Two digital data layers were created for the Oswegatchie-Black River drainage area using GIS. These geographical data layers were interpreted with an existing lake chemistry database for the region. The new data layers were watershed boundary maps of ponds with labels capable of identifying complex (i.e., multiple pond) watersheds and wetland maps showing covertypes using National Wetlands Inventory labelling conventions.

Originally it was proposed that all subcatchments identified by permanent streams on U.S. Geological Survey (USGS) 1:24000 topographic maps be delineated. However, careful examination of the base maps indicated that it was neither practical nor ultimately meaningful to delineate watershed boundaries for all streams. Stream-level watersheds would have been extremely complex and time consuming to map, digitize and label and beyond the scope of this project. Because the smallest unit for which water quality data were available was at the pond or lake level with only a few exceptions, stream-based watershed delineations would not have been meaningful. It was decided, therefore, that the most appropriate watershed units for this effort would be all ponded waters identified by the NYS DEC Biological Survey Unit as having water quality or fisheries data. In reality, more ponds than are identified and numbered for this study appear on USGS topographic maps.

All ponded waters identified by the Biological Survey Unit were verified using the procedure described in Appendix A. A total of 1223 ponds were identified within the Oswegatchie-Black River drainage area. Pond identification numbers were then transferred onto two complete sets of the most recently available USGS topographic quadrangles. When they were available, the newer basemaps (1:25000) were used, and when these were not available, the older maps (1:24000) were used. Appendix B provides a list of which topographic quadrangles were used and a location map.

Independent determinations of watershed boundaries were made by two individuals with the same set of instructions. Watersheds were delineated starting and ending at the pond or lake outlet as shown on the topographic map using conventional delineation methods (USDA 1977). When the exact location of the pond outlet was in question, the field-checked ALSC bathymetric maps were used.

Boundary differences on the two sets of watershed maps were reconciled by a third person overlaying the maps on a light table. The discrepancies were of two types: either simply mistakes or errors. Errors resulted from different judgement calls which usually occurred in areas of low relief or where the pond outlet location was uncertain. Judgement differences were checked with 1968 (1:20000) panchromatic aerial photos and the ALSC bathymetric maps. In some cases, watershed boundaries were also field checked. Use of backup photography and field checks at this stage served as an initial watershed QA/QC. Final watershed boundaries were transferred onto a clean set of topographic maps by a fourth person in preparation for digitizing, providing additional QA/QC. For the final map product, only non-folded maps were utilized and watersheds were digitized from these sheets. All watershed boundaries were hand digitized with additional QA/QC.

Initially, watershed labelling was not expected to require much time and effort. The plan was to adopt a labelling scheme compatible with the existing 11-digit hydrologic code developed by the Soil Conservation Service (SCS) for New York State at 1:500000 (USDA 1981). Early in this project, SCS and USGS were consulted and it was decided to use the 11-digit code and attach an additional 3-digit code unique to each of the 1223 watersheds within the Oswegatchie-Black (OB) basin.

It was discovered that although the existing codes did reflect nesting of 11-digit subwatersheds within larger 8-digit watersheds, the existing 11-digit code was not a flow-oriented code. Since a watershed label related in a flow hierarchy was essential and fundamental to this project, one had to be found or developed. The difficulty was that many watersheds in the study area are complex, draining perhaps as many as 200 ponded subwatersheds (e.g., Unnamed pond 040260). In order for the watershed boundaries to be meaningful for this study, it was necessary to identify the entire watershed area that flowed into each of the ponds, not only the immediate watershed surrounding the pond. An existing method was not available.

As a result, a literature search was conducted on watershed labelling as well as consultation with experts in both GIS and watershed modelling. Regional experts as consultants were essential (i.e., New York and Vermont) because of their familiarity with the surface water flow patterns and because many are involved with ongoing GIS developments through the Lake Champlain Basin Program (LCBP). Unfortunately, the LCBP did not label watersheds beyond the 11-digit SCS code nor did they develop a flow hierarchy for ponded waters.

The following steps were employed in creating the watershed data layer:

Delineating boundaries

Note: There is no distinction between the term 'lake' or 'pond' in this discussion; they both mean a ponded water, regardless of the size of the waterbody or its biota.

1. Ponds to be delineated were identified on 1:24000 or 1:25000 scale USGS topographic map and labelled on the paper map by their unique ALSC pond number. Metric maps (1:25000) were used when available because they were recent and more accurate. When 1:25000 scale maps were not available, the existing 1:24000 scale maps were used. Maps used are listed in Appendix B.
2. A list of identified ponds was made for each 7.5 minute quadrangle map (Appendix C). This became an important reference list.
3. On the topographic maps, watersheds were delineated for each pond starting from the outlet following along high land peaks at the rim of the watershed basin and back to the outlet (USDA 1977). This step was conducted independently by two delineators using two sets of identical topographic maps.
4. Watershed boundaries identifying the SCS 11-digit watersheds were transferred from the 1:500000 scale map (USDA 1981) to the topographic maps.
5. Once the two sets of watershed delineations were complete, they were reconciled by a third individual using a light table and overlaying the two maps. Mistakes were corrected and errors were minimized. Potential for error occurred largely in areas of little topographic relief (flat) or where the exact pond outlet location was indistinct. In several instances more than one outlet was present for an individual pond. Errors were minimized by using collateral 1:24000 black and white photography and using the ALSC field notes defining locations of inlets and outlets.
6. Once the reconciled watershed(s) map was finalized, all the boundaries were redrawn on new flat (unfolded) topographic maps using a light table. After a final check was made for mistakes, the watershed boundaries were digitized from these final maps.

Labelling watersheds

7. All watershed boundaries were digitized with a single value assigned to all the arcs. Using screen editing techniques, the watershed database ws updated by assigning unique values to the SCS 11-digit watershed arcs, the watershed boundary between the Oswegatchie and Black rivers, and the outer-most boundary of the Oswegatchie-Black River watershed study area which was the Adirondack Park boundary.
8. A basemap was prepared for watershed labelling by combining the watershed polygon map with map files of quadrangle boundaries, DEC pond ID numbers, and pond locations as reference. The pond locations were taken from a separate file locating pond centroids developed by the DEC Biological Survey Unit. The DEC pond location file showing pond outlets was not as useful because pond outlets were too close to the edge of the watershed polygons. The entire watershed area was plotted at 1:70000 which proved to be a workable scale for labelling. The pond locations were verified on the original topographic maps and corrections made. For instance, some polygons had no pond location labels or had two labels because either the watersheds were small or the label locations were in error in the locator file. For polygons that contained no ponds because the pond was outside of the Park, the polygon was identified with an 'X'. After QA/QC, two sets of maps (4 sheets per map) were printed for the two independent labellers.
9. The SCS 11-digit watershed boundary locations were transferred onto the basemap (Step 8) from the original topographic maps.
10. Working independently, both labellers were provided with a set of the original marked topographic maps and with a blank watershed map (1:70000 paper). The first step was to determine the sequence of watershed flows from the topographic elevations and transfer flow-into arrows onto the 1:70000 watershed map.
11. After Step 10 was done on a couple of the 11-digit watersheds, the independent work was compared. An experimental 14-digit code, actually a 3-digit extension to the 11-digit code, was adopted. An accounting format was developed to identify simple and complex ponds using the SCS 11-digit code (Appendix D). The two labellers agreed to assign numbers in intervals of 5, that is, starting with 005, then 010, 015 etc. beginning at the top (highest elevation) of the watershed working down through the flow sequence of arrows.
12. Flow direction maps were produced independently by two people, with arrows showing directions of flow from the watershed polygons. Then a pond list was developed identifying all the 14-digit polygons that flowed into each of the ponds (Appendix E). This also was done by independently by two people.
13. The pond lists were independently typed as wordprocessing files. They were compared, checked for mistakes, corrected and finalized. Within the 11-digit watersheds, partial watersheds with no ponds occurring within them because they were outside the Adirondack Park boundary, were labelled 'X'. This became the pond identification flows-into file showing the 14-digit code for each pond (Appendix F).
14. The file in step 13 was imported and combined with DEC pond locator file from the ALSC using UTM coordinates. Each watershed polygon was identified by the ALSC 7-digit pond ID code.
15. Working from the screen (and with an error checking routine developed by J. Barge, APA) the map file was checked for errors. Errors appeared, for example, when watershed polygons and pond centroids UTM coordinates did not match or when polygons did not have ponds in them because the ponds occurred outside the Park boundary. The latter were labelled with an 'X'.
16. The final step in this process involved using the 'regions' function available in ArcInfo 7.0 to process the two files.

In summary, Figure 2 provides a task flow chart for the watershed mapping process. Data tolerances are described in the depicts Metadata section and Appendices of this report. Figure 3 the entire watersheds Oswegatchie-Black River watershed within the Adirondack Park showing of 1223 ponded waters. The whole park is not shown at the shown. The Oswegatchie drainage, top of the watershed, is The angular separated from the Black drainage by the heavy black line. boundary to the left is the Adirondack Park boundary. Some of the waterbodies appear in major grey within the watershed polygons.

Figure 2. Task flow chart for watershed mapping.

Collateral Data		Task and Primary Data Sources		QA/QC
		7.5' USGS Topographic Maps (1:24000,1:25000)
>Imagery: 1968 1:20000 B&W 1985-86 1:58000 CIR >Field Notes		WATERSHED DELINEATION		>Two separate interpretations with resolution by third party.
>ALSC database (with NYS Biological Survey pond locations)		WATERSHED DIGITAL DATABASE		>Determination of watershed flow patterns by two independent parties. >Add 3-digit subwatershed codes within 11-digit codes on map. >Create database file combining ALSC pond numbers with 14-digit watershed codes. >Use ArcInfo Regions function to join complex watersheds.
>USDA SCS Hydrologic Watershed Unit Map 1980 (1:500000)

Figure 3.

The aerial photography used to delineate and classify wetlands in the Oswegatchie-Black River drainage basin was flown for the U.S. Geological Survey National High-Altitude Photography (NHAP) Program. U.S. Geological Survey criteria for acquiring this imagery were aircraft altitude of 40,000 feet (12,190 m) above mean terrain, sun-angle of at least 30 to minimize shadows, stereoscopic coverage (approximately 60% forward overlap), no cloud cover, and minimal haze. The intent of NHAP was to obtain photography when deciduous vegetation was largely dormant (i.e., leaf off). NHAP photographs were taken with two mapping cameras: (1) a 6 in. (152 mm) focal-length camera loaded with panchromatic film produced photos at 1:80000 scale, and (2) a 8.25 in. (210 mm) focal-length camera loaded with color infrared film produced photos at a 1:58000 scale. The primary mapping imagery for the Oswegatchie/Black project was 1985-86 color infrared NHAP photography transparencies (1:58000). In addition, all wetlands delineated for the project were verified on 1968 large scale (1:20000) panchromatic paper prints. Panchromatic transparencies (1:80000) flown 1977-78 and field notes on 1:24000 USGS quadrangles also were used as collateral information.

The 1:58000 color infrared imagery (146 photos) was purchased from the U.S.Geological Survey - EROS Data Center. The 1:80000 and 1:20000 panchromatic photographs were acquired for other projects conducted by the Adirondack Park Agency and State University of New York at Plattsburgh.

The quality of the color infrared photography was quite variable. All photographs had excellent resolution, but color contrast on some flight lines was poor (e.g. dominant blues on 239 63-76), and leaf-out had occurred on flight lines in the western most part of the watershed (e.g. 329 45-56).

The following steps were employed in creating the wetlands data layer:

1. A clear acetate overlay was prepared for each 1/2 quad-centered aerial photograph (2 photographs per 7.5' quadrangle) and the quadrangle border was inked on the appropriate overlays.
2. All wetland areas were delineated and classified according to National Wetlands Inventory (NWI) techniques (Cowardin et al. 1979) on the photo overlay (Appendix G). It is noted that an additional Special Modifier was used in mixed upland/wetland habitats too homogeneous for separate delineations. It is described as '/U'. Although this designation is not an official Cowardin et al. (1979) definition, it is in use by various NWI offices and was suggested for use in this project by Ralph W. Tiner, Jr., U.S. Fish and Wildlife Service. The most commonly identified wetland covertypes encountered during this study are described in Table 1.

3. Polygon lines and labels were drawn using a 5x0 tungsten carbide pen with acetate ink. Because of the complexity of the delineations and the lack of space, shorthand notation was employed to label wetlands. Wetlands were verified on both the 1:80000 and 1:20000 panchromatic emulsions and topographic maps with field check notes were examined. As a QA/QC step, all overlays were examined by a second photo interpreter. Additionally, quadrangles with existing wetlands maps or detailed field inspections were examined by a third photo interpreter. Maps were compared with existing APA wetlands maps for Hamilton County. Forty-two Oswegatchie-Black wetland map quadrangles were examined using either APA official or tentative wetlands maps, NWI maps, and/or photo enlargements of 1:80000 wetlands interpretations.
4. A Bausch and Lomb Stereo Zoom Transfer Scope was used to remove scale changes and image displacement in transferring wetland polygons and linear features from aerial photographic overlays to mylar overlays on stable base 1:24000 B&W orthophoto quads. A separate mylar label overlay was made using an optical reducer/enlarger and a QA/QC check was conducted on wetland polygon boundaries and labels during this operation.
5. The wetland polygon data layers were either hand digitized into the digital data base or, on very complex overlays, were commercially scanned by Computer Sciences Corporation, Beltsville, Maryland. A total of 19 wetland polygon overlays were contracted for scanning. These maps were selected because they were too complex for hand digitizing. One complex map (Number Four 7.5' quadrangle) was hand digitized as a test, and approximately 10 working days (7 digitizing/3 editing) were necessary to complete the task. This was considered an inefficient use of time and the decision was made to use automatic scanning techniques.
6. A hard copy was printed for each digitized wetlands quadrangle, overlaid with the original 1:24000 polygon overlay, and checked for accuracy, including precision of line work, missing or extraneous linear features, polygon closure, and inappropriate dangles.
7. The wetland labels were entered into the digital data base as a series of columns using a digitizer and custom designed menu. The customized menu allowed labels to be entered by simply clicking the digitizing cursor, thus eliminating key entry errors. All polygons, including upland areas, received a label designation, and all unlabelled areas were clearly highlighted on the computer screen. Unified NWI labels were created by concatenating the columns in Dbase IV.

In addition, the labelling procedure was designed as an additional QA/QC on wetland polygons and linear features. Data tolerances are described in the Metadata section and Appendices of this report. A more detailed description of the wetland map QA/QC is provided in Appendix H. A task flow chart for the wetlands mapping process is shown in Figure 4. Maps showing wetland boundaries, wetland polygon labels and linear wetland labels for a sample 7.5' quadrangle (Oswegatchie) are provided in Figure 5, Figure 6, and Figure 7.

The distribution of wetlands varies across the region. For example, the Oswegatchie quadrangle contains 725 wetland polygons (Figure 5), whereas the Copper Lake quadrangle contains 2101 wetland polygons making it the most complex quadrangle in the 48-quadrangle study area. Figure 8 is a map of the wetland polygons of the Copper Lake quadrangle which can be compared for complexity with Figure 5.

Selected areas in Hamilton, Oneida and Lewis counties had existing detailed field inspection data from past visits with landowners to verify wetlands maps. These data were used to check the new interpretations as additional QA/QC. Generally, the new interpretations appeared to match well with the field information. A few quadrangles appeared to be missing small PFO1 wetland areas on the Oswegatchie-Black project wetlands overlay, probably due to leaf out conditions on the 1985-86 CIR photography. In other cases, the new interpretation picked up additional wetlands.

Figure 4. Task flow chart for wetlands mapping.

Collateral Data		Task and Primary Data Sources		QA/QC
		1985-86 1:58000 CIR Transparencies
>Imagery: 1968 1:20000 B&W 1980-81 1:80000 B&W >Field Notes		WETLAND DELINEATION		>Independent interpreter check of photo overlay.
>1985-86 1:58000 CIR imagery		WETLAND TRANSFER (ZTS)		>Superimpose photo/wetlands overlay with 1:24000 rectified wetland map using Kail Reflecting Projector.
>1985-86 1:58000 CIR imagery		WETLAND DIGITAL DATABASE		>Computer based quality control protocols. >Hard copy of digital files compared with 1:24000 rectified maps. >Wetlands and labels checked by two individuals.

The purpose of the small wetlands project was to develop and evaluate a method to systematically identify and accurately delineate wetlands smaller than one acre. It has been recognized that a considerable number of small wetlands are situated in Adirondack watersheds and, taken together, may have significant functional attributes in any given watershed. Therefore, mapping small wetlands is critical to any inventory or effects analysis.

Project staff proposed the following method:

1. Locate a representative geographic area with sub-acre sized wetlands via field inspection and air photo interpretation.
2. Locate airphoto targets near the center of several wetlands using Global Positioning System (GPS). Collect floristic data, assign covertype and water regime modifiers.
3. Differentially correct GPS data and download to GIS. Print maps on appropriate scale base.
4. Take appropriate scale airphotos of the small wetlands project area using fixed-wing aircraft.
5. Interpret the airphotos using a double-blind method. Compare interpretations and determine success of the approach.
6. Prepare recommendations for modification or acceptance of approach.
7. Prepare evaluation of techniques including GPS and GIS.
8. Compare to "conventional" techniques.

The standard resolution of one acre using National High Altitude Photography Program (NHAP) 1:80000 BW and 1:24000 BW collateral photography is based on the Agency's experience using these techniques to prepare and promulgate wetlands maps for five counties in the Park. The small wetlands project utilizes new NHAP 1:58000 CIR with 1:24000 BW collateral photography to attempt to reach a finer resolution. Aerial imagery was also obtained with twin EL 500 Hasselblad cameras with 50 mm Distagon lenses, coupled to a command unit and intervalometer. Color and color infrared vertical aerial photographs at 1:20000 scale with 60% forward overlap for stereoviewing were obtained for two locations. The film was processed by Precision Photo Laboratories, Inc., Dayton, Ohio.

A Pathfinder Basic model GPS from Trimble Navigation was acquired and a Trimble Community Base Station was installed at Hudson Hall, SUNY Plattsburgh. This included a roof antenna and a computer hook-up in the Remote Sensing Laboratory. UTM coordinates required to the nearest centimeter for this system were established by survey on April 5, 1994. The GPS computer software was field tested in 1994 and was found to have an acceptable 2-5 meter accuracy.

On May 4, 1994, a pre-leaf-out aerial photographic mission was flown over a portion of the Oswegatchie-Black study area to test the feasibility of small wetland (< 1 acre) delineation and classification from large scale 1:20000 aerial photographs and to assess the reliability of mapping small wetlands from 1:58000 NHAP imagery. Color (Kodak 2448 Aerochrome MS) and color infrared (Kodak 2443 Aerochrome) vertical aerial photographs at 1:20000 scale with 60% forward overlap for stereoviewing were obtained for two locations: the Cranberry Lake test site (Cranberry Lake 7.5 minute quadrangle) and the Inlet/Oswegatchie River test site (Newton Falls 7.5 minute quadrangle). The test sites, selected for numerous small wetlands and covertype diversity, were field-checked on foot and pre-targeted on May 2, 1994, with 2'x 2' white plywood targets whose positions were logged with a GPS rover unit linked to the base station. The targets were placed in selected wetlands to provide ground control points on the imagery and to determine coordinates for accuracy assessment of wetland planimetric location in the GIS data base. In addition, an approximately 20 mile swath with numerous small wetlands was photographed on the Number Four 7.5 minute quadrangle for future use in evaluating wetlands delineation and classification accuracy from National Aerial Photography Program (NAPP) 1:40000 imagery. NAPP photography is expected to be completed for the Adirondacks in 1996. Figure 9 provides a task flow chart for the steps involved in the small wetlands mapping investigation.

Figure 9. Task flow chart for small wetlands project.

Collateral Data		Task and Primary Data Sources		QA/QC
		GPS Data & Field Observation
		Specially Flown 70mm Color and CIR Imagery
>Imagery: 1968 1:20000 B&W 1980-81 1:80000 B&W 1985-86 1:58000 CIR 1995 1:40000 CIR		SMALL WETLANDS PILOT PROJECT		>Independent interpreter check of photo overlay.
>Field Notes

Poor fens and ombrotrophic bogs were designated as critical wetlands. These peatlands are susceptible to perturbation from anthropogenically derived inputs of acidic and nutrient deposition. The project attempted to identify and locate these wetland covertypes via airphoto interpretation coupled with field investigation.

The following method was proposed to identify critical wetlands:

1. Identify a representative area in the Oswegatchie-Black River watershed that had typical emergent marsh or wet meadow signatures (EM), and coniferous swamp or shrub swamp (FO4, SS3, SS4) signatures.
2. Assign each polygon with the above signatures a classification of fen/non-fen or bog/non-bog.
3. Conduct field investigations to the above polygons to determine if the classification assigned is accurate. Field investigation included filling out standard National Wetland Inventory field data forms.
4. Analyze data collected to determine feasibility of identification via airphoto interpretation.

Hydrography is a separate digital map coverage of lakes, ponds, rivers and streams derived from NYS Department of Transportation (NYSDOT) 1:24000 planimetric maps. When this project was proposed, it was hoped that the existing data would be suitable, but when the existing data were acquired from the NYSDEC Biological Survey Unit, several problems were encountered. First, total coverage of the 48 quadrangle area was not available. As of December 1, 1993, 39 quadrangles were available from NYSDEC. The APA staff scanned and vectorized the remaining 9 quadrangles.

Also of concern was the quality of the digital hydrography layer relative to the newer 1:25000 USGS topography maps used for the watershed delineations. The existing data layer was developed by the Biological Survey Unit using NYSDOT 1:24000 planimetric basemaps with dates ranging from 1946 to 1977. Originally, updating the maps based on best available USGS base information in association with field staff expertise was considered and all available quadrangles were compared and marked ready for digitizing. However, in the final analysis, because the task was large and the gains were relatively small, it was decided to wait until some wetlands maps were digitized to assess the hydrography registration with the wetlands layer.

After the wetlands layer was completed, a number of maps were randomly checked for existing hydrography registration with the open water 'OW' label of the wetlands layer. There were differences as can be shown on an example map (Figure 10).

The following are explanations for the map differences:

1. different photography, difference in emulsions and scales. The hydrography layer was scanned from topographic maps developed during 1960-70 using 1:20000 black and white photography, while the wetlands map was developed from 1:58000 color infrared photography flown in 1985-86;
2. natural differences are likely to have occurred in the 15 to 25 year time span between maps. Another natural difference could be seasonal water level variation (imagery time of year);
3. different mapping missions. For example, shorelines on the topographic map might have included a cattail emergent marsh area, whereas the same area on the wetland map would not be considered openwater;
4. beaver influence during intervening years;
5. terrain, flat areas have a greater potential difference than steep areas;
6. error, for example, the photogrammetric rectification process provides that at least 90% of all horizontal features are presented to within 1/100 inch, and;
7. mistakes.

Although planimetric differences do exist for hydrography in the two databases (i.e., existing hydrography and wetlands), it does not affect the spatial relationships that are being evaluated for this project. For instance, it does not affect the extent of wetland area in any watershed. Also, for relationships between wetland location and lake open water, the 'OW' label is available on the wetland layer and is the more reliable. When new hydrography becomes available, the differences described above will be minimized.

Metadata refers to information about the information. Metadata are only available for watershed and wetlands data layers, the two databases that were created for this project. Appendix I contains the watershed metadata and Appendix J contains the wetlands metadata. Other GIS layers used in some of the map presentations, such as the hydrography, are available elsewhere. Likewise, the metadata on the ALSC data are not available from this project.

Continue reading next section of OB1 Report -- Results and Discussion