In July 2023 another work trip took me to Hokkaido, Japan. I stayed in Obihiro, a city of about 165,000 people. I brought along a RTL-SDR dongle and mag-mount antenna, and if I had some free time, I would try and decode nearby radiosondes.

From my earlier trip back in September 2022, I knew there was radiosondes launched from Sapporo twice a day. However, it was unclear if the Kushiro autolauncher had been replaced after the fire.

After tracking the radiosondes in Hokkaido, I would travel to Tokyo for a few days and try and track them in the city, labeled "Tateno" in the above map.

Tracking Hardware

The equipment I use is very simple but effective, just a RTL-SDR Blog v3, a mag-mount antenna, laptop, and car power supply.

Pro tips for using this mobile setup:

• Have someone else drive the car, or pull over whenever viewing the laptop.
• Use a short USB extension cable instead of plugging the RTL-SDR dongle directly into your laptop. This prevents torque on the laptop's USB port.
• Use a car DC-DC converter for laptop power. Nothing increases blood pressure more than running out of batteries when the sonde is at 1,000 meters elevation and falling fast.
• Viewing the radiosonde on tracker.sondehub.org takes significant bandwidth (~500 kBytes/sec), which chews through a lot of mobile data quickly. If you search for the sonde you're tracking, it will be the only thing shown on the map, saving lots of mobile bandwidth.

Running the docker container version of radiosonde_auto_rx is the easiest and best way to run the software. Nothing to build or install, and no dependency hell.

Morning Launch: Tracking and Recovery

On the morning of July 30th, I drove to the Tokachi Hill Observatory to view the sights while waiting for the (0000 Midnight UTC) radiosonde to launch.

While the overlook is surrounded by lush foliage, there was also a hornets nest nearby, and the hornets were about 2 inches long. The view also wasn't great due to the smog. Looking at the RF spectrum scans, there were very strong signals in the center of the 400-406 MHz radiosonde frequency band. Was this a new type of radiosonde, or some other service transmitting in this band?

I drove away from the tall broadcast antenna nearby, and the wide offending signal rapidly disappeared. After moving away from the nearby noise source, I quickly decoded the Vaisala RS41-SG radiosonde launched from Sapporo.

The maximum altitude for this radiosonde that I received was 24695 meters (~81,000 feet). However, I stopped receiving the radiosonde for about 12 minutes near maximum altitude due to technical difficulties, so the actual maximum altitude is higher.

After watching the radiosonde fall for a while, I noticed that as the altitude decreased the descent rate also decreased. Just after popping, the decent rate is was about 11 m/s (36 ft/sec), which is what I typically see at my station in San Francisco throughout the entire descent (unless it's a floater!). However, as the altitude descended below 10,000 meters (~33,000 ft), the decent rate slowed to about 4 m/s. Was the latex balloon inflating itself, increasing drag? Or was there a parachute?

I looked at Sondehub, which runs forward predictions as a radiosonde is falling, based on the decent rate. Predictions showed that this radiosonde would cross over the mountains and land in the fields outside Shintoku, about 30 minute drive northeast of Obihiro, where I was located. The race was on!

The last packet I received when I was driving was at 621 meters, and about 30 minutes later I was in the vicinity. Immediately I started receiving the radiosonde, showing it was on the ground in a nearby field.

The field where this radiosonde landed was full of white mustard plants (why can't the radiosonde land in the barren field next door?).

After looking at aerial imagery and the topology of the land, walking around along the creek on the north side seemed to be the best approach. The plants were only about 0.5 meters tall (~1.5 feet), so it was pretty easy to get to the immediate vicinity of the sonde, but the mustard plants were just tall enough to make finding it difficult. Also, the datums of my phone and the GPS receiver in the radiosonde are different, so you're going to be off by a few hundred meters.

The radiosonde did have a parachute! That's what was causing the very slow decent rate, allowing the radiosonde to drift long distances after balloon burst. Typical American radiosondes (without a parachute) take about 30 minutes to fall to the ground from 30,000 meters, but this one took over an hour. Aside from the parachute, it is a standard RS41-SG radiosonde.

And in typical Japanese fashion, next to this field there is a Monument to the Pioneers who settled this land, celebrating the 100th anniversary of pioneering.

Evening Launch: Not Recovered

After getting back to the hotel, I ran the predictions on Sondehub for the next 7 days. Marker 1 shows the path and landing prediction of the evening (1200 UTC) radiosonde.

Not surprisingly, the evening launch was also predicted to land near Obihiro. After dinner I loaded up radiosonde_auto_rx and watched the radiosonde fly, and this time I left early enough to be in the area when the radiosonde was landing.

I was only a few hundred meters away from it when it touched down. But instead of landing in the barren field next door, it landed in a corn field with stalks that were over 2 meters tall. Since it was very late at night, I decided to go home without a successful recovery.

Kushiro Radiosonde

The next day I tried to receive a radiosonde launched from Kushiro. Within radiosonde_auto_rx, I set the "never_scan" option to 405.5 MHz, which is the frequency of the Sapporo radiosondes. I spent the whole morning listening, but nothing was heard, so I'm assuming that the Kushiro station is still out of service from the earlier fire.

Tokyo Radiosonde Reception

I had a few days in Tokyo after my Hokkaido work, so I threw my mag-mount antenna on the balcony to see what I could receive. The Tateno launch site is about 50 km (30 mi) north of downtown Tokyo, where I was staying.

This antenna was very compromised, as it was on the other side of the building from the Tateno launch facility. All of the radiosondes that I received were either through the building (not likely), or reflected off some other building nearby.

But even with the non-optimal antenna placement, I was able to receive two radiosondes over the course of three receiving sessions. Both radiosondes were Meisei iMS-100 models (pdf).

I have never seen one of these iMS-100 radiosondes in real life, so it would have been cool to recover one of those. Next trip!