusing data from the model with that from the radar. Our future efforts will include: (a) bringing in ha. values from a wave model, (b) using a more rigorous modeling of the evaporation duct (i.e., using stabilitydependent profiles rather than the neutral profile), and (c) examining a broad range of cases. [1] [2] [3] [4] [5] [6]