[SVC46-02] Problems on high-accuracy topographic measurement near the crater of active volcano
Keywords:UAV, measurement, SfM/MVS, volcanic eruption
1.Introduction
When a volcanic eruption starts, it is very important to measure the amount of topographic change near the crater. However, it is also dangerous to enter the vicinity of the crater during an eruption or fly over the sky. For this reason, measurements using artificial satellites, photography from high altitudes, oblique photography from a sufficiently long distance, and the like have been performed. Therefore, the MEXT volcano project has been conducting SfM / MVS surveys from oblique low-altitude photography using UAVs. He has also created models using images from manned helicopters and Cessna. The results obtained have been reported to the Volcanic Eruption Prediction Association. However, during this time, there are still many problems in terms of providing results in a short time due to constraints such as weather conditions, takeoff and landing points, flight time, and altitude to ground. Unfortunately, there were cases where a model could not be created. Here, we summarize our efforts so far in terms of required time and accuracy, and summarize the issues in future research. 2. Shooting
2. Photogrametry
2.1 Shooting from UAV
Shooting UAV shots around the crater were taken at Mt. Aso in 2016, Mt. Mihara in 2017, Sakurajima in 2018, and Mt.Mihara in 2019. Of these, the autonomous flights are Aso in 2016 and Mt. Mihara in 2019. Miharayama in 2017 and Sakurajima in 2018 were shot manually.
2.2 Shooting from manned helicopters and Cessna
The ones with good 3D terrain models used wide-angle lenses and had a sufficiently high lap rate. Those that failed were those that used telephoto lenses, those that changed terrain during shooting, images that were erupting, and images that were far away.
3. Doming
Because the GCP could not be obtained near the center of the crater, the completed model sometimes deformed into a dome shape.
4. Alignment and terrain change calculation
For the alignment of the model and the calculation of the amount of change in the terrain, precise data that has already been obtained is indispensable, and a resolution of 1 m or more is desired. However, in the case where the preparation has not been completed before the eruption, it is necessary to check the laser measurement status and apply for licensing or borrowing, which is a problem for calculating the amount of topographic change. In Izu Oshima, eruptions near the central crater of Mt. Mihara were repeated, so I was able to shoot in advance and acquire topographic data. It is desirable to carry out all craters that are likely to erupt. In addition, when the GCP cannot be obtained or when the position accuracy is insufficient, the 3D model is often distorted by the SfM / MVS method. In such a case, it takes a lot of labor and time for alignment and distortion, which is a big problem.
5.Summary
We conducted a demonstration experiment to apply the SfM / MVS topography from oblique photographs to the crater of an active volcano. To summarize the four-year results, it is practical enough to be able to get the GCP right. In the future, by performing precise GCP measurements at locations where eruptions are likely to occur, it may be more accurate than laser measurements.
When a volcanic eruption starts, it is very important to measure the amount of topographic change near the crater. However, it is also dangerous to enter the vicinity of the crater during an eruption or fly over the sky. For this reason, measurements using artificial satellites, photography from high altitudes, oblique photography from a sufficiently long distance, and the like have been performed. Therefore, the MEXT volcano project has been conducting SfM / MVS surveys from oblique low-altitude photography using UAVs. He has also created models using images from manned helicopters and Cessna. The results obtained have been reported to the Volcanic Eruption Prediction Association. However, during this time, there are still many problems in terms of providing results in a short time due to constraints such as weather conditions, takeoff and landing points, flight time, and altitude to ground. Unfortunately, there were cases where a model could not be created. Here, we summarize our efforts so far in terms of required time and accuracy, and summarize the issues in future research. 2. Shooting
2. Photogrametry
2.1 Shooting from UAV
Shooting UAV shots around the crater were taken at Mt. Aso in 2016, Mt. Mihara in 2017, Sakurajima in 2018, and Mt.Mihara in 2019. Of these, the autonomous flights are Aso in 2016 and Mt. Mihara in 2019. Miharayama in 2017 and Sakurajima in 2018 were shot manually.
2.2 Shooting from manned helicopters and Cessna
The ones with good 3D terrain models used wide-angle lenses and had a sufficiently high lap rate. Those that failed were those that used telephoto lenses, those that changed terrain during shooting, images that were erupting, and images that were far away.
3. Doming
Because the GCP could not be obtained near the center of the crater, the completed model sometimes deformed into a dome shape.
4. Alignment and terrain change calculation
For the alignment of the model and the calculation of the amount of change in the terrain, precise data that has already been obtained is indispensable, and a resolution of 1 m or more is desired. However, in the case where the preparation has not been completed before the eruption, it is necessary to check the laser measurement status and apply for licensing or borrowing, which is a problem for calculating the amount of topographic change. In Izu Oshima, eruptions near the central crater of Mt. Mihara were repeated, so I was able to shoot in advance and acquire topographic data. It is desirable to carry out all craters that are likely to erupt. In addition, when the GCP cannot be obtained or when the position accuracy is insufficient, the 3D model is often distorted by the SfM / MVS method. In such a case, it takes a lot of labor and time for alignment and distortion, which is a big problem.
5.Summary
We conducted a demonstration experiment to apply the SfM / MVS topography from oblique photographs to the crater of an active volcano. To summarize the four-year results, it is practical enough to be able to get the GCP right. In the future, by performing precise GCP measurements at locations where eruptions are likely to occur, it may be more accurate than laser measurements.