Home

Static Maps
Interactive Map
Methods
Well Log Scans
Data Sources

Methods of Study

Well Log Collection
Database Creation
Georeferencing the Wells

2.0 Methods of Study
The methods of study were based on the requests of the County and based on strategies used by other agencies such as the United States Geological Services (USGS) and the Virginia
Department of Mines, Minerals and Energy (DMME). The primary tool in this evaluation involved the collection and interpretation of existing groundwater well logs. The principle source of this data was the Virginia Department of Health (VDH) and local well drillers. The intent of this process was to develop a representative database of well logs from which interpretations could be made regarding the nature of the resource. From the outset is was determined that a companion web-page would be developed to allow for access to the data by interested parties and to provide an interactive platform for the study. The following sections describe the process taken to achieve these goals.

2.1 Well Log Collection
As stated previously, collection of well logs was the primary data collection vehicle. Information pertaining to a total of 1780 well logs collected. Of these 1669 were positively georeferenced onto a map using a Geographic Information System (GIS) platform. Figure 1 of this plan presents the location and distribution of this data. As described below the data from each of these logs was compiled into a database for analysis.

2.1.1 Well Logs: Stafford County VDH Office
The vast majority of the data came from the Stafford County VDH office in the form of well logs. With the cooperation of the Stafford County of VDH, Draper Aden technicians set up a laptop and scanner in their office for a period of four weeks. Each log was scanned directly into a PDF file and named for its VDH permit number. All of the well log scans are presented on the webpage at daagis.com/Stafford.

The well log records consist of hard copy paper folders organized by tax tile. These folders contained well records pertaining to that tax tile with no further organization or indexing. The technician went through each file with the goal of scanning about three dozen (36) well logs with a random distribution within the tax tile. Figure 2 presents the distribution of wells per tax tile. Each log’s general location was marked on a tax map. In many cases all of the available logs were scanned. Some tiles contained no well logs. In a few tiles significantly more than three dozen logs were scanned. The attempt was made to ensure that the collected log files were distributed across the tax tile as best as possible given the limited availability of logs and the fact that the true location of many of the logs could not be determined during the scanning process. No attempt was made to count the total number of well logs stored at the VDH office. A total of 1390 logs were scanned in this manner.

2.1.2 Well Logs: Culpepper Regional VDH Office / EPA EnviroFacts
Public or community well information is stored at the Culpepper Regional VDH Office. The VDH compiles information on these wells in a manner that does not conform with a typical well driller’s log. The information on total of 11 of these wells were scanned. Ultimately the list of wells supplying public or community wells was augmented from the EPA EnviroFacts website. Additional research was applied to accurately locate these wells on the map. Figure 3 presents the location of these community wells.

2.1.3 Well Logs: Private Driller’s Records of John Danielson
Local well driller John Danielson provided a spreadsheet of his company’s Stafford County drilling since September 1998. Data from this source provided well depth and yield for each of the wells, but did not include several of the other data fields collected from the actual well logs collected at the Stafford County office of VDH. A total of 379 records were imported from the spreadsheet.


2.2 Database Creation
Microsoft Access was chosen for the database platform due to its widespread use and its proven integration with ESRI’s ArcGIS line of products. A master table was created to represent the well logs and a form was built to facilitate data entry. A technician then entered selected attributes from the well logs. The ‘Danielson spreadsheet’ was also imported into the table.

There is a tremendous variation in quantity and quality of information reported on the VDH logs. Some have very little information. Some have a great deal. Different drillers interpreted things in different ways and placed different priorities on what to record and how to record it. Logs ranged from remarkably detailed to marginally complete. All these factors influence the quality of the database. Accordingly a process of “filtering” the data was applied to ensure that that values entered into the data fields were valid and reproducible.


2.3 Georeferencing the Wells
Georeferencing of the wells was a critical component of accurately depicting and analyzing the data. Georeferencing refers to the process of accurately locating the wells with respect to an “X, Y and Z axis”. The X and Y axis refers to a wells horizontal location, or latitude and longitude. The Z axis refers to the elevation of the well.

2.3.1 Goal and Approach
The goal was to place the wells on the map as near to their true horizontal and vertical coordinates as possible using available information. Wells are generally drilled in close proximity to the building they are intended to serve. Accordingly, it was determined that the most accurate location would be to place the wells on the respective buildings wherever possible. In most cases this is a home. On larger lots this approach is significantly more accurate than placing the well in the middle of the lot, i.e. on the lot centroid.

Larger lots are more likely to encompass a greater range of elevation. It was determined that that the elevation of the parcel centroid was often significantly higher or lower than the elevation of the house. By putting the well on the home we better approximate the true elevation of the well and more closely ensure that the vertical or ‘Z’ position has useful meaning.

The County was able to provide GIS layers for tax parcels, buildings, roads and 5-foot elevation contours. Our georeferencing assumptions were:

1- The well is near the house. Since we don’t know which side it is on we placed it on the centroid of the house.
2- The well is at the same elevation as the house.
3- On lots with no house, if small (<4ac.) - use the centroid, if larger - individually evaluate for best placement.
A comment field was added to the data table to record how each well was georeferenced.

2.3.2 Horizontal Placement
The objective was to would match one well log to one tax parcel that had one building. To accomplish this goal the following preparation and geoprocessing was required.

2.3.2.1 Preparing the Parcel Layer
The first task was to reformat the County’s coded multi-field parcel identifier into a single piece of text written as an average person or well driller would write it.

Next, a centroid point was created for each parcel. (Actually a ‘label point’, which is the centroid, unless the centroid falls outside of the parcel in which case another algorithm calculates the point to be within the parcel.) The new ‘parcel point’ layer was limited in application for matching because the tax-ID was not unique. Parcels that crossed a road or stream, for example, were represented as two polygons with the same ID. Therefore, those portions of the parcel point layer that included wells were given a priority in analysis.

2.3.2.2 Preparing the Well Log Tax ID
An initial match of the wells to parcels resulted in about 60% matched. This was primarily a function of mis-identification of the tax ID on the log by the driller. In some cases the tax ID had changed through renumbering or land subdivision since the well was installed. Many were simply written in a manner inconsistent with the parcel naming conventions. These 40% unmatched wells needed to be manually reviewed to determine the correct or most probable location and proper tax ID. This was a critical, although labor intensive, process. A flag was added to the database to record the confidence level of each manual location.

2.3.2.3 Resolving the Well/Parcel Match
The next step was to relate the well log database to the tax parcels and extract duplicates, both allowed and not-allowed. It is possible for a parcel to have more than one well, but it is not possible for a well to be on more than one parcel. The previously created parcel point layer became the primary tool of matching effort. Duplicate parcels were resolved by simply deleting the parcel point that appeared less likely to contain the well. In the few cases where there were more than one well on a parcel, a second parcel point was added with a suffix “_DUP1”. The well record was revised to correspond. In this manner every well log record had a one-to-one correspondence with a parcel point centroid on the map.

2.3.2.4 From Parcels to Buildings
Having the selection set of all parcels with wells, it was logical to select all buildings that are on or cross the border of those parcels. Of course many parcels have more than one building or buildings that are only partially on the lot. To remove the non-home (duplicate) buildings, such as sheds and barns, a zoom-to, swipe, and delete procedure was followed. Though this was a manual process and did involve a degree of professional judgment, it was necessary. The primary filter of automatically selecting the largest building on each lot would have worked in most cases, but would have failed to provide the best solution if a barn was present.
Once the building subset was revised down to one primary building for each lot having buildings, a building centroid was calculated. However, a few buildings that spanned lot lines received a centroid that was positioned over the wrong parcel thus preventing proper tax ID numbering. These were highlighted and manually relocated. Then, a simple automated task was applied to assign the tax ID number to the building point and delete the corresponding parcel centroid. In effect the working parcel point was moved to the building, our original goal.

2.3.2.5 Final Match
Each well, whose location could be determined, now had a single corresponding parcel point on the map. The database was linked to the parcel points creating a selection that was renamed “Wells”. The final graphic well locations on the map linked to our well log database with a one-to-one correspondence by way of the slightly modified tax ID numbers.

2.3.3 Vertical Placement
After the unique (x, y) location or ‘match point’ was determined for each well an elevation was calculated for each well. The County’s 5-foot contour layer proved to be not only the best information available, but also an excellent vehicle for determining well elevations. In an effort to maximize the use of the elevation data, the large file size of the layer was reduced to more manageable files sizes by segmenting the contour layer to individual tax tiles. An ESRI elevation TIN and GRID, with a cell size of 5-feet, was then processed from each of the contour files. This process allowed work on a tile basis, which provided for the refinement and validation of elevations at each point with a high degree of confidence.