PFRA Watershed Dataset Details

Contents

• Introduction
• Background
• Digital Watershed Products
• Watershed Product Details
• Concepts
• The Manuel to Digital Migration Effort
• Cascading Network Methodology
• Products
• Contributors
• Summary
• References

INTRODUCTION

The Watershed Project is a long-term Prairie Farm Rehabilitation Administration (PFRA) undertaking. It entails creating and maintaining a hydrometric gauging station based watershed database for the Prairies. Hydrometric gauging stations, watershed boundaries, and watershed areas are its core data from which digital data products are generated for use in hydrological mapping and analysis. The spatial extent covers most of Alberta, Saskatchewan and Manitoba, and small portions of British Columbia, the Northwest Territories, Ontario, and the United States into which the Prairie watersheds extend. Since the early 1970s, the process has been evolving from manual manipulations to digital processing. This migration from paper maps and index cards to a Geographic Information System (GIS) environment is a significant milestone in improving analysis, access, and distribution of the Watershed Project data.

BACKGROUND

In 1970, as part of an International Hydrological Decade study, the PFRA Hydrology Division undertook the tasks of delineating and of measuring the areas of gross and effective drainage for catchments tributary to active and discontinued hydrometric gauging stations in the Prairies. The results were intended to provide the basis for updated regional flood and runoff studies. In 1975, at the request of the Prairie Provinces Water Board, PFRA formally agreed to delineate gross and effective drainage basin boundaries and also to provide a set of standardized drainage area maps (1:250,000 scale). This work was to be conducted in consultation with the Water Survey of Environment Canada (EC) and the representative agencies from the provinces of Alberta, Saskatchewan and Manitoba. Prior to this agreement, each government agency had independently delineated watershed boundaries. This practice resulted in area calculations different enough to generate controversial results. Consequently, the interested parties agreed that PFRA should be the sole developer and custodian of a set of commonly-accepted watershed delineations. In 1994, the decision was made to move the watershed delineations and the area calculation process into a digital environment. It was triggered by the time-intensive nature of the manual updating process, the deteriorating condition of the maps, and fiscal constraints. It involved digitizing the drainage delineations and gauging station locations from nearly 3000 maps, edge-matching the digital "pieces" into a single coverage, and creating an automated process for calculating and summing the watershed areas. This spirit of cooperation and commitment has created a unique and useful database that is used not only by government agencies, but also publicly by a spectrum of non-government agencies, consultants and academia.

DIGITAL WATERSHED PRODUCTS

The current version of the PFRA Watershed Project publishes twelve digital watershed products. These are derived from five source products:

Source Products

1. A coverage of hydrometric gauging stations (as points) and their effective and gross watershed boundaries (as arcs);
2. A coverage of arcs that maintains the upstream/downstream relationships for hydrometric gauging stations;
3. A TIC coverage for digitizing purposes;
4. A coverage of Canadian mapsheet outlines; and
5. A coverage of american mapsheet outlines.

Output Products

1. Gross and Effective Drainage Areas for Hydrometric Gauging Stations;
2. Effective Drainage Area;
3. Effective Drainage Areas of the Incremental Effective Drainage Areas;
4. Incremental Gross Drainage Areas;
5. Lines of Gross and Effective Drainage Area Boundaries;
6. Major Basins;
7. Major Drainage Systems;
8. The Hydrometric Gauging Station Network;
9. Areas of Non-Contributing Drainage;
10. Hydrometric Gauging Stations;
11. Environment Canada Sub-basins; and
12. PFRA Sub-basins.

WATERSHED PRODUCT DETAILS

Source Products

All five source products are ESRI coverages in an Albers Conic Equal-Area projection in the North American Datum of 1927 (NAD27). They include the following:

1. ARCPNT - A coverage containing both hydrometric gauging stations (as points) and their effective and gross watershed boundaries (as lines). Each hydrometric gauging station point includes 19 pieces of information, namely its number, name, location, type (stream, lake or relocated), six levels of small- to medium-scale watershed identification, regional location, naming authority, six drainage basin statistics, and a "selection" identifier. In developing polygons of incremental areas of gross drainage, several stations may share the same polygon. For calculation purposes, only one station links the network downstream. That "selected" station is declared primary (SELECT = 1). The lines (effective and gross watershed boundaries) contain a code which differentiates between gross and effective boundaries.
2. NET - A line coverage that "connects" the hydrometric gauging stations using a network of downstream-oriented straight-line segments. This source product is fundamental to the digital summing of incremental drainage areas. It maintains upstream/downstream relationships between stations with line topology that is structured to "point" network segments downstream. That structuring allows for gross and effective drainage areas to be summed (by station) using upstream traces. The line attribute GROSS_ONLY signals whether or not the effective drainage areas above the network segment are to be included in effective drainage area summations for stations downstream of it. (See CASCADING NETWORK METHODOLOGY Section).
3. TICS - A coverage of some 11,000 tics, used in digitizing, to coordinate map table coordinates to ones of the mapsheet. The tic coverage contains all the tics from all the maps from which the drainage boundaries have been digitized. The tics are accessed when portions of maps need re-digitizing for various reasons. The tics "know" their mapsheet's name, number, edition, year and the RMS value of its map registration.
4. SHTS_CA - the mapsheet outlines of all digitized Canadian maps. This index coverage consists of the boundaries and metadata for each of the project's Canadian mapsheets (1:50,000 and 1:250,000 scales). The metadata consists of the mapsheet number, name, edition, year of printing or copyright, datum, the root mean squared (RMS) digitizing value.
5. SHTS_US - the mapsheet outlines of all digitized american maps. This index coverage consists of the boundaries and metadata for each of the project's american mapsheets. The metadata consists of the mapsheet number, name, edition, year of printing or copyright, datum and the root mean squared (RMS) digitizing value for 1:24,000 and 1:62,500 scale mapsheets. For 1:250,000 scale mapsheets, only the mapsheet number and name were recorded.

Output Products

All output products are generated from the source products ARCPNT and NET. They include eleven ESRI coverages and one ASCII or INFO file. All output coverages are in the "GEOGRAPHIC" projection (latitude and longitude coordinates) in the North American Datum of 1983 (NAD83). They are downloadable from www.agr.ca/pfra/gis/gwshed.htm.

1. AREAS.TXT or AREAS.DAT - GROSS AND EFFECTIVE DRAINAGE AREAS FOR HYDROMETRIC GAUGING STATIONS OF THE PFRA WATERSHED PROJECT (two files: a comma-delimited ASCII text file, areas.txt , and an INFO file AREAS.DAT) Both files contain the information found in addendum No. 8, namely, station number, station name, station location, and gross and effective drainage areas.
2. EFF - EFFECTIVE DRAINAGE AREA OF THE PFRA WATERSHED PROJECT (polygon coverage) - the area of effective drainage of the project's extent. The EFF polygon is built by joining (dissolving) all the EFFECT polygons. Many non-contributing pockets remain, in fact the coverage is in essence, the complement of the NON polygons for the project's extent. This product was created primarily for mapping purposes, and as such, retains no attributes.
3. EFFECT - EFFECTIVE DRAINAGE AREAS OF THE INCREMENTAL DRAINAGE AREAS OF THE PFRA WATERSHED PROJECT (polygon coverage) - the collection of areas of effective drainage of the incremental gross drainage areas (gross polygons) of the project's extent (See next section: Concepts). These EFFECT polygons are built by removing the non-contributing areas from the gross polygons. For analysis purposes, all information found in the GROSS polygons, except the two related to non-contributing areas, is retained.
4. GROSS - INCREMENTAL GROSS DRAINAGE AREAS OF THE PFRA WATERSHED PROJECT (polygon coverage) - the collection of incremental gross drainage areas (gross polygons) of the project's extent (See next section: Concepts). The gross polygons are built from the incremental drainage area lines of the drainage area boundaries and the primary hydrometric gauging stations. In developing these GROSS polygons, several stations may share the same polygon. For calculation purposes and topologic reasons, only one station may link the gross polygon downstream via the network.
5. LINEWORK - LINES OF GROSS AND EFFECTIVE DRAINAGE AREA BOUNDARIES OF THE PFRA WATERSHED PROJECT (line coverage) - the line component of the source product ARCPNT. Lines delineating the incremental drainage areas of the hydrometric gauging stations have their LINEWORK-IDs = 3 while the boundaries between the effective and non-contributing portions of those areas have their LINEWORK-IDs = 2.
6. MAJ_BAS - MAJOR BASINS OF THE PFRA WATERSHED PROJECT (polygon coverage) The major basins are agglomerations of the project's gross drainage areas into 14 basins of major river or lake reaches (of the Saskatchewan or Athabasca River scale) for the project's extent. They average around 160,000 km2 in size.
7. MAJ_SYS - MAJOR DRAINAGE SYSTEMS OF THE PFRA WATERSHED PROJECT (polygon coverage) The major drainage systems are agglomerations of the project's gross drainage areas into the sub-continental sized basins of the project's extent. It consists of the 3 main drainage systems: Arctic Ocean, Hudson Bay and Missouri River.
8. NETWORK - HYDROMETRIC GAUGING STATION NETWORK OF THE PFRA WATERSHED PROJECT (line coverage) This is a copy of the source product, NET (See NET in section Source Products, above). It is provided to help explain how upstream/downstream connectivity is handled. It is a network for hydrometric gauging stations and drainage basin area calculations. It is NOT a river system network!
9. NON - AreaS OF NON-CONTRIBUTING DRAINAGE OF THE PFRA WATERSHED PROJECT (polygon coverage) - the collection of all non-effective (non-contributing) areas of the project's extent. Each area "knows" of what catchment (gross polygon) it is a part.
10. STATIONS - HYDROMETRIC GAUGING STATIONS OF THE PFRA WATERSHED PROJECT (point coverage) - approximately 2800 hydrometric gauging stations. Each gauging station has 19 pieces of information: its number (STN_NUMBER), name (STN_NAME), location (STN_LOCATION), type (stream, lake or relocated) (STATIONS-ID), six levels of small- to medium-scale watershed identification (WS_MAJ33, WS_MAJ12, SUB_EC, SUB_PF, MAJ_BAS & MAJ_SYS), regional location (REGION), naming authority (AGENCY), six drainage basin statistics (AREAG, AREAC, AREAN, AGSUM, ACSUM & ANSUM), and a "selection" identifier (PRIMARY). (See section above, ARCPNT)
11. SUB_EC - ENVIRONMENT CANADA SUB-BASINS OF THE PFRA WATERSHED PROJECT (polygon coverage) The Environment Canada sub-basins are 41 agglomerations of the project's gross drainage areas whose station numbers share the same first 3 characters. The Environment Canada convention for hydrometric gauging stations is hierarchical. This means all stations whose number starts with "05" would fall in the same basin, as would those with the first 4 letters (e.g. "05HA") fall in the same sub-basin. Sometimes the stations were mistakenly numbered as being in one sub-basin while in fact they lie in another. Those "mis-numbered" stations have been reassigned to their proper 3-character basin (example "05H") while their numbers have remained unchanged. The sub-basins follow EC's convention. They average around 55,000 km2 in size.
12. SUB_PF - PFRA SUB-BASINS OF THE PFRA WATERSHED PROJECT (polygon coverage) The PFRA sub-basins are agglomerations of the project's gross drainage areas into 40 PFRA-tailored sub-basins. Like the SUB_EC polygons, these average around 55,000 km2 in size. These match the Environment Canada sub-basins except in areas of PFRA interest, where the sub-basins are refined, and in areas of less interest, where EC sub-basins are fused.

CONCEPTS

The Watershed Project is based on the hydrologic concepts of gross, effective, and dead drainage areas. The gross and effective drainage concepts originated with a paper by Stichling and Blackwell (1957). Central to this research is the idea that "contributing" drainage areas on the Canadian Prairies fluctuate by year, by season, and by event because of the glacial landscape and the climate. For the Watershed Project, Godwin and Martin (1975) restated Stichling and Blackwell's concepts of gross and effective drainage areas as follows:

The gross drainage area of a stream at a specified location is that plane area, enclosed by its drainage divide, which might be expected to entirely contribute runoff to that specified location under extremely wet conditions. The gross drainage boundary is the drainage divide (i.e. the height of land between adjoining watersheds). The effective drainage area is that portion of a drainage basin which might be expected to entirely contribute runoff to the main stream during a flood with a return period of two years. This area excludes marsh and slough areas and other natural storage areas which would prevent runoff from reaching the main stream in a year of "average runoff". A third important concept is that of dead drainage. Drainage is considered dead if there is no outflow from an area even under very wet conditions. This situation is common on the Canadian Prairies where major depressions having sloughs and shallow lakes with no outlets are usually associated with dead drainage. A dead drainage basin includes all of the area tributary to such a depression.

Both the gross and effective drainage boundaries appear to be distinct lines, but in practice they are not. In theory, a gross drainage boundary is a definite line because it is based solely on topography. However, in areas of poor drainage, gross drainage boundaries become less distinct and other physiographic factors such as slope, drainage patterns, and depressional storage are used as visual cues in the delineation process. Effective drainage boundaries are more conceptual because they pertain to the natural average runoff (approximately the two-year flood event) and are based mostly on hydrologic factors rather than on topography alone. Because of the non-distinct nature of the boundaries, an appropriate workable method for delineation was developed. A complete discussion of the drainage boundary delineation methods can be found in Hydrology Report #104 (PFRA Hydrology Division 1983). In the process of migrating the database to a GIS environment, a major conceptual change was made which involved the consideration of dead drainage areas. Previously (Hydrology Report #104, 1983), the five major dead drainage basins (Maple Creek, Old Wives Lake, Pakowki Lake, Quill Lakes, and Sounding/Eyehill Creek) were not considered to be part of any adjacent major drainage basin. That is, their respective areas were not included in the tabulated gross drainage area values of gauging stations in the external basin into which they would ultimately drain if water levels rose high enough. However, it was found that many (but not all) less significant dead drainage areas were included in the tabulated gross drainage area of values of gauging stations in the external basin into which they would ultimately drain. Thus, a decision was made to provide a consistent approach by including all dead drainage areas in the tabulated gross drainage area values of their associated drainage basin.

Drainage Delineation Methodology

The paper maps are topographic maps onto which gross and effective watershed boundaries and hydrometric gauging station locations have been marked. The watershed boundaries have been delineated to represent the area tributary to each hydrometric gauging station and it is important to note that this method is different than a stream-order method where each watershed is defined on the basis of map scale and stream order.

Gauging Station Source

The large majority of gauging stations in this project are those established by the federal Water Survey of Canada. They are augmented with a small number of provincial and american gauging stations. Gauging station locations were determined from published latitude/longitude coordinates or from Dominion Land Survey descriptions.

THE MANUAL TO DIGITAL MIGRATION EFFORT

The gross and effective drainage boundaries were converted (digitized) into digital format for the development of a GIS database to provide a more automated process.

The Manual Creation/Updating Procedure

In the traditional manually-maintained Watershed Project, the process for delineating drainage basins and tabulating their gross and effective areas was as follows. The hydrometric gauging stations were marked on a paper topographic mapsheet (usually 1:50,000 scale), and the gross and effective drainage basin boundaries were delineated. The areas of the resultant gross and effective polygons were measured with a digital planimeter and the results carefully checked. An index card was kept for each planimetered mapsheet. Each gross and effective area was recorded on the index card. The stations and drainage area boundary representations were transferred to 1:250,000 scale mapsheets. The areas from the index cards were ascribed to the respective polygons on the 1:250,000 scale mapsheet. (The use of 1:250,000 mapsheets permitted viewing sixteen 1:50,000 mapsheets at a glance which provided a better overview of the gauging station basins.) Tabulation sheets were set up for each gauging station. The gross and effective drainage area values from the 1:250,000 scale mapsheets were added to the tabulation sheets and summed. Note: each (gross/effective) polygon area is part of each downstream gauging station's drainage area, and therefore must be recorded on their respective tabulation sheets as well. The gross and effective drainage areas for each gauging station were published in Hydrology Report #104 and subsequent addenda (Martin 2001). In addition, gross and effective area boundary representations were traced onto mylars. Updates, caused by a change in gauging station location or a change in drainage, required that the gauging station(s) or boundaries be redrawn on the paper maps and the entire process redone. The revised values would be incorporated into an addendum to the Hydrology Report #104. It was felt that this labour-intensive process could be automated and shortened by moving the database into the digital environment.

The Migration

Before digitizing commenced, an accounting of all required mapsheets was done. Most of the mapsheets were Canadian 1:50,000 scale National Topographic Survey (NTS) maps, but the project also included some 1:250,000 scale NTS maps, and American United States Geological Survey (USGS) 1:24,000, 1:62,500; and 1:250,000 scale maps. Digital index map coverages (Canadian and American) of the outlines of all project mapsheets were created. Most Saskatchewan maps were digitized by Saskatchewan Water Corporation using PC arcInfo. All other maps (from Alberta, Manitoba, British Columbia, Ontario, NWT and United States) were digitized by PFRA Hydrology Unit using UNIX arcInfo. All subsequent GIS work was done with ArcInfo. A digitizing RMS error standard was set at 0.003. Some mapsheets exceeded the standard because of their poor condition. Digitizing metadata (consisting of mapsheet number, name, edition, publication year, projection and datum, tic numbers, who digitized, date digitized, and RMS error) were recorded in a binder and later transferred to a spreadsheet. Digitizing was done in the map's native projection. Most were Universal Transverse Mercator (UTM) projections in the North American Datum of 1927 (NaD27). Many others, mostly American, were either Transverse Mercator or Lamberts Polyconic (State Quadrangle maps). Some of the newer maps, or ones that were re-digitized, were NAD83. To allow all the data to be placed into one large coverage, a common projection was needed. An Albers Conic Equal-area projection was chosen because it is well suited to mid-latitude land masses that are predominantly oriented east-west while preserving the property of area well. Minimal distortion of areas was very important for watershed area calculations. All coverages were projected into Albers from their digitized projection. Edge-matching tied all the maps into one seamless coverage. It was an arduous process because arcs of adjacent mapsheets often did not match. This discontinuity was attributable to low relief topography, digitizing being done without the context of the adjacent mapsheet, different scales of adjacent sheets, and the thickness of the hand-drawn lines. Hand-drawn boundaries on the mapsheets range in thickness from 0.5 mm to 1 mm. A 1 mm thick line represents 24 metres on the ground at a scale of 1:24,000; 50 metres at 1:50,000; and 250 metres at 1:250,000. The edge-matching process resulted in a seamless coverage of arcs and points that required quality checks. It was decided that quality checks would best be handled during the creation of data products. The manually-generated database helped in establishing the network of linkages between hydrometric gauging stations.

The Digital Updating Procedure

Currently, when a change in drainage pattern is noted, or a hydrometric gauging station is added or relocated, the paper map is altered to reflect the changes. The digitizer is automatically oriented to the map's coordinate system and the changes are digitized into a temporary coverage. Those changes are reintegrated into the source digital product under supervision. Once incorporated, the required area calculations are processed, and the 10 output products are reproduced within 75 minutes. Eventually, digital NTS maps or ortho-rectified images will be used to directly update the source coverages.

Considerations and Discussion

During the migration, discrepancies between the manually- and digitally-generated databases occurred. Differences were attributed to problems in the digitizing, duplicate or missing gauging stations, or changes that were implemented to accommodate the requirements of digital environment. At the time it was thought that perhaps the discrepancies may be due to cumulative errors (boundary delineation, boundary delineation line thickness, digital planimetering, digitizing, and projection). However, the reason for such discrepancies became clearer as the migration progressed. The manually-generated database was used to initially corroborate the validity of the digital network that was established. During the process, a substantial number of errors in the manually-generated database were identified and corrected through a rigorous reconciliation process. Ultimately, the comparison of manually-generated and digitally-generated values of gross and effective drainage areas was quite good (differences generally much less than 1%). In most cases, the differences between the two databases were generally found to be planimetering errors in the manually-generated database. No attempt was made to rectify all differences because the considerable time and effort that would have been required was deemed to be unjustifiable. Spot checks indicated very close agreement when manually-generated drainage areas were re-planimetered. Thus, the digitally-generated values are considered to be the most current and appropriate. Henceforth, the GIS database will be maintained and should be considered the authoritative database for gross and effective drainage areas.

CASCADING NETWORK METHODOLOGY

1. For each catchment in a selected area of interest, gather the gross watershed arcs and a single selected hydrometric gauging station to create a gross watershed polygon coverage which supplies the gross (incremental drainage) watershed area for each gross polygon (catchment).
2. Create another polygon coverage of effective/non-effective polygons by gathering all the gross and effective arcs, and the same "selected" gauging stations together. Identify all polygons as either effective (contributing) or non-effective (non-contributing).
3. Combine both the gross and the effective/non-effective polygon coverages from steps 1 and 2 to tie each effective and non-effective polygon to its gross watershed polygon.
4. Calculate gross, effective, and non-effective areas for each gross polygon.
5. Transfer the area information from the polygon labels to the nodes of the network. This step is necessary because upstream area calculations are summed using network functionality (i.e. through upstream node "areas" summation).
6. Through upstream tracing, identify all upstream nodes of each gauging station and sum the gross, effective and non-contributing areas. The sums are stored in each node (selected station). In cases where drainage occurs through non-contributing portions of a catchment (i.e.dead drainage area), only gross area calculations are summed.
7. Transfer the newly-calculated area sums from the network nodes back to the original gauging stations (points).

PRODUCTS

While versions 1 to 3 have been static products, the current version (4) is dynamic. It will be updated as changes are noted, possibly changing on a daily basis, to provide the most up-to-date rendition of all output products as possible. It is the final version of the PFRA Watershed Project. Two additional output products have already been added: effective drainage area for the project (eff); and effective drainage polygons for incremental drainage areas (effect). all products are time stamped in the metadata. Presently the paper mapsheets are still being maintained but they will eventually be phased-out and all revisions to the database will be done digitally. Future updating will be done directly into the source coverages using either digital NTS maps or ortho-rectified images. The entire process will be moved to NAD83.

CONTRIBUTORS

Many people have contributed to this project. It was initiated by L. K. Szojka, Engineer-in-Charge of the PFRA Hydrology Division in 1970. The scope of the project was subsequently modified and expanded over the years by PFRA Hydrology Division Chiefs/Managers R. B. Godwin, D. W. Lawson and F. R. J. Martin. Skilled and experienced Hydrology Division technicians turned the vision into reality. The bulk of the work was done by P. O. Meid and G. T. Miller with database maintenance support provided, as required, by D. a. Schoenfeld and M. J. Hammer. The long-term commitment of Miller and Meid to this project provided consistent drainage area delineation efficiency and uniformity which would otherwise not have been achieved. In 1994, the decision was made to move the watershed delineations and the area calculation process into a digital environment. The GIS Unit in PFRA was charged with this task. W. R. Harron and D. L. Ackerman managed the move into a digital (ArcInfo GIS) environment and still oversee the GIS component of the project. Digitizing was done by Miller et al. (Hydrology Division) with K. Graham (Saskatchewan Water Corporation) digitizing most of Saskatchewan. The edge-matching and subsequent validating of gauging station locations against files from Environment Canada was completed by P. L. Cherneski. The cascading network and updating processes were facilitated by the development of AML scripts by Cherneski, G. M. Prince and Ackerman. J. T. Ashton has ongoing responsibility for project development, product validation and distribution.

SUMMARY

REFERENCES

Godwin, R. B. and F. R. J. Martin, 1975, Calculation of Gross and Effective Drainage Areas for the Prairie Provinces, Canadian Hydrology Symposium - 1975 Proceedings: Winnipeg, 5 pp.

Martin, F. R. J., 2001, Addendum No. 8 to Hydrology Report #104, agriculture and agri-Food Canada PFRA Technical Service: Regina, 109 pp. PFRA Hydrology Division, 1983, The Determination of Gross and Effective Drainage areas in the Prairie Provinces, Hydrology Report #104, Agriculture Canada PFRA Engineering Branch: Regina, 22 pp.

Stichling, W. and S. R. Blackwell, 1957, Drainage area as an Hydrologic Factor on the Canadian Prairies, IUGG Proceedings, Volume 111, Toronto, Ontario, np.