Come join the workshop in Chadwick 217, at 14.00 on Friday Feb 17th to see how you might play a part in this.
Introduction: for some time during 2011 Dr Tim Sparks and Peter Gingold have been pursuing an idea of Tim’s, the use of dated photos to provide vivid evidence of climate change. An example from the USA:
Note that the photos must be taken during the spring or autumn (and of deciduous trees or shrubs); summer and winter present a less interesting picture!
Work to Date
The project thus far has focused primarily on photos of iconic locations already in the public domain – from a variety of online photo archives. A good example is Remembrance Day in Whitehall, UK, which takes place on or near November 11.
Other examples include Remembrance Day (November 11) in France, Queen’s Day (a national holiday in The Netherlands on April 30) and the May Day Parade in Red Square – though there are challenges in finding photos of trees rather than rockets.
The methodology in use to date is as follows:
- A 5-point scale has been defined: 5 = full leaf cover, 1 = bare.
- A panel are invited to ‘score’ a given photo from 1 to 5. Marked divergences illustrate the need to find additional photos.
- Average results can be simply plotted on a graph, or correlated with the instrumental record. They provide abundant evidence of climate change – and at iconic locations and events.
No major methodological problems have emerged that seem to be insuperable.
Notwithstanding the fact that there is a gap in the (readily available) photographic record in the seventies and eighties the evidence seems crystal clear – an increase in the length of the period of being in leaf has clearly taken place (vertical bars indicate unresolved differences between separate photos reportedly from the same year).
Tim is currently working on a paper for publication.
It seems clear that this approach has enormous promise. Iconic locations and events have been used to date for ease of access in the public domain, and also because they tell a powerful story. But many other records exist: archives of local and national newspapers, records of parks and arboreta, local sports clubs, local history societies, family albums etc. With the caveat that photos are frequently not marked with the date on which they were taken, often just season or year, there are enormous resources waiting to be explored. And looking forwards, the fact of digital photos often being geo-tagged by default means that this approach to recording climate could scarcely be simpler.
Citizen Science: the approach we have outlined obviously lends itself to a citizen science web site, whose functioning and protocol could be roughly as follows:
1. The aim of the site is to create options for:
- Uploading a series of (minimum 2) dated photographs of a given location
- Allowing registered users to score the leafiness of photographs
- Scores to be assessed and approved by some form of moderation process.
- An average score to be displayed
- Some form of output to be displayed automatically.
2. A key initial feature could and should be a number of image series from iconic locations: Remembrance Day at the Cenotaph, ditto inParis, the Moscow May Day parade would be good starts, but the broader geographic spread the better. For example, there are innumerable St Patrick’s Day parades across the US.
3. At start-up it should ideally also have several series of more photos from more domestic or everyday locations in different parts of the world.
4. Users need the option to be able to upload a new set of photos.
5. In summary, therefore, a registered user could:
- Score an existing photo
- Add photos to an existing series that has gaps, or weak photos
- Start a new series with their own photos
- The site has to be sceptic- and vandal-resistant. Proof in some form of the reliability and accurate dating of new photos would presumably be needed (validation on a sample basis retrospectively?) which might be challenging and/or administratively demanding.
Most citizen science sites come stocked with a defined set of data which users are invited to interpret; the scope for this one is much more open-ended, since there are no limits on the number or geographic spread of photos to be covered, and every reason to invite ever larger numbers to take part; a lot of the benefit would probably come from international comparisons. Given the largely common-sense nature of the evaluation and analysis, it may also be possible to take the citizen science element to a further level, by giving user groups a level of control over or at least input into functionality.
There is a fascinating challenge in exploring what forms the ‘output’ should take. Turning photographs into a number, then summarising those numbers in a graph or some such appeals to the traditionally minded! But is it really in the spirit of the project? How might the information be presented in a more lively way that does justice to the originality and individuality of the data?
Possibilities: presented here are some – by no means all – of the questions that seem to need exploration. More are welcome, as well as answers to them.
- How much more valuable is a time series than, say, a pair of photos?
- How to present the data? For example, might an animation or some other visual representation be more powerful than traditional forms of data presentation?
- How near (in location and date) do photos have to be to count as being at the same site?
- Would it be desirable to extrapolate from ‘scores on a given date’ to ‘season length’? And if so, how?
Citizen Science web site
- How tricky is the validation-of-new-photos problem?
- How big a job is developing the site?
- Are copyright issues significant?
- What is the extent of local photo archives?
- How well are local photo archives documented (i.e. with dates)?
- How enthusiastic would people be to join in on such an exercise?
- How well would this work in areas of less seasonality?
- How would this project be best structured as a research programme?
- What disciplines should be involved? Can it extend to social, historic and cultural changes detected from photographs?
- What would the research question(s) be?