By Lizzie Wallace
The Arizona Shuttle rolls up to Flagstaff, Arizona, quite a different scene from the miles and miles of barren desert I’ve been riding through on my travel from the Phoenix Skyharbor International Airport to Northern Arizona University (NAU). At 7000 ft, Flagstaff is a cool 70 degrees surrounded by trees, mountains and large statues of lumberjacks, the mascot of NAU. I clamber out of the shuttle into an empty parking lot and venture through a deserted campus to find my hotel. Classes haven’t started yet, so the 30,000 undergraduate students that usually haunt this spacious campus have not yet arrived. I am not here for classes, though, I’m here for a workshop on using GeoChronR, an open source tool for analyzing and visualizing time uncertain data, in particular paleoclimate data.
Paleoclimatology is the study of ancient climate on Earth. Since we are unable to travel in time to investigate what climate was like millions of years ago, scientists build records of past climate called proxies to infer Earth’s climate. These proxies include ice cores, tree rings, coral cores, and my personal favorite, sediment cores. Tree ring thickness for example can tell you something about a region’s temperature and precipitation, as optimal conditions will result in thicker tree ring growth.
However, there are a lot of uncertainties associated with paleoclimate which is what has drawn a variety of young scientists from around the world and myself to this workshop. In particular, age uncertainties in dating these proxy records can introduce tens to hundreds of years of uncertainty in the timing of climate events. This makes it difficult to combine regional climate records into a global dataset or to compare your climate proxies with other similar ones. And more importantly, there is no accepted framework within the paleoclimate community to deal with these age uncertainties. So this is what the organizers of GeoChronR, Nick McKay and Julien Emile-Geay, professors from NAU and University of Southern California, hope to do with their new toolbox.
GeoChronR allows users to load data in a uniform format, generate many different types of age models for records, and create time-uncertain ensembles of data. These ensembles refer to the 1000 possible age models that could fit my data generated by age modeling software. In practice, most paleoclimate scientist only use the mean age model. This neglects a lot of the uncertainty associated with your data. GeoChronR allows me to work with these ensembles easily using a number of commonly-used techniques and to visualize results in an intuitive way.
Nick and Julien are really excited to get scientists working with their new tool. Indeed, most of the workshop involved actively using the code with our own data. Two and a half days of sitting at a computer playing around with all the cool things that the GeoChronR code could produce. The code for GeoChronR is available in R, an open source and community supported platform. Being an R novice myself (preferring the glories of MATLAB), picking up this package was pretty difficult for me. Most of the workshop involved struggling to load my data and process my data in this package. Indeed, working out kinks in the code was another goal of the workshop for Julien and Nick, who spent most of both days running around trying to fix bugs in the code and help everyone get GeoChronR up and running on their computer. While GeoChronR is still a work in progress, the hope is that someday all paleoclimatologists will use this package to work and share their data with the community.
While I am still working to fix some of problems with using GeoChronR on my data, I am excited to use this toolbox in the future. I think it could be really useful in making uncertainty in paleoclimate studies more apparent and making proxy data easier to use and share in the scientific community.