1. Final Report
Objective
Data Structure
Digitizing Procedure
Error Checking and Correction Procedures
Analysis of the Level of Error
Future Mapping Needs
3. Buildings Feature Class Metadata
4. Street Center Line Feature Class Metadata
Objective:To receive a contract from the Albemarle County (Virginia) Historical Society (ACHS) to create GIS data layers from early 20th-century Sanborn maps of the City of Charlottesville. The ACHS envisions a number of different GIS applications for the Sanborn information:
Land
use change studies:
using layers depicting the city at different snapshots in time, it is possible
to analyze changes in the relative proportion and distribution of land use
categories through time.
Genealogy:
digitizing the address information contained in the maps will allow the
ACHS to pinpoint the
locations of persons/businesses listed in historical city directories and
to aid individuals looking for long-lost relatives.
City
visualization:
creating animations and 3-D views of the city would be an effective means
of visualizing the landscape at different points in time and communicating
development patterns.
Data Structure:
In order to fulfill the requirement of the ACHS, a geodatabase entitled “Charlottesville” was created to hold two feature classes. The feature classes were created and digitized from Sanborn maps 14A and 14B and were entitled “BUILDINGS” and “STREETCRTLINE”. The following tables give details of the attributes of the two classes as well as their format and purpose.
StreetCtrLine Feature Class: line features which will aid in the genealogy application described above
| Attribute Field | Attribute Type | Notes |
| OBJECTID | Object ID | Unique whole number automatically assigned |
| SHAPE | Geometry | Line feature |
| CITY | Text | |
| STATE | Text | |
| ST_DIR | Text | Direction of street (NE, SE, etc.) |
| ST_NAME | Text | Street Name (Hinton, Belmont, etc.) |
| ST_TYPE | Text | Street Type (Ave=Avenue, St=Street, Al=Alley) |
| ADDULEFT | Long Integer | See Note below |
| ADDLLEFT | Long Integer | See Note below |
| ADDURIGHT | Long Integer | See Note below |
| ADDLRIGHT | Long Integer | See Note below |
| SHAPE_Length | Double | Length of the street using internal units |
*Note: Address ranges were constructed using potential address numbers rather than actual to minimize the number of updates required in the event of future development. Values were chosen at logical breakpoints. For example, potential ranges of 0-98, 1-99, 100-198, 101-199 were assigned. The left and right sides of the street are determined by hypothetically standing in the street and looking in the direction of increasing address number. For example, 6th Ave SE begins with 200 for the Lower Left Address and ends with 398 for the Upper Left Address. The Lower Right Address is 201 and the Upper Right Address is 399.
Buildings Feature Class: polygon features which will aid in land use change studies and city visualization described above
| Attribute Field | Attribute Type | Notes |
| OBJECTID | Object ID | Unique whole number automatically assigned |
| SHAPE | Geometry | Polygon feature |
| TYPE | Text | Com (Commercial), Res (Residential), Pub (Public) |
| NUM_STORIES | Short Integer | Number of stories a building has |
| X_COORD | Long Integer | Location of x-coordinate of the building's centroid |
| Y_COORD | Long Integer | Location of y-coordinate of the building's centroid |
| SHAPE_Length | Double | Length of the building using internal units |
| SHAPE_Area | Double | Area of the building using internal units |
Both feature classes utilize the Universal Transverse Mercator Projection, Zone 17. They are based on the North American Datum (NAD) of 1983. Shapefiles were exported based on these feature classes in order to allow users of ArcView 3.2 to utilize the digitized information. The shapefiles do not utilize a projection, but rather are based on the geographic coordinate system, NAD 1983.
Digitizing Procedure:
The digitizing procedure was the most pain-staking and time consuming part of the project. The street center lines were digitized first and were completed fairly quickly. The approximate centers of the streets were estimated visually; no calculations where made. The attributes for each street (type, name, direction, addresses, etc.) were entered after each individual street was digitized.
The individual buildings were digitized last and required the most time. Each individual section of a building had to be digitized separately due to the differences in height. This was important in order that the ACHS can make a 3-D representation of the City of Charlottesville in the future. The number of stories (i.e. height) for each part of the building must be known. Again the attributes for the buildings were entered.
Error-checking and Correction Procedures:
Error-checking and corrections occurred in a variety of ways. The digitizer double-checked that all streets in the feature class were in fact located on the original Sanborn maps. This was done by highlighting the street in the attribute table and making sure that it was found in the feature class. During this process, it was noticed that some features in the feature class were not digitized properly and had to be removed or extended. The feature class was also checked for overshoots and undershoots and adjustments were made where needed. In addition, all streets on the digitized maps were checked to make sure that they had been labeled correctly in relation to the streets on the Sanborn maps. The names of streets, their direction and the address ranges were re-checked to make certain that they were spelled and written correctly within the attribute table.
A similar procedure took place within the buildings feature class. The digitizer double-checked that all digitized buildings were also located on the scanned Sanborn maps. Some buildings were noted to be absent on the digitized map and had to be added. Polygons were checked over to make sure they were enclosed by four sides and shared polygons were checked to see if they each had their own four sides. Separate sections of the same complex or building were checked to make sure that they all had the same address.
Analysis of the Level of Error:
Since perfection is known to be impossible, it is a fact that error has entered this project in many ways. A number of decisions and assumptions were made during the digitizing process that may have induced error. Some of these are outlined below:
- groups of buildings all received the same address pending on proximity to each other
- groups of buildings belonging to a single business (e.g. Charlottes Lumber Co.) received the same address
- when two separate addresses referenced one group of buildings, the address chosen depended on where the dashed line of a building was found (e.g. if dashed line of building was on South St SE, then address was 207 South St SE)
- the letters GRO. on a building were assumed to mean "grocery" and therefore the building was labeled as a commercial building type
- the letters VAC. on a building were assumed to mean "vacant" and therefore the building was labeled as a residential building type
- buildings without distinctly labeled addresses, such as those individual buildings in unlabeled plots, received null values in the attribute table
- streets without buildings, and therefore no addresses, also received null values in the attribute table
- numbers of stories were recorded for buildings so therefore, when the ACHS creates a 3-D map of the city, they will have to make an educated guess as to the height of each story
- some of the roads found on the Sanborn maps no longer existed on the more current orthophoto, which made georeferencing less accurate
- the Sanborn maps were scanned and therefore creases and bends (such as those found between map 14A and 14B) caused the maps not to align properly and therefore some of the buildings were digitized only according to map 14A and didn't match up with 14B and vice versa
- control points used in georeferencing were difficult to match up because they were sometimes close to the edge of the map, therefore centers of intersections of streets were not utilized. Rather, if a control point was located at 5th St SE and Levy Ave SE, the NE street corner (the edge of a property boundary) was used instead
In regards to georeferencing the scanned Sanborn maps to the orthophoto, it is not possible to quantitatively determine the error in georeferencing (in terms of meters, for example). This is due to the fact that the projection and/or coordinate system of the Sanborn maps is unknown. In order for the Root Mean Squared (RMS) error value (found to be ~3 for each map) to have significance, one must know the coordinate systems for the maps/images involved in the georeferencing process. Although the projection is known for the orthophoto used in the georeferencing (Universal Transverse Mercator (UTM), Zone 17), the RMS error has no significance because the projection is unknown for the Sanborn maps.
Despite the assumptions and decisions mentioned above, it can be said that the digitized maps will appropriately serve the purposes intended for by the ACHS. For the purpose of 3-D visualization, the data is clearly accurate and precise to the degree necessary. Again, a height must be determined for each building story and an educated decision must be made. For the purposes of genealogy, the locations of specific houses will be possible by looking at the two feature class attribute tables and locating a particular building on a particular street. The land use application of this database is also accurate enough even when taking into account the assumptions mentioned above. Localizing land use changes between decades will definitely be possible.
Future Mapping Needs:
For future needs of the ACHS, the approximate amount of time needed to produce the same result for a different temporal series is as follows:
| Task | Time (hrs) |
| Georeferencing | 0.5 |
| Digitizing Streets | 1 |
| Digitizing Buildings | 2.5 |
| Error Checking | 0.5 |
| Edge matching | 0.5 |
| TOTAL TIME | 5 |
Therefore, the estimated total hours required to digitize 30 maps covering the area of the City of Charlottesville will equal approximately 150 hours. This would require a full-time employee (working 40 hours per week) to devote approximately 3.75 weeks of time in order to complete the project for a different decade.
The above thematic map shows the land use pattern in the south eastern portion of the City of Charlottesville, Virginia. This map demonstrates that the Charlottesville database will satisfy the future mapping needs of the Albemarle County Historical Society (ACHS).
back to gis coursework