Improvement needed for Tomnod in finding MH370 missing plane
|I can understand how excited when people will be able to handle the satellite remote sensing imagery. The first time I am using satellite remote sensing a decade ago, my purpose was to know how my hometown looks like from the eye in the space. I was so excited when received the satellite image from MACRES, I want to find some useful information from the Landsat imagery. At the end, I did study the landuse changes and land reclaimation using a time series landsat images.
Many people is frustrated with the news of “Nothing was found” even at this 5th day of Search and Rescue operation. They are asking “What else can we do besides praying for MH370?”. Some Google Map and Google Earth active user started to locate the airplane using these user friendly tools, they found the airplane, crash site, wreckage etc. However, they just didn’t realise that Google’s satellite imagery is not real time images which is back dated months or years ago.
The news on Tomnod is a real hit to their question, to their enthusiasm and passion to assist. Tomnod has provided an excellent platform for the public to participate in searching for the MH370 by doing it in a proper manner. It is now overwhelming and the server is always down. You can find Tomnod Malaysia MH370 campaign here
http://www.tomnod.com/nod/challenge/malaysiaairsar2014
As a trained expertise in satellite remote sensing application, I would recommend Tomnod to improve some of the issue below:
- Faster server that can perform huge data handling – many user found the server downtime problem is serious. For life saving or emergency response, we need the server to be responsive.
- Need to limit the user access to reduce the server load
- Need to provide the recommendation of computer system, internet speed etc.
- Enable the image browsing with ease, like auto-navi
- Enable zoom-in, zoom-out function with tool box or icon
- Enable simple image enhancement
- Provide more example of verified object as the reference for other user
- Showing the pointer Latitude and Longitude
- Comparison of Pre and Post Disaster Image
- Discussion group
- Prepare training imagery set to get the participants ready
You are viewing from the Space
Public participation in the satellite image analysis is very interesting. However, if they are unfamiliar with satellite imagery, it might turn them away. Many public having difficulty to imagine that they are actually looking from the space, where the viewing angle will be totally different from our normal vision, they will be disoriented. In reality, satellite image analysis need some professional skill and experience to make a good interpretation on the observation.
Missing Detail Image Information
The satellite image that I am now navigating is taken on 12 March 2014. That’s is the only information I can get. After browsing the image for a while, I noticed that it could be a high resolution Panchromatic image, not the RGB color composite. This satellite imagery information should be provided to the participants so that they can understand what are they looking at.
Boring Satellite Imagery at the Ocean
For the case of monitoring the ocean, the public might feel boring in viewing the imagery as the homogenise color of the blue ocean, white cloud etc (picture below). I spent two hours working on Tomnod, and my vision became “saturated by the dark green and blue color”. On top of that, high resolution sensor commonly share a weakness for ocean monitoring, where they are design to monitor the terrestrial object, their sensor is not sensitive enough for the ocean area. Sometimes, high noise appear in the ocean region.
Difficulties of Image Analysis and Interpretation
- “What is that?”
- “Who can I ask?”
- “Should I mark it even if I can’t understand what I saw? But it seem important”
- “How do they verified my findings?”
- “I saw an airplane above the ocean, do I found MH370?”, then he share it on social media and it go viral
Image interpretation is the most difficult task in satellite remote sensing application. You need to understand what you are looking at, what you are looking for, why you are participating in?
Object identification for the South China Sea area need some experience in this particular part of the ocean. There are various ocean phenomena that could not understand by them, e.g. mangrove, ocean front, algal bloom, oil spill, cloud shadow, marine debris etc. The South China Sea area is in tropics, no floating iceberg, etc. There is a need to prepare a team if expertise who know this ocean very well to assist in the object verification. For example, “if you don’t know the suspicious object, send it to ask@tomnod.com.” And possibly there are hundreds of inquiries need to be look at daily, where it will be a heavy work load.
When participate in Crowdsourcing campaign like Tomnod, the participant need to follow the code of conduct. The participant task is to find suspicious object, but not to become a media center. The information from this campaign if not properly control, could be the source of hoax that can get viral in the internet. It is seriously affect the reputation and credibility of Tomnod. Who will be responsible if this case occur?
Preparing a Training Imagery as what we practice in normal satellite remote sensing analysis might be able to assist the participants perform their task better. It can be design as a quiz, and need to browse the ocean imagery for object. Once detected certain object, they will be requested to classified it, and answer will be given if they do it correctly. For better interpretation, we can actually show them the object photos taken on the ground/ocean. Everyone need to practice with this training imagery before start their own campaign.
The Ocean is Dynamic
Compare to the terrestrial remote sensing application, ocean remote sensing is way much complicate, and need to be handle with care. Even if the object is identify, for example, oil spill, it will be swipe away by the ocean current, and no longer in the same location after a few hours. Sometimes, it submerge and reappear again. The shape of the oil spill could be changed as well. When the “sea truthing” taking place, they might not be able to locate the identified object as there is time lag.
Overwhelming Identification vs. Planned SAR Operation
- What if there are way too many reports of marine debris, wreckage, oil spill that submit to the Search and Rescue authorities?
- Which of the objects that they have checked in the last 5 days of operation?
- Which are the new identified objects?
- Which are important?
- Who should be handling the search? Malaysia, China, Vietnam?
- Who should they send the feedback to if the object is found?
If we are expecting the authorities to check every of the locations submitted by this campaign, it will be a huge burden added on their current planned task.
I would like to apology if this article causing disagreement or bad image on the current Tomnod platform. My purpose to highlight the Tomnod issue here is hoping we as the “crowd” that participate in the campaign, can contribute successfully towards the searching of MH370 with a more proper planned manner and reducing the error that might cause disturbance to the SAR operation.
This project is using the resource from Digital Globe, Tomnod, world media and all the time and effort input from the participants. We all share the same responsibility on the success this crowdsourcing platform, which will be something important in the future disaster as well, not just for the MH370. This is just the beginning. We need to be clear that we contribute just a small part to the huge SAR effort, we shouldn’t be overwhelming by our voluntary effort.
We can only prepare the RIGHT information, but we need to pass it to the RIGHT person, at the RIGHT time, for the RIGHT decision.
#PrayForMH370
Latest News Update on the Application of Satellite in Detecting MH370
Missing MH370: Bombardier dispatched to site of Chinese satellite images, confirms Hisham
Missing MH370: Vietnam yet to find objects seen by Chinese satellite