“Statistics from Space”: Fun with the Gates Foundation and Data as a Public Good
November 11th, 2008by Sean Gorman
I had the good fortune last week to be invited to the “Statistics from Space” workshop held by the Gates Foundation. “Statistics from Space” is a some what misleading name for the subject of the meeting: using remote sensing to help sustainable agriculture with a specific focus on Sub-Saharan Africa. It was a fascinating collection of experts in remote sensing and agriculture policy from the public and private sector with both domestic and international representation. I’d done some basic remote sensing classes in grad school and was definitely out of my league, but it was great getting to listen to real experts talk about the state of the art and how remote sensing could be leveraged to help promote sustainable agriculture.
David Lobell of Stanford gave a great talk on some of the latest techniques for estimating crop yields through work they had done combining remote sensing with ground surveys in Mexico. Javier Gallego of the European Commission discussed their work in the same area. Gilberto Camera provided a state space agency perspective on running a remote sensing program from womb to tomb for Brazil. It was great to hear his efforts to make the imagery available freely to the public and the software available under open source licenses. Brad Doorn of the USDA provided a glimpse into the work surrounding traditional agriculture statistics and also clued me into the PSD (Production, Supply and Distribution) database. We’ll be adding it to GeoCommons shortly.
Kevin Little (Surrey Satellite) and Greg Koeln (MDA Federal) gave insightful updates on the latest remote sensing advances from the private sector. Naman Keita and Jeff Tschirley of FAO (United Nations) gave an “on the ground” and policy perspective on sustainable agriculture efforts in Africa. Kass Green gave her perspective and impressive experience on the intersection of remote sensing technology and policy. Finally Joanne Gabrynowicz provided her world leading experience on the legal aspects of remote sensing. Reminding us all there is a difference between data being openly accessible to the public and date being available free of charge. Governments can and often do charge for data as part of “cost recovery”, but that does not mean the data is not openly accessible.
I hope I did not leave any presentations out, but I wanted to spend a bit of time on two presentations in particular. The first was Gerry Nelson’s of IFPRI who provided and economist’s view on agriculture statistics. Specifically he delved into the idea of data as a “public good“. If you dozed through that part of economics class a public good is where the “consumption of a good by one individual does not reduce availability of the good for consumption by others”. A sunset would be a public good and an apple would not be. Gerry elegantly pointed out that data falls under this category. Any one person gaining benefit from a set of data does not reduce the availability of the data for consumption by others.
This is a big part of the philosophical motivation behind GeoCommons and Gerry felt the same held true with agriculture data and remote sensing data. Fortunately this is coming to fruition with Brazil’s remotely sensed data which is freely available and now American Landsat data being put in the the public domain free of charge. The more data that is freely available to be combined in new innovative ways the bigger the network effect we all enjoy and the greater the positive externalities for the public good.
The same concept was embodied in Enrica Porcari’s (CGIAR) talk on a new Gates funded project called the AgCommons. Leveraging the same concept of data as a public good - the project is aggregating statistical data on agriculture to make it easily available to those interested in promoting sustainable agriculture. The hope is to make data available not only to governments but also farmers, agri-services and agri-businesses in Africa. It is really encouraging to see the concept of data commons spreading and data as a public good being operationalized. USAID’s “Global Development Commons” is another great example. We’ll keep posting on other success stories as we hear of them and please pass along any you might come across.
Popularity: 11% [?]
November 11th, 2008 at 11:56 pm
Sean,
Nice recap of last weeks’ meeting. I just wanted to add that the AgCommons program includes more than top down effort to collect agricultural statistics. It takes a very user centered approach to collecting and delivering information down to the individual farmer and farmer group level in Africa.
November 12th, 2008 at 3:08 pm
We are all children of the incredible legacy of the Landsat program.
However, I still believe there is an important place for an accountable framework, even if no actual currency trades, to track usage. Indeed, we are now seeing the importance of “ecosystem services.”
Really valuable data, like GeoEye, Digital Globe, RapidEye, and others will never be free; only selected scenes will be freed of their intellectual property restrictions. if there were such a framework, perhaps programs might trade “Digital Earth bucks” or something, and have hard metrics. If the data is a common good, in the Darwinian struggle for resources, donors want hard metrics of good it is!
November 13th, 2008 at 10:23 am
I strongly support free distribution of data collected from government sensors. However, in the case of the ‘free’ Landsat data, this freedom comes at a high price. USGS will no longer distribute raw Landsat data and other lower level projects, which up to this point have been available for purchase for a nominal fee (~$500 and can be redistributed to other users).
The data that USGS will offer for free is a specific data product (L1T - Terrain Corrected) that has been processed with a specific choice of parameters: resampled to a UTM grid using cubic convolution resampling. Resampling satellite data, by cubic convolution or other sampling methods, smooths the dataset and removes some of the ‘information content’ contained in the original data. While this likely does not matter to a user that just wants a ‘pretty picture’ image as a base layer, it is of great concern to researchers that use Landsat data for quantitative analyses (such as classification, change detection, feature detection, etc.). While these analyses can be (and are) performed on the L1T data product, the results will tend to be less accurate and precise than working with non-resampled data products.
In summary, USGS has moved in the direction of distributing fewer data products. They will no longer distribute low-level Landsat data products, which is a tragedy for the research community. USGS has been pretty good at downplaying this fact, while strongly highlighting that a single Landsat product will be made available for free.
Links:
Announcement of ‘free’ data: http://landsat.usgs.gov/products_data_at_no_charge.php
Description of Landsat data products: http://landsathandbook.gsfc.nasa.gov/handbook/handbook_htmls/chapter11/chapter11.html
November 13th, 2008 at 10:29 am
Thanks for the elaboration Jubal. I think going bottoms and top down is definitely the best way to both interconnect and validate data. Not to mention creating the feedback loops between both users and policy makers. Has the potential to create a new paradigm for sustainable agriculture and development IMHO.
@Chris - I think there is always an important role for commercial data that users pay for, but the value of that data only increases with more data that is put in the public domain. Creates a network effect so that “for fee” data can be combined with public data in new innovative ways. Commercial providers need to make money, but I think there is a positive externality of giving some data away for free to jump start the network effect. For goverments and NGO’s I believe making the data publicly available free of charge creates a much larger economic multiplier than any revenue they would generate selling it. Just one very biased opinion on the topic.
November 17th, 2008 at 10:49 am
[…] “Statistics from Space”: Fun with the Gates Foundation and Data as a Public Good - Off the Map […]
November 25th, 2008 at 8:59 am
[…] vs sdi. I gave a brief overview of the system and philosophy, with a brief example derived from Sean’s demos for the Gates Foundation. Good to see the acknowledgment that GC is bringing the accessibility of web tools with the […]
November 28th, 2008 at 5:36 am
Just wondered if there’s an email address I can use to contact you Sean? I want to speak to you about agricultural research tools but a post on here will look like a cheap sales pitch!
Many thanks
Carey Bigs
December 2nd, 2008 at 11:23 am
Hi Carey -
You can reach me through sean at fortiusone.com.
best,
sean
March 12th, 2009 at 2:33 am
I tried to signup for your RSS feed but it didn’t work. How can I do this?
March 12th, 2009 at 2:36 am
I can’t add your rss feed to my reader, what could be the problem?
March 11th, 2010 at 2:16 pm
This is a significant blog posting, im glad I found it. Ill be back again in the future to check out other posts that you have on your blog.