Mapping the Past to Predict the Future

Posted on Oct 11, 2013 | No Comments

Mapping the Past to Predict the Future

It is well know that maps have always been about much more than just geography. Maps also serve as a medium for political and social commentary, becoming a reflection of the mapmakers and map financier’s ideas of how the world is or, perhaps, should be.

Map with Europe in the center

Maps have often been used to make political and social commentary. By placing Europe near the center of the map, this mapmaker is attempting to convey Europe’s power.

Knowing this, historians understand that 19th century maps that placed the British Isles at the center of the world did so to emphasize the sea power of the British Empire. Likewise, early maps of the Christian world would orient viewers to a spiritual rather than geographic course by putting Jerusalem at the center of all things. It is not surprising, then, that when historians look at maps of the Great Lakes from the late seventeenth to early nineteenth centuries and find discrepancies between how the French and British drew the size and shape of the lakes, the cause of representational differences is assumed to be the political war over ownership of the lakes during that time period. Another common theory for “inaccurate” early maps is simply that our ancestors had more primitive tools of measurement. Lacking satellite and GPS, the changes in coastlines that early mapmakers recorded over time could easily be due to measurement error.

I-CHASS researchers Dr. Michael Simeone and Dr. Robert Markley at the University of Illinois at Urbana-Champaign approached their analysis of the changing maps of the Great Lakes with the assumption that the cartographers of the 1600-1800s were making their maps in good faith, as accurately as possible. “We decided to take historical maps of the Great Lakes seriously to see if they could tell us something about the surface areas of the lakes,” Simeone said. One major barrier to comparing the maps scientifically was that the varying scales and sizes of maps made it difficult to calculate the surface areas of the lakes by hand. To overcome this problem, Simeone and Markley worked with University of Illinois graduate students Tenzing Shaw and Deepak Neralla to develop new image analysis software. The software works by first distinguishing between land and water on the map by segmenting out the shape of the water. Each map was drawn to a slightly different scale, and so the maps were standardized using spatial calibration, which converted pixel coordinates to geographic coordinates. Finally, the shapes of each lake were used to calculated surface area. In total, 40 historical maps (18 British and 22 French) of Lake Ontario were analyzed for their surface area.

A comparison of the surface area of Lake Ontario on historical maps to today’s satellite-confirmed figure of 7,540 square miles revealed some interesting results.

The process of determining the recorded surface area of maps in old maps.

The software developed by I-CHASS and ISDA researchers works by standardizing the scale of the different maps using spatial calibration, which converted pixel coordinates to geographic coordinates (Image 1), distinguishing between land and water on the map by segmenting out the shape of the lake (Image 2). Then, the outline of the lake (Image 3) was used to calculate the surface area of the lake (Image 4). Finally, the surface area determined by the historical map was compared to the satellite-confirmed surface of the lake (Image 5). (The map pictured is from the Map Division of the Library of Congress)

When comparing the maps drawn by the French and the British, both countries were about equally accurate or inaccurate, even though the French occupied the Great Lakes region from an earlier time. Map accuracy for lake areas increased slowly over time, and then dramatically around 1760, likely due to the invention of more reliable chronometers. The amount of error in each historical map was found by subtracting the modern confirmed surface area of each lake from the surface area found from each historical map. The image below shows a graph of the surface area of each historical map with British maps in Green and French maps in Purple. The larger the size the circular marker is for each map, the larger is the amount of error that the map contains. Almost identical surface areas and error between maps seems to indicate that some mapmakers simply copied maps that were already published. The surface area graph of Lake Huron stands out, however, because surface area and error are very inconsistent over time.

Graph visualizing the surface area of Lake Ontario as portrayed in historical maps

British maps are green, French are purple. Over years A.D., measured in sq. miles. Radius of balloons = .1*sqrt of absolute error (real size of lake is aprx 7k sq. miles)

Measurements of Lake Huron remained inaccurate for far longer than any of the other Great Lakes.

Maps of Lake Huron remained inaccurate and inconsistent for far longer than any of the other Great Lakes maps.  (The maps pictured here are from the Map Division of the Library of Congress)

It wouldn’t make sense for mapmakers to be continually improving in the map-making skills over time for all of the Great Lakes except for one, and so Simeone and Markley looked into the historical records of the Lake Huron area during the 1600-1800s and found that there were years of unusually cold weather. Ice charts from the time period show that some areas of lakes, especially the archipelago in Lake Huron, developed large amounts of ice cover that likely made it difficult for mapmakers to distinguish between ice and land. The lake likely appeared bigger or smaller depending on the amount of ice. Since it seems that maps were becoming more accurate over time, variations in surface area that are outside the normal realm of error are likely giving us information about the climate of the past.

This climate information is very useful because no climate models exist before 1856, when the British Meteorological Society began collecting temperature records from around the world. Scientist can currently only learn about historical weather patterns by looking at information such as the size of growth rings on trees and the composition of ice taken from Antarctica and Greenland. These records used for these indirect measurements of temperature are called “proxy data.” The preliminary results of Simeone and Markley’s study shows that ice accumulation around lakes are another type of proxy data that indicates changes in climate over time. It is essential to find climate information before 1856 because with this information we can expand the climate timeline and gain a more accurate picture of global warming. The biggest critique of climate models is the length of the timeline, but this research project gives hope that by studying changes in lakes and other features on maps, a deeper understanding of both past and future climate changes is possible. This information would make our predictive capabilities regarding climate change much stronger, and will better inform us as a society about the impacts of human beings on the world.

This research is in the early stages of analysis and the next stage is to analyze more maps of the Great Lakes and maps of other lakes and coastlines around the world. The algorithm that is used to calculate surface area is also being revised to better handle complicated coastlines and maps with little contrast between water and land. To learn more about this project, contact Dr. Michael Simeone at mpsimeon@gmail.com .

« Previous post:
» Next post: