Super Typhoon Bavi: The CIMSS Perspective
Earlier this week, we discussed the rapid development of Super Typhoon Bavi, a major tropical cyclone in the western Pacific Ocean. It delivered a strong hit to the island of Rota, part of the American territory known as the Commonwealth of the Northern Mariana Islands. You can read more about Rota’s devastation and see photos of the aftermath in this article from the Stars and Stripes. Even for a typhoon, Bavi was an extraordinary event. Note how the Joint Typhoon Warning Center described Bavi in one of its warnings:
TYPHOON 09W (BAVI) REMAINS A GARGANTUAN TYPHOON, WITH A WIND FIELD SIZE RANKING
IN THE TOP 3 PERCENT OF ALL WESTERN PACIFIC TYPHOONS OF THE PAST DECADE.
The National Weather Service forecast office in Tiyan, Guam was even more direct with its language when it issued an extreme wind warning issued in anticipation of Bavi’s landfall on Rota:
THIS IS AN EXTREMELY DANGEROUS AND LIFE-THREATENING SITUATION! TAKE COVER NOW! Venturing outside can result in DEATH from flying projectiles.
Their warnings weren’t unwarranted. By some estimates, this was the most powerful tropical storm to hit a part of the territory of the United States. In this post, we’re going to delve into some additional products that could be used to assess the storm and how it was evolving over time. This is a bit of a deeper dive that we normally post to the CIMSS Satellite Blog, but an extraordinary storm deserves an extraordinary analysis.
First, we’ll look at a standard product in a somewhat unique way. This movie shows the Band 13 infrared view from Himawari-9, but instead of having a fixed field of view it follows Bavi’s center of rotation. We can see most of the life cycle of this cyclone, from its initial identification as a tropical depression until it loses its clearly discernible eye. Watch as the circulation organizes then forms a tight, clear eye as it intensifies. The storm categorization is visible in the lower right of the image.
The infrared loop is the classic way of monitoring tropical systems, but there’s so much more that satellites can tell us. For one, it’s helpful to know about the environment in which a tropical system is forming. Here’s a map of some sea surface temperatures, courtesy of NOAA’s Coral Reef Watch. Remember, the Pacific is currently experiencing an El Niño. That’s easy to see when you look at a sea surface temperature anomaly map, which shows how far above or below normal the observed temperatures are. While the greatest sea surface temperature (SST) anomalies are centered right at the equator, the tropical western Pacific is experiencing temperatures that are 1-1.5 C above normal. The figure below shows the satellite-observed sea surface temperature anomaly map for 25 Jun 2026, the first day that the Joint Typhoon Warning Center issued an alert for a new invest in an area where cyclogenesis was favorable. The location of that invest is depicted with a purple star on the SST anomaly map.
Following several days of favorable shear and warm tropical waters, the storm intensified into a tropical depression on 1 July, and after that, further intensification was quite rapid. Analyzing the intensification of tropical systems is one of the applications where polar orbiting satellites excel. Geostationary satellites give us a nice, consistent view of particular place over time. However, geostationary platforms can’t support microwave antennas; the energy emitted from the earth and its atmosphere is so low in the microwave band compared to infrared emission and visible reflectance that it is effectively impossible to collect from the much larger geostationary altitude (22,000 miles vs 500 miles).
The CIMSS Tropical Cyclone Group has developed tools and products to monitor each tropical storm in every basin to help forecasters determine the strength and evolution of these powerful systems. Many of these products leverage the unique strengths of microwave imagers. Microwave has some particular advantages for tropical observations that are worth investigating here. One of the big ones is that microwaves are so long compared to infrared waves that they are generally unaffected by clouds and thus can tell us information about the precipitation structure within the cloud. This allows us to see some interesting things. Much of the time, the top of a tropical cyclone can be obscured by thick cirrus: as low level flow converges it ascents and condenses to form upper level clouds. This can really obscure where the center of circulation is, especially for developing storms. For example, the rightmost image is the Band 13 IR window view of Bavi from Himawari-9 on 2 July. The system really just looks like a large blob of convective cloud. However, looking through the cloud with several different microwave channels (the left three panels) shows us exactly where the eye of the storm is. Correctly identifying the center is critical for a number of reasons, including determining motion and properly assessing the storm’s intensity via the Dvorak technique (which we’ll discuss later).
The microwave images above are from a special data source. These datasets are produced via public-private partnership between NOAA and Tomorrow.io. Seeing the utility of microwave observations, Tomorrow.io has designed relatively low-cost sensors and put them into orbit to help fill in the temporal gaps between NOAA and EUMETSAT observations. NOAA, on the other hand, recognizes that these observations have many uses for numerical models, operational forecasting, and other applications, so they purchase these data on behalf of the public.
Microwave imagery also helps us understand the thermodynamic structure of the storm. Here’s an animation of the Advanced Microwave Sounding Unit (AMSU) Band 6, at 54.46 GHz. These are storm-centric images that follow the known center of rotation as it progresses westward. As the storm gets stronger, the temperature in the middle gets warmer. This is because hurricanes are warm core storms, a fundamental characteristic that separates them from the cold core wintertime low pressure systems that dominate weather in the midlatitudes. All of the humid (but unsaturated) air that flows into the center of a hurricane releases extreme amounts of latent heat as it condenses which causes the air at the center to be warmer. Just look at how the warm core emerges over the course of several days of microwave observations. The stronger the system, the greater the difference between the core and the surrounding environment.
Note that the temperatures outside of the core are on the order of -28 C. Clearly, this channel isn’t seeing all the way to the surface but instead is only seeing partway down, to around 350 mb or so. If we look at a different channel, we’ll see a horizontal cross section at a different depth of the atmosphere. Here’s channel 7, 54.94 GHz, for the same period as the earlier loop. Environmental temperatures are much cooler here, signifying that our cross section is much higher in the atmosphere (in this case around 200 mb).
We can put the various channels from AMSU together, in fact, to create a vertical cross section of the storm. This next plot shows the temperature anomalies relative to normal for 6 July. Here, we see that the warm core gets warmer with height from the surface up to around 150 mb or so. Clearly, the higher up an air parcel goes, the more condensation takes place and the more latent heat gets released. This only stops once the tropopause is reached, when the air collides with that super-stable layer and moves outward instead. You can even see that effect with the elevated temperatures between 100 and 200 mb.
As we’ve already discussed, the microwave images are the best tool for diagnosing where the eye of the storm is and how strong the eyewall is. But we’ve also discussed how the microwave pictures aren’t continuous like the geostationary ones are. The ability to monitor the eye characteristics like you can with the microwave imagers but at a geostationary like cadence would be a huge asset to forecasters. This is where the CIMSS MIMIC product comes in. MIMIC takes the irregularly-spaced microwave satellite swaths and applies image morphing techniques to them to, well, mimic the effects of a rotating tropical cyclone, including increasing angular velocity closer to the eye and increasing rotational speed with increasing observed wind speed. It focuses on the 85 to 89 GHz channels from the various microwave radiometers in orbit, as these channels are highly sensitive to precipitation and can be used to create a radar-like product for locations where radar observations are impossible. This animation shows MIMIC in action for all of 4 July 2026 in UTC time. MIMIC stays focused on the center of the storm’s rotation and tracks it across the ocean. Note how the islands of Guam and Saipan appear to move in from the left as the typhoon tracks westward.
Something very interesting is happening at the end of the above loop: we’re seeing an eyewall replacement cycle. Note how the eye gets bigger and the brighter colors disappear right at the end. During a hurricane’s life cycle, the outer bands are robbing the center of energy and momentum as they move inward, causing the original eye wall to die out. The new eye is bigger and the storm is less intense as the pressure gradients are reduced, but it distributes the storm’s heavier impacts over a larger area. Strong cyclones can also re-intensify after an eyewall replacement. The above loop cuts off right as the diameter of the eye is rapidly growing. Let’s see what happens over the next 24 hour period:
The eye seems to decrease in diameter, causing the gradients to reintensify and the winds to increase. This happened right before the storm passed between Guam and Saipan, enabling the typhoon to hit Rota with such ferocity.
Products like these can help us identify the position of the storm and assist in qualitatively assessing its strength. But CIMSS also has a long history in using satellites to quantitatively measure storm intensity. The earliest such products were based on the Dvorak technique. In principal, the Dvorak technique is simple: tropical cyclones of similar intensity all have similar appearance on satellite. Thus, if you can use a satellite to identify what stage a storm is in, you can then correlate that to a reasonable estimate of wind speed and central pressures. This is especially useful in the Pacific, where hurricane hunter flights are much more rare than they are over the Atlantic and thus remotely sensed techniques are desired. Even in places where aircraft reconnaissance is more common, Dvorak enables forecasters to fill in the temporal gaps. The primary issue with the original Dvorak technique is that it reqires human inspection for satellite images which can bring subjectivity and bias. CIMSS developed the Advanced Dvorak Technique (ADT) to automate this process and correct for some biases that became known after the original technique was developed. But CIMSS didn’t stop there: they have continued to develop new techniques that include artificial intelligence and machine learning to better assess tropical cyclone intensity from satellite imagery.
The CIMSS D-MINT product is the culmination of decades of automated tropical cyclone intensity estimates from CIMSS. Here is a time series of D-MINT wind speed estimates that combine the always present geostationary infrared images with observations from the polar orbiting microwave radiometers. The black line shows the JTWC’s estimate of wind speed, which looks to be slightly overestimating the wind speed relative to the D-MINT observations. Of special interest in this plot are the periwinkle observations, which are from the Tomorrow.io satellites discussed above. Here we can see how the storm rapidly intensified from a high end tropical storm to a Category 5 super typhoon in about 24 hours. It stayed at or above Category 5 status for about three days, including its overpass over Rota, before starting to weaken. It’s really interesting to note the fluctuation in cyclone intensity right around 0000 on 5 July, which is the same time as the eye wall replacement we saw above! Given the paucity of aircraft reconnaissance in this basin, these observations are critical for determining the strength of Pacific typhoons, and CIMSS is at the forefront of developing and distributing these observations.
Of course, as we’ve alluded to already, there are other ways of measuring tropical cyclone intensity by satellite. The CIMSS SATCON product blends a number of these different estimates together to produce a best-estimate of satellite-derived tropical cyclone strength. This next figure shows the time series of SATCON for Bavi over the last several days. The thick red line shows the SATCON best estimate while the thin red lines give a measure of the uncertainty about that measurement.
These products are critical for operational forecasting, especially in the Pacific. Models can often struggle in these data-sparse regions, and these satellite-based tools are often the only way that forecasters can validate how well a model has been doing and if it can be trusted in the future. There’s many more products to explore, and this blog post could easily be twice as long if we described them all. Instead, we encourage you to visit the CIMSS Tropical Cyclones website at tropic.ssec.wisc.edu, click on a storm (if one is active) and explore all the products that then come up.