Validation of Parametric Hurricane Models by ENVISAT ASAR
ABSTRACT In the past decade several satellite borne synthetic aperture radars (SAR) have been put into orbit. A synthetic aperture radar (SAR) not only records the intensity of the returned signal, but also the phase history of the backscattered radar signal and is processed to high spatial resolution images. To observe the structure of tropical cyclones at the sea surface RADARSAT 1 and ENVISAT ASAR Scan SAR images (400 x 400 km coverage) are the preferred datasets used. Together with optical imagery that yield information on the cloud tops they yield the possibility to investigate the three dimensional structure of tropical cyclones. The intensity images are calibrated and information on wind speed is derived. In addition sea surface features relating to cloud structure and rain rate are analysed. We give an assessment on the possibility to retrieve wind speed from SAR using the CMOD algorithms.
In addition we have been investigating the following features in SAR images of hurricanes:
- Wavelength and Direction of Boundary layer rolls for information of mixed boundary layer depth - Radius of maximum wind speed - Sea State in terms of wavelength and -direction
These image parameters are related to parametric models of hurricanes and validated by aircraft measurements from the national hurricane center (NHC). The work aims at the improvement of prediction of the cyclone track, intensity and sea state in such high wind speed conditions.
- SourceAvailable from: Rasmus Tonboe[Show abstract] [Hide abstract]
ABSTRACT: The high spatial resolution and large coverage of satellite-based synthetic aperture radars (SAR) offers a unique opportunity to derive mesoscale wind fields over the ocean surface, providing high resolution wind fields near the shore. For this purpose, algorithms were developed and tested using the ScanSAR aboard the Canadian satellite RADARSAT-1, operating at C-band with horizontal polarization in transmit and receive. Wind directions are extracted from wind-induced streaks visible on most SAR images. Wind speeds are derived from normalized radar cross sections (NRCS) using empirical models. The models were developed for scatterometers (SCAT) operating at C-band with vertical polarization and must be modified for horizontal polarization. Several available C-band polarization ratios were considered, including theoretical and empirical forms. To verify and improve the algorithm, wind speeds were computed from several RADARSAT-1 ScanSAR images and compared to colocated measurements from the SCAT aboard the European remote sensing satellite ERS-2 and to the results of the Danish high resolution limited area model (HIRLAM). Using the colocated measurements, the polarization ratio was estimated and applied to improve the wind retrieval algorithm. In addition, the main error sources in SAR wind field extraction are discussed with respect to the RADARSAT-1 ScanSAR data. Sensitivity studies were performed under different atmospheric situations using the modified C-band model to compute the errors due to wind direction and inaccuracies in NRCSIEEE Transactions on Geoscience and Remote Sensing 10/2000; · 3.47 Impact Factor
Article: Ocean winds from RADARSAT-1 ScanSARCan. J. Remote Sens. 01/2002; 28(3):524-533.
- [Show abstract] [Hide abstract]
ABSTRACT: This article reviews several microwave instruments employed in research and analysis of tropical cyclones (TCs), typhoons, and hurricanes. The instruments discussed include scatterometers, microwave radiometers, synthetic aperture radars (SARs), and rain radar from space. Examples of the particular contribution by one or more of these instruments in analysis of several storms illustrate the comprehensive new views provided by the SeaWinds scatterometers, the detailed high-resolution wind field provided by RADARSAT-1 SAR, particularly inside and in the vicinity of hurricane “eyes,” and the presence of secondary flows in the region between rainbands in TCs. The high spatial resolution of precipitation data from the Tropical Rainfall Measuring Mission's rain radar, combined with scatterometer or SAR data, give a significant improvement in the details that can be seen from space, at the surface, and in the precipitating areas of TCs. The microwave instruments provide a penetrating view below the upper level cirrus clouds.Journal of Oceanography 01/2002; 58(1):137-151. · 1.05 Impact Factor
Validation of Parametric Hurricane Models by ENVISAT ASAR Images
S. Lehner, A.Reppucci, J. Schulz-Stellenfleth
German Aerospace Center (DLR), Oberpfaffenhofen 82234
Wessling, Germany. Tel. +49 8153282101, email: firstname.lastname@example.org
In the past decade several satellite borne synthetic aperture radars (SAR) have been put into orbit. A synthetic aperture
radar (SAR) not only records the intensity of the returned signal, but also the phase history of the backscattered radar
signal and is processed to high spatial resolution images. To observe the structure of tropical cyclones at the sea surface
ENVISAT ASAR and RADASAT 1 Scan SAR images (400 x 400 km coverage) are the preferred datasets used.
Together with optical imagery, that yield information on the cloud tops, SAR images yield the possibility to investigate
the sea surface properties of tropical cyclones. The SAR intensity images are calibrated and information on wind speed
is derived. In addition sea surface features relating to cloud structure and rain rate are analysed. We give an assessment
on the possibility to retrieve wind speed from SAR using the CMOD5 algorithm.
In addition we have been investigating the following features in SAR images of hurricane Katrina: wavelength and
direction of Boundary layer rolls for information of mixed boundary layer depth, radius of maximum wind speed, sea
state in terms of wavelength and –direction.
These image parameters are related to parametric models of hurricanes. The work aims at the improvement of
prediction of the cyclone track, intensity and sea state at these high wind speed conditions.
While images taken with optical sensors are dependent on sunlight for illumination, radar signals can be used to observe
the earth`s surface at day and night and in all weather conditions even through cloud coverage. A synthetic aperture
radar (SAR) not only records the intensity of the returned signal, but also the phase history of the backscattered radar
signal and is processed to high spatial resolution (~30 meter) images.
In the past decade several satellite borne synthetic aperture radars (SAR) have been put into orbit to help monitor the
Presently there are several satellites in a low earth orbit acquiring SAR images. They are the Canadian RADARSAT-1,
and from ESA the European Remote Sensing (ERS-1/2) and ENVISAT satellites. While all these satellites transmit in
C-band, RADARSAT-1 is horizontally (HH), ERS-1/2 are vertically (VV) polarized, and ENVISAT has multi-
The independence of SAR measurementson on daylight and cloudiness together with their high resolution and large
spatial coverage make them a valuable tool for measuring and observing geophysical parameters like the surface wind
field. In the past it has been shown, that SAR imagery offers a unique opportunity of observing features associated with
hurricanes . Goal is to understand the processes involved in the air sea interaction in the boundary layer and thus to
improve numerical modeling and forecasting of sea surface wind fields and the sea state.
We have been investigating the following features in SAR images of hurricanes:
Eye size and eccentricity as a measure of hurricane development
Wavelength and Direction of Boundary layer rolls for information of mixed boundary layer depth
Sea State in terms of wavelength and direction
Synthetic Aperture Radar (SAR) data is available in various formats, depending on processing level, station and satellite
type. ERS SAR images of 100 x 100 km size are available in the so-called ERS SAR image mode and at high resolution
of 30 m. This mode can only be acquired when the satellite is in line of sight of a ground receiving station, due to
limited onboard storage capability. RADARSAT and ENVISAT have the ability to acquire data in the so-called Scan
SAR mode, in which several beams are combined to generate an image up to 500 km x 500 km in size. In Scan SAR
mode a large part of a tropical cyclone (hurricane) can be imaged synoptically.
Envisat simultaneously acquired images of Katrinas showing the eye at 1550 UTC (1150 US Eastern Daylight Saving
Time) on 28 August, with its Medium Resolution Imaging Spectrometer (MERIS) and Advanced Synthetic Aperture
Radar (ASAR), when the hurricane was a Cat 5 hurricane. In the following we analyse the ASAR image of Katrina.
To compute wind fields over water by SAR, the normalized radar cross section, referred to as Sigma0, has to be derived
from the image. Sigma0 is independent of satellite type, and therefore can be used to identify the radar properties of
objects like the roughness of the sea surface. To recalibrate SAR images to Sigma0 values, several satellite sensor,
Figure 1a: Subimage of the calibrated ENVISAT ASAR Image around the eye of Katrina acquired on 28 Aug. 2005
15:20. The circle shows the radius of largest wind speed.
Figure 1b: Cut through the eye at constant incidence angle for the calibrated image.
product and scene specific operations have to be performed on the image data. Figure 1a shows a 200 x 200 km
subimage around the eye of hurricane Katrina, the radius of maximum wind speed Rmax = 25km as given by the
NOAA/HRD model data is shown in the image. The cross section of the calibrated data shows a strong dependency on
incidence angle. Figure 1b is a cut at constant incidence angle of Katrina. The dark hurricane eye and a ring around the
eye at about 35 km distance can be observed. From the MERIS image given at the ESA webpage it can be seen that the
dark ring is related to the top of the eye wall and thus probably related to ice clouds.
SPECTRAL ANALYSIS OF WIND ROLLS
On the SAR images often so-called wind streaks are visible, ranging in wavelengths from 600 to 2,000 meters. This
variation in sea surface roughness is explained by change in surface wind speed due to the formation of boundary layer
rolls. The direction of these streaks is thus used to derive the wind direction and the wavelength is taken to be a measure
of roll size and thus mixed layer depth.
Figure 2: Wind Field of Katrina according to NOAA AOML, HRD acquired at 1330 UTC.
Figure 3: Direction and wavelenght of boundary layer vortices of hurricane Katrina derived from image shown in Fig.1.
To determine wavelength and direction a 2-dimensional spectral analysis is performed. The image is split into
subscenes of an appropriate size of several kilometers. The image spectra are then calculated on these tiles using a Fast
Fourier Transform (FFT) algorithm.
In the following the SAR image of Katrina is divided into subareas of 12.5 km and on each subscene the image
spectrum is calculated. Length of the windrolls is determined by finding the maximum for wavelength between 600m
Figure 3 shows the direction and the wavelengths of the boundary layer rolls for hurricane Katrina. The lengths of these
roll vortices are between 1000 and 1200 m , longer wavelengths occur only when image features due to rain cells are
prominent on the image
Thus the wavelength of the rolls are definitely longer than measured for all other hurricane images, investigated before,
were the usual roll size was measured to be between 600 and 800 m, e.g. in Ivan, Floyd at a Cat 3 stage. As the roll size
is a measure of the boundary layer depth with an aspect ratio of horizontal to vertical roll size of about 2.4 (Ralph
Foster, pers. comm). A first preliminary conclusion is, that in a Cat 5 hurricane the boundary layer is deeper with
moisture being transported up higher into the hurricane thus reinforcing the circulation.
SAR WIND RETRIEVAL
SAR backscatter from the ocean surface is dependent on the small scale roughness, which in turn is strongly dependent
on the local wind field. Therefore the backscatter can be used as a measure for the local wind. SAR ocean surface wind
retrieval is a two-step process. In the first step, wind directions are estimated from wind induced streaks and in the
second, the wind speeds are derived from the normalized radar cross section (NRCS).
The SAR wind speed retrieval is based on the C-band scatterometer (SCAT) model usually the CMOD5, which gives
the empirical relation between the VV polarized NRCS, the wind vector and the incidence angle. CMOD4 is used by
ESA as the standard wind retrieval model for the ERS scatterometer was only tuned to wind speeds up to 15m/sec and
should thus not be extrapolated to much higher wind speeds. For hurricane cases CMOD5, which has been retuned for
high wind speeds ought to be used .
CMOD 5 saturates though, depending on incidence angle at about 20m/sec for wind speeds blowing towards the SAR
antenna. For hurricane wind speeds the NRCS even diminishes again, see figure 4. This together with the fact that the
NRCS is highly dependent on the wind direction and incidence angle and is additionally influenced by rain, does make
a C band radar in this set up not ideally suited to measure hurricane wind speeds in a robust manner.
In figure 5 we show the wind speed for hurricane Katrina derived by CMOD 5, using tangential wind direction. Several
imaging artifacts and rain bands can be observed. As CMOD 5 is obviously not suited to derive wind speeds in
hurricanes for VV polarized SAR imagery, we thus suggest to improve the CMOD for high wind speeds using a
hurricane model fitting wind speed and NRCS.
Figure 4: Calibrated Radar Backscatter derived by CMOD 5 for incidence angle of 23, 33 and 42 degrees (from top to
bottom) for wind direction towards the radar antenna.
Figure 5: Wind Speed and Direction derived from the 400 x 400 km ENVISAT ASAR image of hurricane Katrina
derived from image shown in Fig.1.
Figure 6: Holland model of hurricane Katrina, calculated with radius of maximum wind speed of 25 km and a central
pressure of 908 mbar.
Figure 7: Ocean wave lengths and direction of hurricane Katrina derived from image shown in Fig.1.
Figure 6 shows, e.g. a Holland model of hurricane Katrina for a central pressure of 908 mbar and a radius of maximum
wind speed of 25 km. For further explanations see Reppucci et al. this issue. In a next step we will fit a new hurricane
CMOD to such model results
SAR images show features due to ocean waves, usually waves longer than 100 m are imaged depending on travel
direction. Figure 7 shows the wave length and direction of the ocean waves as observed on the Scan SAR image of
hurricane Katrina. Ocean waves of wave length 250 to 300 m can be observed., with wavelength increasing, as the
swell travels away from the hurricane center.
A first analysis showed that ENVISAT ASAR hurricane images can be used to infer information on hurricane structure,
eye size and the sea surface roughness from the image.
Due to saturation CMOD 5 is not suited to derive wind speed from SAR images and parametric models should be used
to make the measurements more robust.
Even on the ASAR Scan SAR Images ocean waves can be observed and their wave length and direction can be
measured. This can be used to improve high resolution ocean wave models.
1. K.B. Katsaros, P.W. Vachon, W.T. Liu, and P.G. Black, “Microwave remote sensing of tropical cyclones from
space,” J. of Oceanography, vol. 58, pp. 137–151, 2002.
2. J. Horstmann, W. Koch, S. Lehner, and R. Tonboe, “Ocean winds from RADARSAT-1 ScanSAR,” Can. J. Remote
Sens., vol. 28, no. 3, pp. 524–533, 2002.
3. W. Koch, Directional analysis of SAR images aiming at Wind Direction, IEEE Trans. Geosci. Remote Sens., vol.42,
no 4, pp 702-710, 2004
4 S. Lehner, J. Horstmann, W. Koch and W. Rosenthal, Mesoscale wind measurements using recalibrated ERS SAR
images, J Geophys. Res., vol103, pp. 7847-7856, 1998
5. H. Hersbach, “CMOD5 an improved geophysical model function for ERS C-band scatterometry,” Internal report,
European Centre for Medium- Range Weather Forecast, 2003.
6. J. Horstmann, W. Koch, S. Lehner, and R. Tonboe, “Wind retrieval over the ocean using synthetic aperture radar with
C-band HH polarization,” IEEE Trans. Geosci. Remote Sens., vol. 38, no. 5, pp. 2 122–2 131, 2000.