Sakurajima from three shores, and the road cameras on the island itself
At 10:07 this morning Japan time the camera at Tarumizu Arasaki had Minamidake's cone filling the right of the frame, a thin plume leaning off the summit, and rows of fish pens on the water in front of it. Four minutes later, from Chuo-ko Shinmachi on the Kagoshima city waterfront, the same mountain was square-on across the bay with a boat crossing underneath it and the plume still leaning the same way. That is the thing about Sakurajima that the single-camera listings miss. It is one volcano with a bay around it, and which side you watch from changes the whole picture.
The Japan Meteorological Agency runs three cameras on it, and most of the pages that come up for a Sakurajima live cam show you one feed from Kagoshima city and stop there.
The three the weather agency runs
JMA publishes them under volcano number 506, one page each, and all three answered 200 when I fetched them on 2026-10-29.
- Tarumizu Arasaki is the one to open first. It sits closest of the three and looks across the Tarumizu aquaculture pens at the south-east flank, so the cone is large in frame rather than a shape on the horizon, and on a clear morning you get the plume against blue sky instead of against haze. The frame I read was stamped 2026/10/29 10:07:58. Its caption is burnt in as katakana, サクラジマ アラサキ, which is how you can tell it apart from the other two at a glance.
- Ushine is the wide one. From the eastern shore the whole island lies out in profile, the long ridge and the skirt of lava down to the water, and this morning there was a mackerel sky over the top of it. A utility pole stands in the left of the shot and always has. Frame stamped 2026/10/29 10:07:53.
- Chuo-ko Shinmachi is the postcard. This is the view from the Kagoshima city side, the one on the tourist posters, with the ferry route running across the bottom of the frame. Stamped 2026/10/29 10:11:55.
All three post a new frame every two minutes. I did not take that from a spec sheet - the JMA page for each camera carries the last hour as a strip of dated images, and on 2026-10-29 the Tarumizu Arasaki strip ran 09:08, 09:10, 09:12 and on through to 10:08 without a gap. Two minutes is quick for an agency still camera. Most of the river and road cameras we hold in Japan are nothing like it.
The catch is daylight. These are ordinary visible-light cameras, and from about half an hour after sunset until half an hour before sunrise all three are a black rectangle with a timestamp on it, so a Sakurajima habit is a morning habit if you are in Europe or the Americas. JMA also warns on each page that distribution can stop during maintenance on the video system.
The road that goes round the island
Sakurajima is joined to the mainland at its south-east corner, and National Highway Route 224 runs right round the south shore underneath the active craters. The Ministry of Land, Infrastructure, Transport and Tourism puts cameras along it, and they are filed as traffic cameras rather than volcano cameras, which is why nobody thinks to look at them and why none of the aggregator pages list one.
They are not pointed at the summit. They are pointed at the road, and that is the point: the road is the thing the volcano acts on. 赤水遮断機, the Akamizu barrier gate at 31.5756, 130.6100, is one of several gate cameras on the island, and the gates exist because debris comes off the flank into the channels that cross the road. 有村 sits at Arimura on the south-east shore, below the slope the lava came down. 道の駅桜島 looks over the roadside station on the west side, which is where you would actually stop if you drove it.
Watch any of them on a day the wind is off the summit and you can see what the city means when it talks about ash: the surface goes grey and the road markings go soft. On the opposite shore the same system runs 牛根麓口 and the other Route 220 gates along the Tarumizu waterfront.
I have not timed these. The catalogue carries an interval for the source and I would rather measure one than repeat it, so take the road cameras as slower than the JMA three and leave it there until I have sat on one.
Across the bay, and the one at the airport
Two more worth knowing. The Sakurajima Volcanic Summit Panorama is a continuous stream rather than a still, shot from the city side about ten kilometres out, and it is the one to leave running in a tab if you want to catch an eruption column rather than infer one from a plume. Sakurajima Crater Kagoshima is a second streaming view from the same side of the water.
Then there is Kagoshima Airport, 23.8 km north-east of the summit by our own coordinates, measured on 2026-10-29. It is not a volcano camera and nobody files it as one, but ash closes that airport, and a glance at the apron tells you something the volcano cameras cannot.
Counting everything inside 40 km of the summit, our catalogue held 78 live cameras there on 2026-10-29, most of them road and river cameras on both shores of the bay rather than anything pointed at the crater (Japan). The density around Kagoshima Bay is unusual for a Japanese prefecture, and it is not an accident. A working volcano sits across the water from a prefectural capital, and the monitoring reflects that.
We re-check every camera on the site once a day, which is how the uptime figure on each page above is built, and the three JMA pages are checked the same way as everything else. What I want next is the real refresh interval on the Route 224 gate cameras, timed rather than inherited from the source. If you know the publishing cycle MLIT uses for the Kyushu road cameras, I would like to hear it.