ACCURACY ASSESSMENT OF HISTORICAL MAPS FROM THE LOWER ROANOKE RIVER WATERSHED,
NORTH CAROLINA
Joseph Roberts
Undergraduate Student, Department of Environment, Earth and Geospatial Sciences of North Carolina Central University
The art of cartography has evolved and progressed ever since the first world maps were created in Babylon thousands of years ago. Looking at early maps has made me realize that you can tell a lot about the cartographer from just looking at one of his works. These maps tell us things such
as his perspective of the world, where he is from, and even his religion. Early maps makers such as Ptolemy, Mercator, and Burton wanted and sought out to do the same thing that every cartographer wants to do in modern times. Make their masterpiece as accurate as possible. Accuracy
will always be the most important aspect of cartography for the simple fact that they are used for location reference.
Unfortunately, most cartographers’ technology was not at today’s qualities. Most cartographic tools were the same tools used for navigation, such as magnetic compasses and wind roses for direction, cross staffs for angular precision, and even simple rope to measure and map out coastlines. This makes sense due to the fact that most early cartography was done at sea while on voyages. I have been motivated to find out just how accurate these tools steered early cartographers. While at my internship at UNC-Chapel Hill, I was provided with a ton of early maps of the Roanoke watershed. As I glanced through these beautiful maps, I thought to myself, “Even though these maps are rich in aesthetic value, they really don’t mean much if they are not accurate.” Thus sprung, the purposes of my research to find out how accurate are these historic maps are.
Study Area
Methodology
Creating “Old Map” Points
Maps of Georeferencing (years)
Jpeg to TIFF
Before georeferencing the 9 maps I chose, I
had to change them all from JPEG files to TIFF
files in Microsoft paint.
1798
1808
1814
1818
1833
What is RMS Error?
A measure of the difference between locations
that are known and locations that have been interpolated or digitized. RMS error is derived by
squaring the differences between known and
unknown points, adding those together, dividing
that by the number of test points, and then taking the square root of that result. (ESRI) This
measure is used for observing and analyzing accuracy.
Hypothesis

-The

-Most of the points will be under 2000 meters
off
-Most of the “Old City” points will be under
the under the mean distance.

most accurate map will be 1833
Georeferencing
After transforming the maps, I began to
georeference the maps. The georeferencing process is key because this will be basis of your calculations and analyzing. Most of my control
points are major curves of the Roanoke River. I
also used smaller streams and rivers and even
some of the earlier cities. After georeferencing
each map, I recorded the RMS error, exported a
JPEG of the control points, how many control
points were used to georeference the map. The
JPEG was created to uphold the integrity of my
georeferencing. I had to make a decision to either use the spline or 1st ordinal projection. I decided to use the first ordinal because it maintained the structure of the map and did not distort the map much. But with the 1st ordinal projection comes an RMS error. I tried to keep the
RMS error between 2000 and 3000.
Next, I normalized the Distance from actual column in
my spread sheet by dividing each city by its mean.
Last, I created a bull’s eye representation of my data by
using the multiple ring buffers in ArcMap. My ring distances were the greatest distance from actual value
and the mean of my distance from actual values
Results
As expected, many of the maps were very inaccurate. To
calculate the most accurate map, I took the distance from actual values and added them together for each city. The map
year with the lowest value was 1822. 67% of all of the “old
city” points were over 2000 meters off there mark of the city.
Unfortunately, only 41% of the points were under the mean
distance. All of my hypotheses were wrong.
1822
1825
Data
I received most of my maps from my supervisor
from UNC-Chapel Hill. As I looked through the lower
Roanoke watershed maps, I recognized that there
were 3 cities that were on most of the maps. These
cities were Windsor, NC, Williamston, NC, and
Jameston, NC which is now Jamesville, NC. Fortunately, the maps I acquired were already dated by
year. All of these cities were portrayed the clearest in
maps from 1798 to 1862. Since I now have my 3
study cities for each map. For the cities actuals locations, I used the United States Atlas cities SDC Feature database from 2004. Next, I obtained the most
important part of my data. The Roanoke River. I
used the United States Rivers and Streams SDC feature database which was also from 2004. Last, to
mark where the cities were on the old maps, I had to
create a new data layer in which I named “Old Cities”. This layer consists of the points in which where
the cities were in a particular year after being georeferenced by the Roanoke.
Normalizing the Data
After all my maps were georeferenced
and recorded in my spread sheet, the next
step was to create the points that would hold
the location for each map. For this task, I
first had to create a new feature layer in
ArcMap. As stated earlier, I named the layer
“Old Cites”. After this layer was created, I
began to “edit”. I went through each map,
plotting and recording the map year and city
name for each study city. I plotted the
points as close as I could to where it was on
the map but some of the previous dots that
were created would overlay with the ones I
was trying to plot. After the old cities layer
was created and the attribute table was completed, I had to add a layer which held the
location coordinates for the old map city locations and the actual city location. These
coordinates were used to calculate the distance from the actual city location using the
distance formula.
Map
1857
1862
RMS
CP
XcordJam YcordJam
DFA
Norm
(DFM)
1798 2165.824
19
799173.3 233999.9
3361.941 1.015172
1808 1995.208
20
799075.9
234488
3859.178 1.165317
1814 2797.012
19
798029.9 232590.4
2522.552 0.76171
1818 2787.353
19
801956.5 233447.9
3592.977 1.084935
1822 2109.883
20
800476.9 232648.7
2131.543 0.643641
1825 2261.39
20
798029.9 232590.4
2522.552 0.76171
1833 2268.427
17
799173.3 233999.9
3361.941 1.015172
1857 2632.997
21
800476.9 232648.7
2131.543 0.643641
1862 3173.921
20
793923.5 233304.9
6321.046 1.908703
Xcord
ActualJam
AverageJam
AverageDFA
Ycord
799671.5 230675.1
798924 233302.1
3311.697
Standard Devia on 1297.648
Future Research
I plan to try to make my maps more consistent by
choosing maps from one specific cartographer at a particular time period. Once I find a somewhat flawless
way to replicate this process, I will repeat this process
to find the most accurate cartographer of the Renaissance Era.
References
Kain, Roger. "History of Cartography." Volume 5: Cartography
in the Nineteenth Century (Forthcoming). 1. 5. Chicago: Association of American Publishers., 1987. Print. <http://
www.geography.wisc.edu/histcart/series.html
Abers, James. "Brief History of Maps and Cartography."http://
academic.emporia.edu. Emporia, 2008. Web. 7 May 2012.