U.S. Geological Survey
U.S. GEOLOGICAL SURVEY BULLETIN 2103
Selected Papers in the Applied Computer Sciences 1994
CHAPTER B
Digitizing Options for Hydrologic Maps:
Scanning or Hand Digitizing?
By Andrew Cohen
CONTENTS
TABLES
ABSTRACT
Geographic information systems require that spatial data be digitized into specific formats to allow comparison of data from different sources. Two methods of digitizing map features were compared in terms of time requirements for data entry, ease of data entry, and accuracy of results. Scanning the data set electronically was generally more efficient for a series of hydrologic and planimetric maps than hand digitizing the same material, and scanning planimetric maps required more time than scanning hydrologic maps of the same quadrangles.
INTRODUCTION
Two methods of digitizing maps to produce ARC/INFO\xaa coverages for geographic information systems (GIS) were compared for time, ease of entry, and accuracy. The first method used a Contex\xaa 8000 scanner coupled with PixelTrak\xaa vectorizing software; the second used an Altek\xaa model AC40 hand-digitizing board. Map separates on scale-stable mylar film showing hydrologic and planimetric features were obtained for each method.
Basic steps in the scanning method included (1) scanning the mylar separate,
(2) converting the resulting raster image file to a vector file, (3) converting the vector
file from scanner units to an ARC/INFO coverage in real-world coordinates, and (4) checking
the accuracy by overlaying plots of digitized work on the original separates. Time results
are summarized in
table 1.
Basic steps in the hand-digitizing method included (1) hand digitizing the map features
directly into an ARC/INFO coverage by using the Altek digitizer and (2) checking the accuracy by
overlaying plots of digitized work on the original separates. Time results are given in
table 2.
The scanning method was then applied to map separates depicting two different data sets---planimetric and hydrologic---to determine which of the two sets would yield a faster production time. Both were scanned and vectorized, and the total time needed to produce a final ARC/INFO coverage was recorded. This comparison was not done for the hand-digitizing method because choosing lines to be digitized from either data set requires the same amount of time. All aspects of the test were performed and recorded by the same person.
APPROACH
The data feature to be digitized was hydrology, including shorelines. Two 7.5-min quadrangles in the Champlain Valley of northeastern New York State (Churubusco, N.Y., and West Chazy, N.Y.) were selected because they contain an average amount of detail. Intermittent streams, indicated by dashed lines, were digitized as solid lines.
Two sets of materials were tested. The first set consisted of planimetric separates of standard U.S. Geological Survey (USGS) 7.5-min 1:24,000-scale quadrangles on clear mylar; these were obtained from the New York State Department of Transportation and contained all features found on the published maps except topographic contours. The second set consisted of hydrology separates for the same quadrangles; these were obtained from the USGS National Mapping Division and contained only the blue-line (hydrology) features found on the published maps.
RESULTS AND DISCUSSION
All items discussed are user and (or) site specific and are not meant to describe general conditions.
Linework
If a feature is represented by a solid line, the scanner (PixelTrak software)
is able to complete a large percentage of its vectorizing in a semiautomatic mode;
for example, it automatically traces that particular line until it reaches an
intersection, at which time the operator must take control. If a feature is
represented as a dashed or dotted line, the operator must take control at each line
break and digitize the vertex by hand. In the hand-digitizing method, all vertices
are entered manually.
Accuracy
The linework resulting from the scanning method was excellent. The software divides the thickness of a line into several pixels and selects those representing the middle of the line. If the process yields an erroneous result, the operator can immediately correct the situation. By contrast, erroneous digitizing on the hand-held digitizer is generally not noticed until a check plot is made.
Eye Strain
The hand-held digitizer can cause more eye strain than the display monitor used in scanning. The operator must carefully view the linework through a small eyepiece on the keypad, often under poor lighting. The scanner displays the work on a large color monitor and can enlarge the image several times to minimize eye strain.
Comfort
Continuous use of the hand digitizer is extremely tedious and thus can lead to errors. The operator found digitizing by scanner more comfortable than digitizing by hand.
Lost Data
Frequently during hand digitizing, the processing would abnormally abort and interrupt the digitizing, and lost data had to be redigitized. If the operator uses too little pressure on the entry key, data are inadvertently omitted. This problem becomes major when thousands of data entries must be made. Using the scanner eliminates this problem.
Correction
If lines must be corrected during the quality-assurance/quality-control process, the linework generated by the hand-digitizing method must be split into separate lines to remove an erroneous section. Also, where nodes are too far apart to "snap" together, gaps can result. The scanner eliminates this problem also.
Time Required
Scanning the hydrology separate to produce a final ARC/INFO coverage took significantly less time than scanning the planimetric separate, because vectorizing the scanned features of a planimetric separate requires that the operator identify the hydrologic features by referring to a paper map---a time-consuming step not needed for hydrology separates.
SUMMARY
Results (tables 1 and 2) indicate that, for a digitizing job containing an average amount of detail, the scanning method yielded more accurate results in a shorter time than the hand-digitizing method.
Table 1. Time required for conversion of hard-copy images to digital files by Contex 8000 scanner and Pixeltrak vectoring software.
[All values are in minutes]
-------------------------------------------------------------------------------------------------------
Scan1 Vectorize2 Transfer3 A/I Conv4 QA/QC5 Correction6 Total
-------------------------------------------------------------------------------------------------------
Planimetric separate:
Churubusco, N.Y. 5 134 7 17 19 25 298
7 17 20 8 (4 h 58 min)
7 17 15
West Chazy, N.Y. 5 170 7 17 22 40 302
7 17 17 (5 h 2 min)
Hydrology separate:
Churubusco, N.Y. 5 152 7 17 10 10 253
7 17 20 8 (4 h 13 min)
West Chazy, N.Y. 5 160 7 17 10 10 250
7 17 17 (4 h 10 min)
-------------------------------------------------------------------------------------------------------
1. Scanning consists of loading material into the scanner, performing the prescan functions (such as adjusting light thresholds and dots per inch), actual final scanning, and moving the resulting .RLC (run-length compressed format) raster output file to the PixelTrak partition of the personal computer.
2. Vectorizing consists of loading the scanned .RLC file into PixelTrak, selecting and vectorizing each desired line, and saving the resulting collection of vectors in a .DXF (Drawing Exchange File Format) file.
3. Transfer refers to physically putting the .DXF file on a diskette, hand carrying it to the personal computer that is linked to the Data General (DG) network, and loading the .DXF file into the ARC/INFO workspace on the DG server.
4. ARC/INFO conversion is the process of converting the PixelTrak file, which is in scanner units and referenced only to an origin on the scanner, to an ARC/INFO coverage in some useful map projection (usually Universal Transverse Mercator units) and referenced to the New York State grid so it can be used in conjunction with other coverages. This process is largely automated but still requires some user interaction.
5. During the quality-assurance/quality-control process, a paper plot is made of the A/I coverage and overlain on the original material to check for line accuracy, additions, and deletions. This procedure is standard industry practice.
6. If any discrepancies are found, the original .DXF file in PixelTrak is corrected. The process then repeats from footnote 3.
Table 2. Time required for conversion of hard-copy images to digital files by Altek AC40 digitizer.
----------------------------------------------------------------------------------------------
Digitize1 QA/QC2 Correction3 Total
----------------------------------------------------------------------------------------------
Planimetric separate:
Churubusco, N.Y. 236 24 64 385
20 24 (6 h 25 min)
17
West Chazy, N.Y. 210 30 75 417
25 60 (6 h 57 min)
17
----------------------------------------------------------------------------------------------
1. Digitizing consists of mounting original material on the Altek digitizing board, logging onto the Data General server, creating a new empty coverage (already in a useful map projection, usually Universal Transverse Mercator), and entering each vertex of the desired line into such a file by hand.
2. Quality assurance/quality control is the quality-checking process. A paper plot is made of the ARC/INFO coverage and overlain on the original material to check for line accuracy, additions, and deletions. This procedure is standard industry practice.
3. If any discrepancies are found, the corrections are made by returning to the Arcedit session described in footnote 1. The process then repeats from this step.
Return
to Chapter A
Continue
to Chapter C
Return
to Bulletin 2103 Contents
Return to USGS Home Page
U.S. Geological Survey, ISD National Center, Reston, VA 22092, USA
URL https://pubs.usgs.gov/bulletin/b2016/chapa/ch_a.html
Contact: webmaster@pubs.usgs.gov
Last Modified: 9/8/95 (hem)