On February 20th, 2016, the Fiji islands were hit by one of the most intense tropical cyclones in history, a cyclone named Winston. With 1-min maximum sustained winds reaching 285 km/h (80 m/s), Winston is second in strength only to Super Typhoon Haiyan. Winston persisted for more than two weeks and took a very unusual track that included a sudden reversal of direction and a rapid intensification.
As is often the case for Southern Hemisphere storms, due to the remote storm locations, there were no in-situ wind observations available from hurricane hunters’ aircrafts or dropsondes. To forecast most Southern Hemisphere storms, forecasters rely on visible and infrared satellite imagery of the storm and its evolution (Dvorak technique).
Space-based instruments capable of measuring ocean surface winds provide critical information about the storm intensity. Multiple space-based microwave sensors are currently operating: The European scatterometer ASCAT (C-band, 5.2 GHz), the NASA scatterometer RapidScat on the International Space Station (Ku-band, 13.4 GHz), the NASA SMAP radiometer (L-band, 1.4 GHz), and the US Navy multichannel polarimetric radiometer WindSat (6-37 GHz). All these sensors can measure wind speed and direction at spatial resolutions of about 12-50 Km.
While the sensors give accurate and consistent wind measurements in the range of 0-30 m/s, the instrument sensitivities and the reliability of the sensor winds at extreme wind speeds are still under investigation. There are very few in situ observations of extreme winds available for comparison.
For Winston, we show here a comparison of the winds from four microwave sensors for two sample days during Winston, one day in the development phase and one day in the mature phase. We also show a comparison of a time series of 10-min maximum sustained wind values with those reported by the National Weather Service and a widely used numerical forecast model.
With this study, we address two questions:
- How reliable are the microwave sensor measurements of extreme winds?
- How consistent are the wind speeds from the different microwave sensors?
The four figures below display the wind fields (m/s) in the development phase, on February 14th. The winds are from (clockwise from top left): ASCAT, WindSat, RapidScat, and SMAP. Below the wind plots is a figure showing the rain rates (mm/h) observed on the same day by WindSat. Note that each of the sensor overpasses occur at slightly different times of the day and the storm is therefore shifted depending on the translation speed of the storm on that day.
The winds from these sensors show similarities in pattern and intensity. However, there are some differences: SMAP measured a higher intensity (44 m/s) compared to RapidScat (36 m/s), ASCAT (35 m/s), and WindSat (32 m/s). This could be due to the combination of two factors: at the low L-band frequencies, the SMAP microwave signals are less impacted by rain. Also because of the low frequencies, SMAP demonstrates a higher sensitivity at high wind speeds. RapidScat and WindSat winds appear to be more affected by rain than ASCAT, as the rain pattern is clearly visible. Both SMAP and ASCAT can sufficiently distinguish Winston’s eye.
The following set of figures show the wind fields (m/s) and rain rates (mm/hr) on February 21st, soon after Winston hit the Fiji islands.
The images from all sensors consistently show that the storm became more symmetrical and more intense. SMAP maximum winds reached 53 m/s, while the other sensors recorded much lower maximum winds (35-36 m/s), again, likely due to rain impact and reduced sensitivity at the C- and Ku-band frequencies. Over the duration of the storm (not shown), ASCAT winds are less affected by rain than either WindSat or RapidScat, as expected.
Summary and Conclusions
The figure below shows a comparison of 10-min maximum sustained winds over the lifetime of the storm for SMAP, ASCAT and WindSat with those reported by the NOAA National Weather Service (NWS) and model winds from the NOAA NCEP Global Forecast System (GFS) (0.25o deg model ouput).
SMAP outperforms all the other instruments, and over the two weeks, displays maximum winds which are very consistent with those reported by the NWS and the NCEP forecast model. The skill in satellite observations of extreme winds with SMAP is remarkable and unprecedented, and shows great potential for tropical storm wind forecasting in remote locations. An analysis of SMAP winds for Hurricane Patricia shortly before landfall (10/23/2016) has shown excellent agreement with the airborne Step Frequency Microwave Radiometer (SFMR) up to 65 m/s. This increases confidence that the SMAP radiometer is able to measure such extreme wind speeds. So far, none of the other sensors (ASCAT, RapidScat, WindSat) have been able to do that.
Data Sources: SMAP, ASCAT, and WindSat winds and rain rates displayed here are developed and processed at Remote Sensing Systems (RSS). ASCAT and WindSat are publicly available at www.remss.com. SMAP wind data are not yet available to the public, as the wind direction algorithm is still under development. RapidScat data are developed at NASA JPL, in collaboration with RSS, and distributed through the Physical Oceanography Data Active Archive Center (PO.DAAC). The RapidScat winds displayed here are the science data 12Km winds, uncorrected for rain, but adjusted for the low SNR III anomaly state. The NWS weather reports for Winston can be found among the references here. NCEP GFS data are available from NOAA.