Tuesday, March 15, 2011

Japan quake data delivery is an ongoing challenge

GeoEye post-quake imagery prepared by Google Crisis Response Team
We're part of a great distributed team doing map support for the search and rescue task forces in Japan.  One of the big challenges we've all been working on is how to get updated data to the field. In the early phases of a response you don't know the situation, so you try everything and stick with what works.  In that spirit, here's a story of some of the comm approaches we've worked on.

  • Hand off hard drive before departure. This was the first thing we did and so far the one that worked best.  We managed to wire 3 GB of map data to JPL in Pasadena, which was on the way from CA-TF2's staging area to their departure point at March Air Reserve Base. That let our guys at JPL do a physical handoff of a hard drive.  Since then we've heard from the team in the field that they made good use of the data during their flight to pre-plan their operations, so thanks again to everyone who helped make that happen!
  • Courier data from Google Tokyo. We figured out on Sunday that there was a solid link between Google's offices in Mountain View, California and Tokyo, sufficient to send ~ 30 GB of data in a few hours.  The plan was to courier the data by hard drive from Tokyo to the operational area at Ofunato, a small town about 100 km north of Sendai. Unfortunately, things stalled there. We've been advised that road conditions are very poor with some bridges out, emergency vehicles and aircraft are all in use, and some roads may be closed to civilian vehicles. As of now we're continuing to stage data to Tokyo and hoping conditions will improve.
  • Send data via satellite internet. The team in the field has a satellite dish that uses the BGAN Inmarsat service to give them a direct low-bandwidth internet connection. For now this is our only working data link and we're trying to generate really tightly focused lightweight maps that cover the high priority areas. I don't have the specs on their particular BGAN device but we've been advised that 50 MB is the upper bound file size for practical transfers under good conditions, and unfortunately conditions are poor. That's probably because the satellites are overloaded, just like cell phone networks get overloaded after a disaster.
  • Courier data from Misawa Air Base.  Misawa is located at the northern end of the main Japanese island of Honshu and it's where the teams originally arrived and started their drive down to Ofunato. We've established a data link to Misawa and are waiting to hear more about the bandwidth. It may be easier to courier data from Misawa rather than Tokyo, especially because we may be able to piggyback a hard drive in the delivery when they resupply the Air Force security team that's accompanying the search and rescue task force. But so far we don't know when a resupply trip will happen.
  • Get help from Cisco. The Cisco Tactical Operations team is investigating a couple of options. One is to deliver data through Cisco's Japan offices and local staff that are providing comms to the Japanese government. Another is to get a higher-bandwidth satellite dish to the field team. We just got in touch with them Monday night.
One really important point about data transfer: the rapid growth of storage media sizes and satellite imagery resolution is an ongoing tech trend and wireless comm bandwidth isn't keeping up. So the trade-off between physical media delivery vs. wireless transfer will probably favor physical delivery more and more as time goes on, and we need to pre-plan how to make that happen for future deployments.

So, hope that is food for thought and I'm sure improving the connectivity will be a fun conversation during debrief.

Sunday, March 13, 2011

Calling for mapping help for Japan earthquake

We're helping to provide map support for the FEMA Urban Search and Rescue Task Forces deployed to help with the Japan Sendai Quake.  I'm looking for help or ideas related to challenges listed below.

Please post comments on this blog post where everyone can see them.  Thanks!

FEMA Urban Search and Rescue California Task Force 2 (LA County) staging for the trip to Japan

  • Cacheable data sets: We're looking for mapping data sets that can be cached and delivered on a hard drive for offline use.  For reference, the task forces will be splitting into small groups which will have spotty internet connectivity in the field.  All their laptops have at least the basic version of the Google Earth client.  We've already grabbed small areas of OpenStreetMap data, and Google has pulled in lots of data sets, but there may be some great resources we're not aware of.
  • Data delivery: We're looking for ways to deliver the data.  We were able to hand off a hard drive Friday evening before they flew out of LA but currently we don't have any pipes.  One possibility is delivering data to Google's Tokyo office and handing off from there, but we haven't worked out the logistics.  Looking for creative solutions.
  • Scripting support: The publicly accessible versions of some data sets need to be massaged into something that's usable offline.  If you have behind-the-scenes access to rsync or great map scripting chops to reformat data sets, we might be able to use your help.  An example data set is this GeoEye imagery KML prepared by the Google Crisis Response Team.  How do we cache it?
  • File sharing admin support: I set up a 1 TB Apache+WebDAV file sharing server on Amazon Web Services (EBS) to help us collect data.  I haven't configured backups yet.  I am not an admin, there are probably other things I forgot to do. If at this point you are thinking "what a moron", please get in touch :)
Other notes on the situation:

  • The latest I have on Task Force 2 is that they will be assessing damage in Ofunato tomorrow morning.
  • Ways to get in touch with me for more information:
    • Email: trey.smith@nasa.gov
    • Twitter: trey9000
    • AIM: treyontheoutside
    • Google Talk: trey.smith@gmail.com
    • Cell: 412-657-3579

GeoCam assisting Japan Sendai quake response

We've been providing map data to the FEMA Urban Search and Rescue Task Forces deploying to Japan to help after the Sendai Quake.  The mission is called Operation Tomodachi (Japanese for "Friendship"). They will likely have spotty connectivity and low bandwidth while deployed, so we need to gather the maps they need up into big chunks of data that can be cached on their local disks.

Tsunami wave of mud hitting a city in northern Japan

Area around Sendai where we cached OpenStreetMap data

Our first data delivery was at around 7:30 PM PST Friday, when we managed to push a bunch of map data to JPL and some friends down there got it onto an external hard drive and handed it off to a member of California Task Force 2 (LA County) before their group departed for Tokyo.  We're still trying to figure out when our next opportunity to send data will be.  For the moment we're collecting data on a server in the  States so we're ready.

This has been a great team effort involving a bunch of NASA folks, OpenStreetMap folks helping us grab their data, Google offering free licensing on their Earth Enterprise product and negotiating data release of Zenrin street and GeoEye satellite data, and even a local Best Buy in Pasadena donating a couple of hard drives.  It is so amazing to work with people who will take that call and instantly drop what they're doing to help out with something like this. I'll have to give it a real write-up when things cool off.

Thursday, December 2, 2010

The GeoCam Motley Crew

At the GeoCam project, the mission has remained, but the people involved often change. We thought that it might be a good time to introduce some of the people making GeoCam better everyday.

Trey Smith leads the GeoCam Project and has been developing technology for disaster response since he joined NASA in 2007. His goal is to make disaster information systems cheap, usable, open, and interoperable so that responders and ordinary citizens can collaborate more effectively. As a result of Trey’s efforts, GeoCam Mobile has been deployed to several major California wildfires and to the 2010 Haiti earthquake. There is no better way to understand emergency response than to immerse oneself in the process. So, in addition to his role at NASA, he is also training to be a Technical Information Specialist with the California Task Force Three FEMA search and rescue team.

Eric is an avid mobile and web developer, having worked with J2ME and Google's Android mobile framework since its early beta days. He led the effort to create the I'm OK! notification system for use in emergency situations, which won First Prize at the inaugural Random Hacks of Kindness code jam. I'm OK! was subsequently deployed with World Bank field agents performing aid work in Haiti after the 2010 earthquake. In addition to his efforts on GeoCam, Eric works as a robotics engineer building systems that help explore and understand remote environments.

Ted Morse works on many projects as a computer scientist for the Intelligent Robotics Group (IRG) at NASA Ames Research Center. He is the lead developer of phone-side applications for GeoCam. While his GeoCam efforts have focused on increasing situational awareness within the disaster response community, he truly believes that this technology can help the general public by combining GPS-enabled mobile platforms with small web applications. Prior to his role with GeoCam, Ted worked on the GigaPan Project, a gigapixel photographic system complete with a web interface for acquiring, stitching and viewing panoramas.

Mark joined GeoCam just a month ago, but he has worked for over a decade to bring robots and other technologies to emergency response and search and rescue. He is certified in multiple aspects of search and rescue including technical search, technical rescue, hazardous material response, and is a nationally certified fire fighter. Before joining the GeoCam team, he was a search specialist for the Massachusetts FEMA search and rescue team. Mark was a technical search robot operator during the World Trade Center Disaster and was a technical search specialist for Florida Task Force Three during the Hurricane Katrina response in Biloxi, Mississippi. His recent research leverages multi-touch tabletop, cell phone, and robot technology to bridge the gap between responders in the field and the incident command structure that supports them.

Wednesday, November 17, 2010

GeoCam new app ideas for 2011

In the coming year, the GeoCam Project will develop several small apps instead of one big app.  Here are some ideas for new apps we might create in 2011.  We'd love to hear your ideas!

App 1: Group coordination.  Monitor team member positions, send geotagged messages, receive geo alerts when somebody enters a danger zone.

App 2: Search coverage heat map.  Initialize a heat map of likely victim locations from sources like Google Local business locations and USGS earthquake intensity.  Any search group, from government to an NGO or CERT team, can contribute to the map, or enter its location and skill level to learn what area it should cover to best serve the coordinated effort.

App 3: Community health and safety forum.  People at the disaster scene post questions or problems by text message.  Anyone can help, by answering a question, translating a request, extracting a place name so the request shows up on a map, classifying what help is needed, or flagging the request for urgent assistance.

Monday, November 16, 2009

Random Hacks of Kindness First Prize!

GeoCam Disaster Response Project members were on the First Prize team "I'm OK". From left to right, Chris Cinelli, Otavio Good, Eric Park, Trey Smith, and Ted Morse. (Our iPhone hacker Hansel couldn't stick around for the awards.)

The Random Hacks of Kindness disaster relief codejam was this weekend. RHoK was an unconference run by all the attendees, not just a few organizers. The goal was to hack together new tools to help in disaster relief. Before the conference, people submitted problem definitions -- their sense of gaps in disaster relief where innovation could make a difference. At the conference we broke up into teams to create open source software and process ideas to address those problems.

Our team "I'm OK" addressed the problem that urban cell networks get bogged down in disasters with people calling their loved ones to tell them they're ok. To address this, we created a mobile phone app called "I'm OK" with just one button. It sends a single SMS to a server that in turn notifies all your friends and family that you're ok through whatever services you use to connect with them, like SMS to their phones, Facebook status update, Twitter update, etc. This is particularly useful because SMS messages don't burden the cell network as much as phone calls, and can often get through when calls can't. Half of the team was GeoCam project members (Eric, Ted, and myself), and we picked up some new friends along the way (Chris, Otavio, and Hansel).

Here's a few more photos.

FEMA Director Craig Fugate gave the keynote address, where he introduced the problem of encouraging people to make a family disaster plan before the disaster strikes.

I talked about NASA's involvement in disaster relief.

My slides.

The team was pretty exhausted by Saturday morning, but the free food helped :)

The presentation Eric gave about "I'm OK". Otavio did most of the slides.

The I'm OK app running on Android and iPhone. (Credit: Elinor Mills, CNET)

Friday, October 30, 2009

Loma Fire 2009/10/27

Santa Cruz County Sheriff Phil Wowak (far right) listens as a resident says he did not receive the evacuation warning from the county's reverse-911 system.

Tuesday was our last day at the Loma Fire. With activity slow on the fire itself, the most interesting event was the community information forum in the evening, where all the public agencies involved in the response discussed what happened with local citizens and listened to their concerns.

It was great to see this, because we're embarking on a new initiative called the Citizen Responder Project, which aims to empower citizens to better help each other in emergencies through use of information technology and social media (more on that in future posts). This fire may become our first case study.

The forum started with several presentations explaining things like how the fire progressed, which roads were closed and why, and how the evacuation warning was sent out using the county's reverse 911 system.

A CAL FIRE representative talks about the history of fires in the area. The Loma Fire is in red and the 2008 Summit Fire is in yellow. (It was neat to see this map, as I was in the GIS trailer when Joe Larson was preparing it.)

Then there was a wide ranging open discussion. Some residents were very expressive about not liking road closures that kept them away from their homes. Very calm professional responses from the presenters kept that thread from getting out of hand.

The part that interested me the most was a discussion of how citizens could stay informed. The people who listened to the radio complained about the quality of information. For example, one woman spent an extra hour driving home because she couldn't figure out which roads were closed. Most media reports were targeted at audiences outside the community and left out the details local residents care about most.

Other residents said that their local Facebook friends were the best source of information. Alex Leman's Loma Prieta Fire Twitter feed was also mentioned as a good way to keep up. Somebody asked how they could stay updated when they were away from home, and the advice was to call a neighbor who's always on the computer. Other communication methods were brought up, like a call center to answer questions or a physical bulletin board for official postings. It was interesting to see how different residents had different levels of comfort with official vs. informal channels, which seems to be related to computer literacy and level of social involvement in the community.

Loma Prieta Fire Chief and Community Emergency Response Team organizer Alex Leman (top left) tells citizens how they can get involved in response.

At the end of the evening I talked to a few of the presenters and got some business cards, so as soon as we figure out what questions to ask we should be able to follow up with them to get more impressions of their experience with this response. The forum was a great wrap-up to our involvement in this fire.