Ray BellSOES6026Ian Robinson
TROPICAL CYCLONE ANALYSIS WITH SATELLITE RADARS
INTRODUCTION
1
Ray BellSOES6026Ian Robinson
Tropical cyclones are amongst the most devastating natural disasters; they can cause substantial loss of life, damage properties and shape financial markets every year. This has recently been highlighted by hurricane Katrina which in August 2005 displaced hundreds of thousands of people and damaged major oil refineries which caused gasoline prices in the USA to soar (Sampe and Xie, 2007). High wind speed events play an important role in earth’s climate. They remove heat and moisture from the ocean which affects the formation of deep water to drive ocean circulation (Bourassa, 2009). The strong air-sea exchange affects gas levels such as carbon dioxide and mixes water masses. They have a noticeable influence on pumping up nutrients from the deep and increasing the number of plankton. Tropical cyclones are defined as storm systems characterised by a low pressure centre with strong winds and heavy rains. There are subtle differences between the terminology of tropical cyclones, hurricanes and typhoons based on location but all have wind speeds of greater than approximately 20 m/s.
Satellite radars have an advantage over conventional observations of tropical cyclones due there global scale sampling (Smith et al, 2009). The intensities of tropical cyclones often make it a hostile environment for in-situ observations. They provide an advantage over other remote sensing satellites as they can penetrate through cloud, which are largely present in tropical cyclones. They have provided additional data about tropical cyclones which have shown a marked improvement in forecasting. There have been many case studies from satellite radars of scatterometers, altimeters and Synthetic Aperture Radars (SARs) which have flown over the cyclone or in very close proximity and obtained results. However, the satellite radars present problems in areas with high rain fall and the calibration of extreme wind speeds.
Figure. 1. QuickSCAT wind speeds and direction of hurricane Katrina in 2005. The eye of the hurricane is visible within the purple swirls and wind speed decreases radiating outward. The barbs reveal wind direction and the white barbs show heavy rainfall. To convert knots to m/s divide by approximately 2. NASA (2006).
SCATTEROMETERS
Scatterometers work by sending out a radar signal obliquely and measuring the normalised radar cross section (σ°) which is related to surface roughness. Scatterometers give surface wind speeds and also offer the ability to measure wind direction as they measure several σ° from different azimuths angles. The values of σ° are placed into an empirical model to obtain surface wind speed and direction (Nadari et al, 1991). They offer a common spatial resolution of approximately 25-50km2, large swath width of e.g. 1800km for QuikSCAT and a repeat cycle of 4 days, which gives unprecedented synoptic observations of surface wind and can reveal cyclone structure with good accuracy. Figure. 1. shows the scatterometer QuikSCAT observations as it flew over hurricane Katrina in 2005.
Although scatterometers have a fairly good spatial coverage of tropical cyclones a main limitation is their calibration against very high winds, for example the QSCAT-1 geophysical model function (GMF) in which the QuikSCAT scatterometer obtains wind speed is known to be limited at 30m/s (Hennon et al, 2006) (Figure. 2.). Understanding the maximum wind speed of tropical cyclones is very important to assess the likely damage it will cause. Fernandez et al (2006) noted that when using airborne backscatter measurements of C and Ku bands to obtain moderate to high wind speeds (25-65 m/s) the σ° response became saturated and stopped increasing at hurricane force winds for both frequency bands and polarisations. Other studies also show the scatterometers underestimate high winds as the σ° response becomes saturated or due to the shortcomings of the GMF (Zeng and Brown, 1998; Weissman et al. 2002; Yuan, 2004). Improved GMF have been suggested to measure these high wind speeds which give a more accurate wind field. However, validations of high wind speeds are hard to come by due to sparse observations especially in data poor regions such as the Southern Ocean.
To improve the GMFs which provide wind speed and direction from scatterometers, parametric equations are now given as wind speed becomes greater than a certain threshold, specifically for tropical cyclones (Yueh, 2003; Draper and Long, 2004; Atlas et al, 2005; Adams et al, 2005). Hennon et al (2006) validated surface wind speeds by using Stepped Frequency Microwave Radiometers (SFMR), onboard a National Oceanic and Atmospheric Administration aircraft and analysed winds for nine hurricanes and one tropical storm. It should be noted the choice of these in-situ wind products for their study are also limited (Uhlhorn and Black 2003). Fernandez et al (2006) gives new GMF for individual C and Ku bands up to wind speeds of 65 m/s. Although most of these GMFs give better results in areas around the cyclone, the strongest winds speeds near the eye can reach > 70m/s. In addition, the GMFs still obtain poor data due the presence of heavy rain near the core. This is the greatest limitation of measuring tropical cyclones from radars.
Figure. 2. Observation of hurricane Fabian in 2003 at 2149 UTC. Retrievals of objective smoothing and empirically tuned wind speeds in the Ku-2001 GMF Vs. Analysed winds. It is also post-processed to be a 2.5 km Ultra High Resolution product (UHR KU). Hennon et al (2006)
Rain interferes with σ° as it affects small-scale surface roughness, attenuation and scattering of the radar signal in the atmosphere; therefore it is complex and hard to take into account (Weissman et al, 2002). Measurements with the Ku band are known to be strongly affected by rain much more than the C band, especially at high incidence angles (Nie and Long, 2008). Sea surface roughness is affected further by the airflow associated with the rain event. Nie and Long (2008) found that additional scattering from rain causes wind speeds to appear higher than expected. They created a simultaneous wind/rain retrieval (SWRR) using the C band with the ERS scatterometer, ESCAT and collocated with the Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR), and numerical predicted wind velocities from the European Centre for Medium-Range Weather Forecasts (ECMWF) for incidence angles > 40°. It is shown to significantly improve wind velocity estimates and the surface rain rate in moderate to heavy rain cases. Heavy rain can also bias the wind direction towards along track, regardless of the true wind direction and therefore disrupt the wind field of a tropical cyclone, which is not improved by the SWRR, and results become noisier.
The scatterometers offer the lowest spatial resolution in comparison to the other radar satellites. Tropical cyclones are complex features, which have small-scale features especially in the eye which regular scatterometer products cannot resolve (Quilfen et al, 1998). Recently higher resolution scatterometer products have become available with resolutions of 2.5 km x 2.5 km. Williams (2008) uses a wind field model of a tropical cyclone to correct rain contamination which provides better observations for the eye of the cyclone and wind direction. However, these are inherently noisier than the standard 25 km products and again underestimate high wind speeds.
ALTIMETERS
Altimeters offer a greater spatial resolution than the scatterometer with a footprint of approximately 5 km, however they are only nadir directed, have a longer repeat cycle and rarely catch the strongest wind speeds of a tropical cyclone. The altimeter works by sending out a nadir directed pulse of microwave radiation continuously along its track which reflects back from the sea surface as backscattered return. The return can then be used to obtain geophysical properties of the ocean (Fu and Cazenave, 2001). The backscattered power (σo) is given by the amplitude of the return pulse which is used to estimate wind speed and the slope of the waveform is related to significant wave height (Hs). Figure. 3. shows when Envisat measured Hurricane Juan in 2003. In addition to studying wind speed to measure tropical cyclones the altimeters have an advantage in offering the sea surface height anomaly (SSHa) and Hs which can also be related to the tropical cyclone and sparsely observed from in-situ measurements.
Figure. 3. Combined sensor output for a single Envisat pass across Hurricane Juan at 14:37 UT on 27th Sept. 2003. (left and right) Swath information derived from the AATSR, with the black lines indicating the nadir track. (middle) The derived attenuation from the RA-2 (in black when statistically significant and related to rain rate) along with the Brightness Temperature values from the MWR-2 channels (pink 24 GHz, red 36 GHz). Quartley et al, 2007.
Similar to the scatterometer most research has been to investigate the effect of high wind speeds and heavy rain on altimeter wave forms in tropical cyclones (Young, 1993; Quartley, 1997; Quilfen et al, 2006; Quartley and Guymer, 2007). Quilfen et al (2006) investigated the effects of winds, waves and rain rates on altimeter signals from tropical cyclone Isabel in 2003. The dual frequency altimeters on board Jason-1 are used to correct for ionospheric effects, which affects the radar signal path delay, but also flag measurements affected by rain. The effects of rain on σo are fairly similar to scatterometers. An increase in wind speed is responsible for the large drops in both C and Ku band σo, but larger attenuation is found for the Ku band σo in the presence of rain. They studied two altimeter tracks within 100 km from the observed eye of the storm and had wind speeds >25m/s. They created a corrected C band σo and fed it back to create a corrected Ku band for observations of extreme conditions. The corrected bands give much better results for the surface wind speed and are larger than the Young (1993) algorithm. The rainfall asymmetry was found to shift with increasing intensity; the stronger the tropical cyclone the more axisymetric the inner core. Satellite altimeters have shown that the maximum rainfall remains in front of the tropical cyclone at all speeds (Lonfat et al, 2004). These are some examples how satellite radars have verified the theory of tropical cyclones.
Altimeters provide Hs as the stretching effect of the signal by the time delay of returns from the wave crest first and the wave trough later. Quilfen et al (2006) showed Hs reached up to 11m and 14.7m either side of the centre by using its own tracking algorithm to process C band waveforms little affected by rain (Quartley, 1997) in tropical cyclone Isabel (Figure. 4.). This verified the theory of tropical cyclones that high wind speeds and sea states are more likely to be larger to the right of the direction of tropical cyclones (Willoughby and Rahn, 2004).
Altimeters have an advantage over the other radars as they provide SSHa which qualitatively gives the integrated water column temperature and therefore the amount of potential energy. This is important to investigate the formation of tropical cyclones by cyclogenesis, as is known to exist in areas with sea surface temperature greater than 26°C, as well as favourable atmospheric conditions.
Figure. 4. Ku and C band Hs in meters for Jason orbit 50 (top) and 152 (bottom). Quilfen et al (2006).
Important processes at the air-sea interface are a key component in driving tropical cyclones and therefore it is essential to improve our knowledge about them. Shay et al (2000) showed the intensification of hurricane Opal in 2005 was likely due to a warm eddy from the loop current in the Gulf of Mexico. Other results indicate that in 31 out of 36 cases hurricane intensification can be linked to an increase in the values of hurricane heat potential of approximately 30 kJ/cm2 under the storm track (Goni et al, 2003). Lin et al (2005) showed that the intensity of supertyphoon Maemi in 2003 went from having 1 minute sustained winds of 41 m/s to its peak of 77 m/s as it passed over a warm eddy (Figure. 5.). The incorporation of this eddy activity yielded an evident improvement on Maemi’s intensity evolution. Algorithms have been produced to map the 26°C isotherm using altimeter SSHa which is shown to influence the track of a hurricane (Goni et al, 2003). Although to fully understand a hurricanes track the atmospheric conditions must be taken into account but cannot be provided with satellite radars. Altimeters like all radars are unfortunately limited when they reach the coast where tropical cyclones have a direct impact on the population. This is currently a large focus of research.
Figure. 5. Detailed SSHa distribution from a composite of TOPEX/Poseidon and Jason-1 measurements during the cycle (30 Aug–8 Sep) before Maemi’s encounter of the eddy WOE-2. Maemi’s intensity (in Saffir–Simpson scale) and radius of maximum wind are also shown. The storm position is denoted every 6 h. Lin et al (2005)
SARs
SARs offer the highest spatial resolution of the radars as they can measure down to 25m x 25m pixels. SARs offer wind speed and wind direction from images and have potential for observing hurricanes. SARs can retrieve extreme ocean surface winds using the same GMF C band - CMOD5 as scatterometers (Shen et al, 2006). The maximum wind speed can be obtained from SAR images by fitting an appropriate Holland (1980) model given by the parametric model with wind speed below or
above 20m/s with an accuracy of 10% (Reppuci et al, 2008). However the results become underestimated at high wind speeds (Pichel et al, 2007; Nie and Long, 2008b). A problem with this method is the SAR uses a priori of Hurricane Research Division (HRD) wind directions upon which to base their GMF. Additionally, errors arise from rain (Weissmann and Bourassa, 2008) and the GMF applied.