Tropical Storm Cempaka 2026
Last Modified:
Live tracking map, satellite, radar, and forecasts of Tropical Storm Cempaka 2026 (台風32号) south of Japan. Current wind speed 110km/h. Max 185km/h.
Cempaka (Japanese Typhoon Number 32) is located 469 km south of Iwo To, and has tracked west-northwestward at 24 km/h (13 knots) over the past 6 hours. Maximum significant wave height is 6.7 meters (22 feet).
Cempaka is forecast to continue tracking northwestward, along the southwestern edge of the subtropical ridge, over the next 36 hours. Afterward, an incoming longwave trough is expected to break down the western portion of the ridge, causing Cempaka to turn slowly northward over the next 2 days before getting picked up by the trough and accelerating east-northeastward.
There continues to be some uncertainty revolving around the speed of advance after 2 days. Some models show the trough zooming by and repositioning the steering ridge eastward in a way that prevents Cempaka from recurving. Additionally, a separate ridge centered near southern China could further inhibit poleward movement depending on how far eastward it is able to build.
The other scenario is a bit more straightforward, and is currently the one reflected by the JTWC forecast. This involves Cempaka simply gaining enough latitude for it to round the ridge axis, turn northeastward, and eventually become absorbed by the frontal boundary associated with the mid-latitude trough.
In terms of intensity, Cempaka is forecast to continue intensifying over the next 2 days to a peak of around 185 km/h (100 knots) as the environment remains highly favorable.
After 2 days, shear and dry air will begin to increase, causing the system to begin weakening. In around 60 hours, the environment will quickly become much more hostile, with strong (over 35 knots) westerly shear and significant dry air intrusion.
Cempaka is then forecast to begin extratropical transition at 3 days with completion occurring in around 4 days, as the vortex becomes stretched and absorbed into the frontal boundary.
Model track guidance is in good agreement only through the first 36 hours with a cross-track spread of 204 km at that time. After 36 hours, models begin to diverge greatly.
The GFS and GEFS mean are now the two primary outliers, showing the slowest track by a significant margin. It is also worth noting that the GFS is the most equatorward model at 2 days, which is about the time of the inflection point coinciding with the northeastward turn.
The ECMWF is also among the slowest and equatorward trackers. On the other hand, GALWEM, JGSM, and the EC-AIFS comprise the fastest group and are all east of 150E at 4 days. Then there is the Google DeepMind and EC-AI ensemble means that are situated between the two extremes.
The JTWC track forecast is placed near the multi-model consensus over the next 2 days, then closer to the AI-consensus and Google DeepMind over the next 4 days with low confidence.
Intensity guidance continues to generally agree on rapid or near-rapid intensification over the next 2 days with multiple RI aids triggering this model run.
Models then also agree on weakening following the next 2 days. HAFS-A remains somewhat of an outlier, showing a delayed and more gradual weakening trend. This is attributed to it being amongst the most poleward and fastest models, allowing for it to take full advantage of the baroclinic forcing and strong temperature gradient rather than becoming absorbed by the frontal boundary.
The JTWC intensity forecast is aligned with the consensus over the next 2 days and then slightly lower over the next 4 days to offset HAFS-A.