Mapping the Ancient Coastlines of Europe
As part of my doctoral work, I became interested in how the land area of Europe changed at end of the last ice age, and how these changes would have affected people living in Europe. Unfortunately, there were no good maps depicting the European continent at that time, so I used the ArcInfo GIS environment to create accurate maps for the European coastline at end of the ice age. You will find below a map of Europe at 10,000 BP (Years Before Present) and a technical description of how that map was made. Additional maps of Europe in past were produced as part of this project at: 18,000; 16,000; 14,000; 12,000; 8,000; and 6,000 BP.
I originally presented the description following the map at the 1995 SAA Conference in Minneapolis, Minnesota during the "Fleshing Out Formal Analyses: New Approaches to Archaeological Problems" symposium. This work has additionally been included in an article presently in review and my dissertation.
How the Map of Europe at 10,000 B.P. Was Created
By Matthew D. Syrett, MA
Department of Anthropology
University of California, Santa Barbara
The creation of the map of Europe at 10,000 BP involved the combination of data coming from variety of hardcopy and electronic media sources. The data on environmental data was not generated by me, rather I transformed them from an existent hardcopy source. I did generate the model of past coast lines from scratch. To my knowledge, this coastline map is the most accurate depiction of the past coastline of the Europe continent at 10,000 BP, due to the coastlines being generated using a model that incorporated data on both eustatic (sea level) changes and holocene crustal distortion.
The area of the European continent is defined primarily by how high the sea level in relation to the region's topography. Although good worldwide sea level curves now exist for the time periods in question, the process of accurately simulating past coast lines is still difficult, since there is much local variation in sea level due to geological processes that can alter the plastic shape of earth's crust. During the Holocene the most powerful of these processes was glacioisostatic rebound, which is the uplifting of land once depressed under the weight of glacial ice. This process is so profound that there are areas of Scandinavia and Scotland are still rebounding from the ice loads they bore in the Pleistocene. Every coastal model for the continent I have seen ignores isostatic rebound, which has led to an inaccurate depiction of the coastlines of Holocene Northern Europe. I desired for my work to create a model that accounted for both sea level changes and isostatic rebound.
To incorporate isostatic rebound into my model, I started with a raster model of modern topography in Europe taken from the Etopo5 world elevation database. This model includes both above and below sea level topography. I then created a series of raster filters for each time period to be simulated. Each of these filters consists of a raster that described how many meters each cell of the modern topography raster would be have deformed up or down to create an accurate description of elevations relative to sea level at the specified period in the past. To visualize this process, it is best to imagine Europe as a flexible surface partly submerged in a bath of water. Instead of simulating past sea levels by altering the water level in the bath, I chose to simulate the changes in sea level by deforming my model of Europe in relation to a fixed water level. This enabled me to simulate sea level changes that are non-uniform across the surface. The filter was created by taking a series of local sea level curves measured from points across Europe and generating a raster by a process of extrapolation called krieging. Some arbitrary points based on a global sea level curve were added to the krieged model to help the filter better reflect reality in areas where data was unavailable and isostatic rebound would not have played an important role defining local sea levels in the past.
Once I had the filter, I subtracted the cells from the filter raster from the topography raster to create a topographic model for 10,000 BP. This new raster of past topography was then transformed into line coverages by generating a contour line at zero meters above sea level. The arc (line) coverages were then manually built into a polygon coverages from which land area could be measured.
The rest of the model was easy to generate by digitizing models of the extent of Northern ice sheets and environmental zones (biomes) from hard copy maps. These digitized models and the model for sea level were then projected into the same coordinate system and then merged together. I projected final map image into a Lambert Equal-Area projection for measuring the area extent of European land and environmental biomes in the past to avoid the distortion of a flat map depicting a spherical globe.
|Projection||Lambert Azimuthal Equal-Area Projection|
|Projection Center||Lat 90 N, Long 0 E|
|Map Extent||NW Corner - Lat 75 N, Long 15 W; SE Corner - Lat 35 N, Long 40 E|
|GIS Environment||ArcInfo version 7.0 running on Sun Sparc Workstation|
|Graphic Environment||Adobe Illustrator 5.5 and Adobe Photoshop 2.5 running on a Power Mac 7100/66AV|
A Lambert Equal-Area projection was chosen for the map coverage due this projection's ability not to distort area measurement for regions as large as a hemisphere. An Alber Conic Equal-Area projection could have been used, but this projection has problems conserving area measurement in regions with a north-south extent greater than 35 degrees. The cartographic cost of using the Lambert Equal-Area projection is a distortion in distance and orientation measurements in all directions except along radii extending from the center of the projection. This was not a problem for my model, since I was ultimately concerned with area measurement.
The ArcInfo coverage for 10,000 BP was created using the Arc identity function to merge graphical data describing whether areas of Europe were above sea level, whether they were covered by glacial ice, what biomes existed for a particular area of land, and what the polygons' general location were using a 5 degrees by 5 degrees latitude-longitude grid. The basic data items stored in the Info relational database for the coverage's polygon attribute table were as follows:
The Etopo5 database is a gridded topographic and bathymetric model of the world. There are 9,331,200 points measured every 5 arc seconds in this database, although in some regions of the globe this accuracy is false since fine resolution data does not exist. The data for Europe and its off shore regions were complied from records of the US Navel Oceanographic Office, US Navel Oceanographic Office, and US Navy Fleet Numerical Oceanographic Center. The Gridded elevation data contained in this database is measured to the center of each raster cell in whole meters above or below sea level. I obtained this model from the Map and Imagery Library at UC Santa Barbara from the Digital Relief of the Surface of the Earth : Bathymetry / Topography Data CD-ROM (National Geophysical Data Center 1988). This model is also available on the Internet through World Wide Web, Gopher, and FTP servers
To gain a model of the modern topography and bathymetry of Europe, a rectangle from 345 to 40 degrees east longitude and 35 to 75 degrees north latitude was cut out of the Etopo5 database. This grid of Europe was then filtered using data on past shorelines and sea levels (see below), transformed into a vector coverage showing past shorelines using an Arc contour function, and then projected into a Lambert Azimuthal Equal-Area projection. The resulting coverages were manually built into polygon coverages and merged with the other data themes. Areas of land were encoded in the LAND variable of the polygon attribute table as 1, while areas covered by water were encoded as 0.
Point data was taken from this volume on sea-level curves worldwide. This data was used to generate filters for the modern topography model to simulate past local sea levels during the Holocene. The locations of points were entered manually by latitude and longitude. The cell values represent how many meters the modern shoreline topography would have to be distorted up or down to model a past shoreline. These points were built into a point coverage within ArcInfo and then krieged (a process of extrapolating raster data from point measurements) into a raster. This raster was then used as a subtraction filter for the modern elevation model. After filtering, the new grid was transformed into a line coverage using a contour function, which was instructed to draw a single contour interval at 0 meters above sea level.
The Fairbanks world sea level curve provided data points for the coverage in areas where other data was non-existent and isostatic rebound would not have been significant.
Isochromes depicting the extent of glacial ice sheets in Northern Europe were digitized from this hardcopy map and merged with the other vector data models. The original coverage was geo-referenced through the use of control points positioned throughout Europe. Areas covered by glacial ice were encoded with a value of 1 in the ICE item of the polygon attribute table, while areas without ice were encoded with a value of 0.
Huntley and Birks' map of the past biomes in Europe based on a pollen reconstruction of 10,000 BP was enlarged, digitized, and merged with the other data models. This environmental coverage was geo-referenced through the use of control points positioned throughout Europe. The areas of the map were encoded into the BIO item of the polygon attribute table as: 0 = no data, 1 = tundra, 2 = boreal forest, 3 = tundra, 4 = grassland, 5 = chaparral/Mediterranean, and 6 = mixed forest.
|Designed by Matthew Syrett|