Why Cyclone Center is the CrockPot of citizen science projects
Much like North Carolina-style barbecue , our project is slow-roasted and prepared for greatness.
The task of analyzing historical satellite images of tropical cyclones from around the globe is no small task. It is an important project, but in our four years, some have said our progress might be slow. Here’s a peek behind the curtain on what we’re doing (the science team) and our plans for the future.
Where are your results?
There are quite a few reasons why our progress may appear slow. This is due in large to the type of project we are working on.
- We’re not discovering new systems – Unlike GalaxyZoo where many of the images have never been seen by human eyes, each image has been analyzed before. However, much of that analysis was done during the storm, where the focus was on forecasting and not necessarily analysis. So we don’t foresee discovering new types of hurricanes. What we do foresee is the largest undertaking of weather satellite analysis ever!
- Citizen scientists aren’t as familiar with the images – Unlike images of African animals, many of the Cyclone Center images look foreign to some. In fact, they likely appear more like Rorshach tests than nascent images of a developing severe typhoon. What does that mean? We are really depending on those citizen scientists who will persevere through the difficult images to find those diamond-in-the-rough eye scenes.
- Progress is only made at a large scale – The goal here is climate-worthy data, which means two things: good data and lots of it. Even limiting our current emphasis on the Western North Pacific still requires analysis of more than 30,000 images. So we can’t reanalyze one storm and publish (though it could be possible). The positive is that there have been published studies that suggest that the typhoon activity is increasing while others have shown a decrease in activity. Cyclone Center is poised to help in this discussion thanks to work of our citizen scientists.
So what are the results so far?
We have plenty. Our peer-reviewed scientific papers show:
- The need for our project – while much progress has been made in recent years toward forecasting tropical cyclones, less emphasis has been on the reanalysis – that is, to understand how they may be changing. This project is the first (and presently only) dedicated to putting human eyes on each and every satellite image to analyze the patterns.
- The initial results of estimating intensity are actually pretty good – for our first paper, we had limited data so we tried to develop a simple approach at estimating intensity from the citizen science data, kind of as a proof of concept. The result was a relatively good approximation of storm intensity. Again, the emphasis of our project is not on the most accurate storm analysis, but the most consistent analysis. That way, we can compare storm intensities in the 1980s to those in the 2010s.
- The results are consistent and repeatable – Repeatability is so important in science. “If we do this project again, would we get the same result?” Our first analysis was to look at consistency between users. What if a different set of citizen scientists looked at the same image? Would they get the same result? We found the result a resounding yes!
- The storm types selected by users tell us about storm evolution – The images analyzed can be simplified into 4 primary storm types or patterns (curved band, embedded centers, eyes and shear). These types are related to the lifetime of the system. We’re learning how these storm types are related to growth and decay of a system.
We’re currently working on a refined algorithm to determine the storm intensity from the selections from citizen scientists. We’re also working to expand the database of storm images – the last ones on CycloneCenter are from 2009 … we’re planning to update with 5 more years soon.
In short, the progress may appear slow, but so does slow cooking meat in my kitchen over night. However, just like the fragrant aroma that I wake up to in the morning tells me that we ready for a gourmet meal, the progress we see from our citizen scientists show that some of our initial goals are nearly complete . The analysis of the Western North Pacific typhoons is so important and so close to complete. Keep classifying! We’ll keep analyzing your results. We’ll reach that goal soon. And we won’t be done there. There are other ocean basins to complete. There are other seasons of data to be included!
Hennon, C. C., and Coauthors, 2015: Cyclone Center: Can Citizen Scientists Improve Tropical Cyclone Intensity Records? Bulletin of the American Meteorological Society, 96, 591-607.
Knapp, K. R., J. L. Matthews, J. P. Kossin, and C. C. Hennon, 2016: Identification of Tropical Cyclone Storm Types Using Crowdsourcing. Monthly Weather Review, 144, 3783-3798.