Figures and text for Poster at NASA/NOAA meeting (Miami, February 8-10, 2005)

The pdf file of the finished poster can be viewed and downloaded from here . The text and figures are listed below.

Click on images for enlarged versions of plots.

The observation of hurricanes over the ocean presents an exceptional challenge for instrumentation design, deployment and data calibration. Proper characterization of all significant scales of a hurricane require large-scale data coverage (Order 1000 km) and the ability to resolve very small dynamic structures (Order 1 km or less) at hourly time intervals. In this study, the wind speed data from three different sources are compared: satellite scatterometers (from QSCAT, using "standard" 25 x 25 km winds and also ultra high-resolution, experimental 2.5 x 2.5 km images) and ocean surface drifting buoys.

Standard QSCAT

The scatterometer instrument aboard QSCAT is a SeaWinds system, which transmits a radar signal of known frequency (13.4 GHz, Ku band) and polarizations to the sea surface, and then measures the backscattered radar return. The radar backscatter is a function of sea surface roughness, the angle of incidence of the transmitted pulse, and other parameters. Several returns, for different geometries and radar properties (e.g. polarization), are used from the same patch of ocean surface to retrieve estimates of wind speed and direction by inverting an empirical model function. The model function relates speed and direction of the equivalent neutral stability wind at 10m above sea surface to the geometric and radar properties affecting backscatter (Freilich, 1996). In this study, the direction interval retrieval with threshold nudging (DIRTH) Level 2B winds are used. "Standard" QSCAT is provided on a 1800km wide swath, and 25 x 25km grid. QSCAT processing provides a rain-flag which is based on the "multidimensional histogram" rain probability (MUDH).

High-Resolution QSCAT

For the standard product, several radar signals from 6x25km surface elements, called "slices", are averaged into "egg" measurements, which have an effective size of ca. 25 x 32km. The standard Level 2B wind product is produced on a 25km grid-spacing using the egg measurements. The slice measurements densely sample the surface and have significant measurement response overlap. This "over-sampling", along with the non-ideal roll-off of the spatial measurement response, makes it possible to produce enhanced resolution of the radar images, down to a pixel size of 2.5 x 2.5km (Long et al., 2003; and Long and Luke, 2003). As in the the standard product case, the high-resolution wind processing produces one to four "ambiguities" with similar wind speeds, but different wind directions. Initially the ambiguity closest to the standard L2B 25km value is selected. Then, a median-filter ambiguity selection algorithm is applied with a window size of 17.5 x 17.5km to select the final ambiguity. Rain has the same effect on the high-res wind speeds as on the standard winds: high winds will be reduced in high rain while low winds will appear enhanced in high rain. Low rain has little effect on high winds, but will enhance low wind speeds.

Drifter Data

During September 2004 several Minimet drifters were air-deployed in front of the projected paths of hurricanes Frances (Class 4) and Jeanne (Class 1). The air-deployment of drifters to study the details of hurricane evolution has successfully been employed for hurricane Fabian in September 2003 (Niiler et al., 2003). Together with other drifters in the area, the data from 58 drifting buoys are used to provide atmospheric pressure, SST, wind direction, and wind speed. For this study, the wind speed observations from seven Metocean drifters and one Minimet are compared with scatterometer measurements. Wind speed observations derive from acoustic pressure data collected by WOTAN (Wind Observation Through Ambient Noise) hydrophones.

The tracks of all 58 drifters present during the passage of hurricanes Frances and Jeanne are presented above (8/31-10/1/2004). Drifters with speed measurements are drawn in blue. The deployment positions or the locations of the drifters on 8/31/2004 are marked with dots. An example of drifter observed wind speed and direction data is shown for Metocean drifter 41539. The distance between drifter and hurricane is also plotted. QSCAT data have been collocated to the locations of the drifter observations, when both measurements are within 1 hour and 100 km of each other. On average, the time difference is 22 min and the distance 15 km. Rain-flagged QSCAT wind speeds are marked with open squares.

The differences of high-res minus standard QSCAT data are plotted vs. drifter observed wind speeds. In these comparisons high-res speeds tend to be slightly lower than standard QSCAT data in low and moderate winds (<12m/s). At higher wind speeds, the differences tend to become even larger. As none of the drifters with speed measurements passed directly under the center of the hurricane, the highest recorded wind speed in the collocation data set is 26m/s.

Acoustic measurements from hydrophones in the ocean suffer from two major problems, resulting in opposite effects. As the wind speed increases, successively higher frequency bands become saturated. The most commonly used 6-10 kHz band becomes saturated at about 17 m/s wind speed, due to saturation of the bubble cloud which emits this frequency sound. On the other hand, the ambient noise can be contaminated by the sound from breaking waves. Longer sampling intervals (ca. 15min) could eliminate these outliers. The power requirements, however, of the onboard acoustic data processing are severe. Large power supplies require large floats and large drogues, which very quickly increase the cost per drifter. Drifter data is valuable only when a large number of low cost units can be deployed.

There is no practical way to assure the performance of hydrophones, compasses and the entire electronic/mechanical system a priori. The response of drifters manufactured in an identical way can differ by as much as 5-20%. Calibration can only be achieved with satellite data in an effective way.

Maps of wind speed from standard and high-resolution QSCAT data are presented for three revs that passed over hurricane Jeanne. The revs are separated by 13 and 10 hours on 9/20 and 9/21/2004. The high-res data is able to resolve the vey high wind speeds near the eye of the hurricane: in rev 27374 (central panel), the high-res winds reach a maximum of 38 m/s while the maximum standard winds are only 29 m/s. Also, the relative wind speed minimum at the center of the hurricane is more pronounced in the high-res data (ca. 10 m/s) than in the standard product (not less than 15 m/s). In areas further away from the hurricane center (e.g. the northwest region in rev 27374), wind speed is quite variable in the high-res data (winds vary between 10 and 15 m/s), while the standard product seems more biased towards higher wind speeds (>= 15 m/s).

Wind directions from standard QSCAT are presented below for the same revs as above. Rain-flagged data are plotted in red. Some of the areas with relatively high wind speeds away from the hurricane center appear in regions with rain, where the wind speed retrieval is less reliable.

From maps of wind speed the center of hurricane Jeanne can be estimated by locating the relative minimum in wind speed near the eye of the hurricane. Estimates from standard and high-res QSCAT are shown below. Only nine revs crossed the track of the hurricane between 9/18 and 9/26/2004. Both the standard and the high-res winds provide reasonable positions when compared with NOAA's "best track" data.

Also shown are the maximum wind speeds recorded in QSCAT data. Most of the maxima are only associated with one or two wind vector cells (i.e. 25 x 25 km or 2.5 x 2.5 km). The NOAA wind speeds plotted below are the "best track maximum sustained winds". NOAA's estimates of "maximum gusts" can exceed those by 5 to 10 m/s. The high-res maximum wind speeds correspond much better with NOAA's estimates. On average, the maximum high-res winds exceed the maximum standard winds by 9 m/s (except for one rev).

QSCAT wind speeds along two across-swath sections are plotted below for rev 27374. The positions of the cross-sections are indicated on the wind speed maps for this rev. Hurricane Jeanne is captured in these images on 9/20/2004, 23hr UTC.

In the section through the center of the hurricane, the high-res winds reach a maximum of 38 m/s, while the maximum standard winds are only 26 m/s. The relative minimum of wind speed is 12 m/s in the high-res winds, and 17 m/s in the standard data. In both cross-sections, there are regions away from the hurricane center where the standard wind speeds are consistently larger than the high-res winds, on average by 2-3 m/s, but sometimes by as much as 6m/s.

Summary

The reliable determination of ultra high-resolution (2.5 x 2.5 km) wind speed and direction holds the promise of providing detailed images of hurricane structures not available from standard QSCAT (25 x 25 km). The deployment of large arrays of inexpensive drifters requires calibration that can only be done with satellites in an effective way.

Besides high-frequency wind speed and direction, drifters also measure other important ocean-atmosphere parameters (most importantly, atmospheric pressure and SST). The combination of large-scale, instantaneous snapshots from satellites with high-frequency drifter observations, allows detailed studies into the genesis and evolution of tropical storms and hurricanes.

Currently, scatterometer wind observations in hurricanes are limited by: a) temporal sampling based on a single satellite in polar orbit, making 3-6 hour sampling of storm systems impossible; and b) conservative rain flags in high winds.

References

Freilich, M.H., 1996: Sea winds algorithm theoretical basis document. Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, 56 pp.

Long, D.G., J.B. Luke, and W. Plant, 2003: Ultra High Resolution Wind Retrieval for SeaWinds. Proceedings of the International Geoscience and Remote Sensing Symposium, pp. 1264-1266, Toulouse, France, 21-25 July 2003.

Long, D.G., and J.B. Luke, 2003: High Resolution Wind Retrieval for SeaWinds. Proceedings of SPIE Vol. 5155 Ocean Remote Sensing and Imaging II, ed. R.J. Frouin, G.D. Filbert, D. Pan (SPIE, Bellingham, WA), pp. 216-225.

Niiler, P., W. Scuba, and D.K. Lee, 2004: Performance of Minimet Wind Drifters in Hurricane Fabian. Journal of the Korean Society of Oceanography, Vol.9, No.3, August 2004.