Index

• Introduction
• Study Area
• Materials
• Classification System
• Project Methodology
• Accuracy Assessment
• Results
• Conclusions
• Recommendations
• Acknowledgments

The riparian lands of Custer County Colorado and the surrounding Pike/San Isabel National Forest are used for a variety of diverse applications such as outdoor recreation, livestock grazing, water supply, wildlife habitat and scenery. In order to properly plan and manage these valuable multiple use lands, the San Isabel Foundation (SIF) and the Colorado Division of Wildlife (CDOW) have developed a set of riparian overlays registered to 7.5' quadrangle maps in digital format for the entire County. These riparian maps will be one of several data layers in a county-wide geographic information system.

The project was undertaken by the San Isabel Foundation and the Colorado Division of Wildlife. It began in April 1997 and was completed in September 1997. The photo interpretation work and the data capture/conversion was done by James F. Ward and Associates. Mr. Ward had previously mapped riparian habitat for the Pike/San Isabel NF lands within the county and it seemed prudent to use the same interpreter for the non-National Forest land within the county. This should provide the necessary consistency between the USFS and non-USFS property.

The objective of the project was to identify and map selected riparian vegetation communities (comprised of one or more plant associations) in the county using high altitude color infrared large format camera (LFC) and NAPP photography. Overlays registered to 7.5 minute quadrangle maps were used to inventory riparian areas on a per class basis. They may also serve as a baseline for prioritizing land conservation projects by the San Isabel Foundation, land use planning, and monitoring the effect of land management practices.  

The project deliverables include a set of riparian overlays registered to 7.5' mylar orthophotoquads and a corresponding set of digital map products for use in an Arc/Info Geographic Information System (GIS). The overlays contain labeled polygons that define the specific riparian vegetation communities being classified. Delineated riparian habitat is edge-matched with existing U.S. Forest Service riparian data to ensure a consistent and seamless data layer. Another deliverable is a final report documenting procedures, personnel, results and discussion.  

The project was a coordinated effort involving the San Isabel Foundation, Colorado Division of Wildlife, and James Ward (Contractor). Responsibilities for each group were as follows:

San Isabel Foundation - Contract development and administration, field data acquisition;

The study area encompassed twenty-five 7.5' quads encompassing Custer County in south-central Colorado. Of these twenty-five quads, seven were mapped during previous projects. The eighteen quads mapped for this project were:

Bear Creek Devils Gulch Beck Mountain

Deer Peak Rosita Aldrich Gulch

Horn Peak Wetmore Hardscrabble Mountain

Mount Tyndal Westcliffe Beckwith Mountain

Electric Peak Florence SE Rockvale

Curley Peak Iron Mountain Hillside

Much of the area in the Wet Mountain Valley was irrigated which caused some difficulty in interpretation. An extra trip was made to the area in July 1997 by Mr. Ward in order to better understand and interpret the riparian lands in the Wet Mountain Valley. In some instances it was quite difficult to interpret true riparian areas from land that is seasonally influenced by irrigation.

The Pike/San Isabel National Forest allowed the San Isabel Foundation and the Colorado Division of Wildlife to borrow the Large Format Camera aerial photographs, several mylar topo maps and orthophotoquads, and the existing riparian maps. Through this loan of data, the Forest Service played a very significant role in the success of this mapping effort.

Large Format Camera (LFC) Aerial Photographs

The LFC, 9x18-inch, color infrared photographs were acquired during August 1988 with a NASA ER-2 aircraft. This photography was used to map riparian areas by the Pike/San Isabel NF and the aerial extent included a significant portion of non-USFS lands in Custer County. The following is a list of selected photo parameters.

Flight Number: 88-117, acquired by NASA, Ames Research Center

Sensor Package: A-4 configuration (HR-732 LFC and Wild RC-10)

Sensor Type: Itek HR-732 Large Format Camera

Focal Length: 24"

Film Type: High Definition Aerochrome Infrared (SO-131)

No. of Frames: 82

% Overlap: 60%

Aircraft Altitude: 65,000 ft. (MSL), yielding a nominal scale of 1:30,000

The quality of the photography was excellent. The color balance and resolution of the film were outstanding.

National Aerial Photography Program (NAPP)

Format - 9"x9"

Scale - the nominal scale was 1:40,000

Lens focal length - 8.25 inches

Date - September 1988

% Overlap - 60%

Number of Frames - 30

Emulsion type - Kodak 2443 transparency

The interpreter had to make some adjustment to convert from one photo type to the next. Some of the cottonwood on the NAPP came close to looking like some of the conifer on the LFC photography.

The LFC photography was used to map quads that had some USFS land on them. The NAPP photography was used to map the north end of the Wet Mountain Valley. The LFC photography has a better color balance and higher resolution than the NAPP photos. It is the preferable photo type to use for riparian mapping when it is available.

Maps

The Colorado Division of Wildlife supplied 1:24,000 scale topographic maps on stable-based mylar, orthophoto quad transparencies, a 1:50,000 scale county map, print outs of the digital riparian maps and a secondary series Forest map (1:125,000). The map products were used for the following purposes:

1. to serve as a base of reference for a photo index;

2. to aid the interpreter with reference to hydrography, roads, cultural features, elevations and contours;

3. orthophoto quads were used to provide 1:24,000 scale base maps for photo to map transfer of riparian areas; and

4. to layout the sampling system transects.

Equipment

All of the photo interpretation equipment used during the project was furnished by the contractor. An ordinary 4' x 3' light table, modified to handle film rolls provided an adequate light source. A Bausch and Lomb Stereo Zoom Transfer Scope was used to view a stereo model, interpret the riparian vegetation and delineate polygons directly on the overlays registered to the 1:24,000 scale orthophotoquads. Occasionally when a more detailed inspection was required, a 10x tube magnifier was used. A KOH-I-NOOR 000 rapidograph pen was used to delineate and label the very small polygons.

The digital maps were created by scanning the riparian overlays at 300dpi with a Hewlitt Packard drum scanner, and using an IBM PC with LTDOS software to create vector type riparian maps.

The classification system (Table 1) was developed by Dave Lovell, Colorado Division of Wildlife (CDOW). It was used for the identification and classification of existing riparian plant communities found within the project area. Potential riparian habitats were not delineated. Plant communities were classified to a minimum mapping size of at least one half acre with many going well below this limit. If the width of the riparian area was less than 80 feet, it was designated as an appropriately attributed line. (Some linear polygons were less than 80 feet wide). The classification system worked very well in relation to the scale, resolution, and emulsion of the LFC and NAPP photography.

The system was designed to be used with a GIS and to be easily aggregated into broader classes. It was also planned to correlate with the Pike/San Isabel system (Table 2) used on Forest Service land within the county. Although the labels are different, the content was very similar. The classification system was used for a similar project on the Arkansas River basin between Leadville and Pueblo, Colorado, and worked very well.

The project was done in two stages, photo interpretation and digitization; converting the hand drafted hard copy to a digital format compatible with Arc/Info geographic information system software.

Seven general project tasks were performed in order to derive the final products described earlier:

1. Photograph preparation

2. Photo interpreter training

3. Polygon delineation

4. Polygon labeling

5. Map digitizing

6. Accuracy assessment

7. Documentation

Photograph Preparation

Upon receipt of the LFC and NAPP photographs, a quality inspection was made and a photo index was drafted for each photo type on a USFS secondary map series at a scale of 1:125,000.

Photo interpreter Training

Training the photo interpreter is a very important step in any photo-interpretation project and should never be under emphasized. Ideally, the photo interpreter should visit the project site and view the objects on the ground that will be interpreted from the aerial photographs. Since the interpreter had previously mapped the USFS land within the county it was thought that a training trip would not be necessary. However, as the project progressed and the interpreter began working in the irrigated valley, it was decided that it would be prudent to visit the site and look at the irrigated lands on the ground. This trip enabled the interpreter to separate the riparian areas within the irrigated valley that would be riparian in a natural setting with a higher degree of accuracy. The Colorado Division of Wildlife and the San Isabel Foundation provided a great deal of very valuable information to the interpreter about the vegetation cover and ecology of the area that made the interpretation more accurate. The field trip lasted three days in July. Mr. Lovell continued to answer occasional questions about the riparian vegetation via telephone throughout the project.
 
 

Polygon Delineation

Riparian polygons and lines were delineated from the aerial photography using a Bausch and Lomb Stereo Zoom Transfer scope. A stereo model is set up on the scope and then optically manipulated to fit the black and white orthophotoquad (OPQ). Literally thousands of control points are available between the stereo model and the OPQ for registering the polygon to a true ground position. As the stereo model is interpreted, the polygons are delineated on a stable based mylar overlay registered to the OPQ rather than directly on the aerial photographs.

TABLE 1 CDOW/SAN ISABEL FOUNDATION COOPERATIVE RIPARIAN HABITAT MAPPING PROJECT CLASSIFICATION SCHEME
CATEGORY	MAP CODE
RIPARIAN TREES
Riparian Deciduous Tree - General	RT
Riparian Deciduous Tree - Aspen	RT1
Riparian Deciduous Tree - Cottonwood	RT2
Riparian Deciduous Tree - Russian Olive	RT3
Riparian Evergreen Tree	RE
RIPARIAN SHRUBS
Riparian Shrub - General	RS
Riparian Shrub - Willow	RS1
Riparian Shrub - Tamarisk	RS2
Riparian Shrub - Alpine Willow	RS3
Riparian Shrub - Gambel Oak	RS4
RIPARIAN HERBACEOUS
Riparian Herbaceous - General	RH
Riparian Herbaceous - Cattails/Sedges/Rushes (With permanent standing water)	RH1
Riparian Herbaceous - Sedges/Rushes (Waterlogged Soils)	RH2
Riparian Herbaceous - Sedges/Rushes/Mesic Grasses (Moist Soils)	RH3
WATER BODIES
Open Water - Standing	OW1
Open Water - Riverine	OW2
Open Water - Canal	OW3
OTHER RIPARIAN
Unvegetated	NV
Sandbar	SB
NON-RIPARIAN
Upland Tree	UT
Upland Shrub	US
Upland Grass	UG
Both polygon features and line features are mapped using this classification scheme, infrared aerial photographs, 7.5 minute topographic ortho-photos, and a minimum mapping unit of 0.5 acres. This classification scheme utilizes a dominant/subdominant methodology of describing riparian habitat. Unless a polygon is at least 75% homogeneous the dominant category is listed first followed by a slash ( / ) and the subdominant category. Example: RT1/RS1 = Aspen/Willow with aspen being the dominant category within the mapped polygon

 

The first interpretation step was to delineate riparian from non-riparian areas using the project color infrared photographs, (which is the best possible type of photo emulsion for mapping any type of wetland). Next, within the riparian areas, the vegetation differences were delineated. Separation of riparian vegetation classes were identified first by color, then texture, and finally by association with riparian topography. It is important to remember that riparian habitat is delineated by the landform it lies within as well as the vegetation it supports. It is for this reason that riparian mapping be done in stereo (3-Dimensional) and not from a single photograph.

A standard rule for all photo interpretation projects is that the boundary lines of all polygons along the border of the quadrangle map must smoothly match with their respective polygons on the adjacent interpreted map. As a new map was started, polygon edges and lines were brought onto the new map from the adjacent maps.

Polygon Labeling

Polygons were labeled on each quadrangle map as soon as all of the polygons on a particular quad had been delineated. Labels were placed on a separate overlay registered to the polygon overlay and the OPQ. Doing it this way keeps the polygon overlay clean for scanning which reduces the edit time on the digital phase of the project. It also adds to the consistency of the project and it keeps the interpreter from writing a label across a riparian area before it has been delineated.

Digital Processing

Each riparian polygon overlay was scanned with a Hewlitt Packard drum scanner at a resolution of 300 dots per inch and the resulting raster image was written to an RLC file. The technician was careful to set the threshold so the lines would stay as thin as possible and the very narrow riparian polygons would not collapse and become a thick solid line. Even with this careful attention, a significant amount of editing was necessary to achieve the desired products.

TABLE 2 USFS/CDOW COOPERATIVE RIPARIAN HABITAT MAPPING PROJECT CLASSIFICATION SCHEME
CATEGORY	MAP CODE
RIPARIAN TREES
Aspen	AS
Cottonwood	CO
Evergreen	E
RIPARIAN SHRUBS
Riparian Shrub - General	S
Riparian Shrub - Willow	W
Riparian Shrub - Alpine Willow	AW
RIPARIAN HERBACEOUS
Mesic Meadow	M
WATER BODIES
Open Water - Standing	L
Open Water - Riverine	ST
OTHER RIPARIAN
General Riparian	R
Unvegetated	NV
Sandbar	SB
NON-RIPARIAN
Upland Grass	GR
Both polygon features and line features are mapped using this classification scheme, infrared aerial photographs, 7.5 minute topographic ortho-photos, and a minimum mapping unit of 0.5 acres. This classification scheme utilizes a dominant/subdominant methodology of describing riparian habitat. Unless a polygon is at least 75% homogeneous the dominant category is listed first followed by a slash ( / ) and the subdominant category. Example: AS/S = Aspen/ Riparian Shrub with aspen being the dominant category within the mapped polygon.

A data capture and conversion software package called LTDOS was used to digitize the maps. The use of this software is a five step procedure:

4. Attribute each polygon and line.

5. Export polygons and lines in separate MOSS files compatible with Arc/Info GIS

The accuracy assessment field work was very intensive and required approximately 386 person hours to complete. The main objective of the accuracy assessment task was to obtain a ground verification and to estimate the overall accuracy of the photo interpretation classification work. It may also be used to make adjustments to the classification system for future riparian mapping in other Colorado counties.

Sample Size and Cluster (Transect) Layout

The accuracy assessment sampling design employed a stratified cluster sampling technique with transects serving as clusters or the primary sample unit (PSU). Each change of dominant vegetation type along a transect was a sample point or the secondary sample unit (SSU). There were 327 sample points spread out along 125 transects. The beginning point for the first transect was randomly selected along the main stream draining the valley. From this point, transects were placed at two mile intervals going both north and south to the boundary of the county. Other transects were placed on randomly selected tributaries, also at two mile intervals from the main stream up to the USFS or county boundary. By using this type of layout procedure, the samples were stratified by elevation, stream order, and vegetation type.

The transects began and ended at the riparian zone boundaries and were laid out perpendicular to the stream channels and then drafted onto the riparian overlays and the 7.5 minute quads which were taken to the field.

Only the dominant vegetation type was compared to the interpreted map because the combinations of dominant and sub-dominant was so large it became unmanageable. The dominant vegetation type was recorded on the field sheet. Ground photos also were taken at each transect site. On the back of each photo is the field crew number, camera number, and exposure number followed by the transect number the photograph depicts. Field forms and photographs are available for review at the SIF offices in Westcliffe, Colorado.

Field Data Collection Procedures

The accuracy assessment field data collection work was performed by seven people from the San Isabel Foundation, three Colorado Division of Wildlife employees, two employees from the Bureau of Land Management, and James Ward of James F. Ward and Associates. The field investigators were divided into five teams and each team was assigned a portion of the county to examine.

The teams were as follows with the team leader being the first name sited:

Team 1 - Pari Morse, Don Mercill, Sherri Patterson

Team 2 - James Ward, John Barnett, Carol Barnett

Team 3 - Derek Brixsey, Eric Wormsley

Team 4 - Chris Kloster, Peggy Kavookjian, Shannon Koch on 8/12/97

Seth McClean, Peggy Kavokjian on 8/13-14/97

Team 5 - Dave Lovell, Seth McClean on 8/12/97

Dave Lovell, Shannon Koch on 8/13-14/97

Although some of the field staff had limited experience in riparian habitat identification and mapping, each team had a team leader that was at least familiar with this subject and should be able to identify riparian habitat and vegetation. The interpreter spent 6 hours training the field crews in how to identify and collect the field data. Obviously, it is critical to the accuracy of the project that all personnel see the ground cover the same way. There were at least two subjective judgements that were encountered by field crews. One is the dominance of a vegetation type in a complex site and the other is the average width (a polygon had to be 80 feet wide to be delineated) of a vegetation type within an odd shaped polygon. The following data collection procedures were used:

1. One end of the transect was carefully located at the edge of the riparian zone. The direction of the transect was noted and the team proceeded to the opposite end of the transect noting changes in the dominant vegetation type as they crossed the riparian zone. Each team had a Global Positioning System (GPS) receiver to aid them in locating and recording the location of the transects.

2. A field data sheet (Figure 2) was used to record information relating to each vegetation class. Field data was later compared with the riparian maps (Figure 3) to develop the error matrix. Figure 4 is a ground photo of riparian transect 57a.

3. 125 transects and 327 samples were examined. Of these, 51 transects were added after the accessible planned transects were visited. Some transects had to be viewed at a distance due to land owners not allowing field crews onto their property. If the transects were not accessible or visible from a reasonable distance, the transect was shifted to a similar location as close as possible to the original site. These transects were designated with an "A" following the number. Figure 5 is a map indicating the location of all the ground checked transects.

Completed field data sheets and photographs are archived by the San Isabel Foundation at their office in Westcliffe, CO.

The results of the project are presented in two parts. First is a review of the resources required to do the job, followed by an analysis of the photo interpretation accuracy assessment.

Resources

Table 3 shows an accounting of the resources, in hours, spent on each phase of the project. The 386 hours identified with the field data collection work includes approximately 78 hours allotted to the training of 12 people in field data collection procedures. The resource figures include all project tasks.

Photo interpretation Accuracy Assessment

Based on the analysis of the collected ground data (comparing ground observation with the riparian map), an overall percent correct classification (PCC) of 74.0% was obtained for the riparian vegetation photo interpretation effort. The accuracy of riparian versus non-riparian is 96.3% and may be seen in Table 4. The Error Matrix shown in Table 5 identifies some areas of possible confusion. It is important to remember that there is a nine year difference between the date of photo acquisition and the date of the accuracy assessment field trip. There is also significant yearly and seasonal change in the valley because of irrigation. Additionally, the year the photography was flown (1988) was a dry year while this year, when the field data were collected, was quite wet. The overall vegetation PCC was 8-10% lower than anticipated, probably due to the above factors. Projects such as this usually yield a PCC of 82%-86%. One such project was done for the Pike/San Isabel NF in 1991-1992. The accuracy assessment done on a portion of the Forest south of Leadville, CO., provided a PCC of 83.6%.
 

TABLE 3 RESOURCE REQUIREMENTS
PROCEDURE	HOURS INVESTED
	Ward	CDOW	SIF	BLM	TOTAL
Photo Preparation	8				8
Field Study	40	32	16		88
Delineation	200				200
Labeling	180				180
Digital Processing	250	100			350
Hard Copy Production		32			32
Statistical Design	16				16
Field Data Collection	40	76	210	60	386
Field Data Analysis	40				40
Documentation	50	10			60
Administration	16	16	80		112
Contract/Project Design		80			80
Acquisition of Materials		40			40
TOTAL HOURS	840	386	306	60	1592

 

Photo interpretation error can be placed within 5 major categories. Three of these are a result of

subjective judgment differences between the interpreter and the field crews, one is a mis-identification because of similar appearance on the photos, and the last is illogical and extreme error that really makes no sense and is therefore assumed to be spatial error rather than classification error.

Much of the error found in this project falls into the first type, which is confusion between complex classes and is a result of subjective differences. For example, the interpreter may have classified a riparian area as RH3/RS1 and the field crews recorded the site as RS1/RH3. Even though both vegetation types were included, the polygon would be recorded as an error. This confusion also occurred when RT1/RS1 was compared with RS1/RT1; US/UG compared with UG/US, etc. There was also error encountered when the interpreter recorded a single vegetation type and the field crew called the interpreted type sub-dominant in a complex vegetation type. An example of this is a field call of RH2/RS1 and an interpretation of RS1. The opposite of this may also be found. The interpreter will continue to improve with experience, but this type of error will never be completely overcome and the user will have to be mindful of the error. There may also have been some error in the field data collection. There was significant difference between the five teams as follows 58%, 83.2%, 54.4%, 74.0%, and 79.5%. This 19% difference between the high and low PCC may possibly indicate a lack of consistency between field teams or that there are some extreme differences within the county.

Another type of subjective error is in transition zones such as when wet meadows (RH2) move into mesic meadows (RH3) or when mesic meadows (RH3) migrate into non-riparian (X). This problem is magnified if the photos were acquired during a wet season and the field verification done during a dry season or vice versa. It was also seen in a single polygon. For example, when field examining a

TABLE 4 ERROR MATRIX RIPARIAN VERSUS NON-RIPARIAN
Ground Checked Points	Photo-Interpreted Points			Omission Error	Percent Omission
	Riparian	Non-Riparian	Total
Riparian	315	12	315	12	3.7%
Non-Riparian	0	6	6	0	0
TOTAL CORRECT	315	6	321
Commission Error	0	12
Percent Commission	0%	67%
Percent Correct Classification (PCC) = 315 Correctly Classified of 327 Total Points = 96.3%

 

linear riparian area with a creek flowing down the center, it was divided into three classes - an RH3, moving into an RH2 and back to RH3. Whereas, the interpreter viewed the entire polygon, which could be a mile long, and generalized calling it RH3.

Occasionally, some very strange and illogical errors occur that do not make sense such as confusing OW1 with RS1 or RT1 with UG. Errors such as these may be due to spatial misplacement on the ground because they are so different in appearance on the photo they should not occur. Temporal changes may also contribute to such error. For example the RS1 which was recorded next to the reservoir may have grown since the photos were acquired. It is also obvious on the ground photograph that the RS1 does not meet the minimum mapping size standards.

A final type of error is when different vegetation appears the same on the photos. The interpreter was surprised at the lack of error or confusion between aspen (RT1) and cottonwood (RT2). It was initially thought that since these classes have a very similar spectral signature, that they would be very confused, but there were only 3 errors in this category. It could be that when the transects were laid out, they skipped the elevational transition zone from one species to the next. There was a great deal of confusion in the RH2 and RH3 classes. RH2 was interpreted as RH3 eighteen times and RH3 was interpreted as RH2 eight times. It may be desirable to combine these for future work on dynamic changing land such as the Wet Mountain Valley, especially when there is a long lapse between photo acquisition and ground verification. If these classes are not combined, the RH3 should be only mesic grasses with no sedge or rush component. In both cases the appearance on the photo between the confused classes were so much alike as to be indistinguishable. The RH1 class had no commission error and seven omission errors. It was confused with RT2 three times, RH2 three times, RH3 four times, and UG once.

Riparian zones were correctly classified 96.3% of the time. There were no omission errors, but twelve times the interpreter said it was non-riparian and the field investigation found it was riparian. This is an extreme error and is probably due to temporal change, or spatial misplacement.

The overall percent correct classification of 74.0% for vegetation communities is 9-12% lower than the interpreter anticipated. The discrepancy in anticipated and actual PCC can probably be attributed to time lapse of photo acquisition and field data collection and too many field personnel. The 96.3% for riparian versus non-riparian is about what the interpreter expected, and what a literature search of similar projects would predict. There were no significant trends in the error that can be globally adjusted unless the user would like to combine RH2 with RH3 which increases the accuracy to 81.7%, or combine all the RH classes to achieve an 83.8% accuracy. Finally, by combining all similar types into RT, RH, RS, U, OW and NV, the accuracy increases to 85.3%.

The primary objective of this project, to inventory the riparian ecosystems within Custer County, Colorado, has been successfully met. The LFC and NAPP color infrared aerial photographs, when used with the current classification system, proved to be an excellent data source for mapping the riparian areas of Custer County.

Photo interpreter

A single photo interpreter performed the delineation, identification, and labeling work on the project photographs. This approach was employed to ensure the consistency of the interpretation work, which is a highly desirable result in any project of this type. The same level of consistency is desirable in field data collection; however, the field crews that collected evaluation data for this project consisted of 13 people on five different teams. The crew members had a diversified background and varying levels of experience. There were only six hours of training and the consistency of field evaluation may be questionable. It may be advantageous for future projects to have one or two field crews collect the evaluation data.

GIS

The riparian areas mapped through this project have been digitally processed in an Arc/Info GIS database by the Colorado Division of Wildlife and will be shared with the San Isabel Foundation, U.S. Forest Service, and Bureau of Land Management. As this photo-derived riparian information is used, changes or errors that are observed in the field should be noted and updated in the GIS database.

There are two recommendations the interpreter would like to make based upon the above results and conclusions. They are:

The authors wish to thank the Great Outdoors Colorado Board, Sangre de Cristo Habitat Partnership Program Committee, and Bureau of Land Management for their financial contributions to this project. We also wish to thank Dave Winters and Mark Ropers, U.S. Forest Service for allowing the use of aerial photographs, ortho-photographic mylars, ortho-topographic mylars, and riparian habitat maps which greatly reduced the cost of this project and increased the utility of the final product. We also wish to thank the many landowners in Custer County who allowed us access to their lands when conducting the accuracy assessment field work. Finally, we wish to specifically thank the following individuals for their many contributions to the success of this project: Don Mercill, Sherri Patterson, John & Carol Barnett, Derek Brixsey, Eric Wormsley, Chris Kloster, Peggy Kavookjian, Shannon Koch, & Seth McClean.

TABLE 5 OVERALL ERROR MATRIX
Ground Check Points	Photo interpretation Points																		TOTAL CORRECT	OMISSION ERROR	PERCENT OMISSION
	RT	RT1	RT2	RE	UT	RS	RS1	US	RH	RH1	RH2	RH3	UG	OW1	OW2	OW3	NV	X
RT	0																		0	0	0
RT1		26	1										1					3	26	5	16
RT2		2	18			1	1					1							18	5	22
RE				9															9	0	0
UT					0														0	0	0
RS						4							1						4	1	20
RS1							26				4	1	1	1		1		1	26	9	26
US								6			1	1							6	2	25
RH									0										0	0	0
RH1										9	3	4	1						9	8	47
RH2							2				36	8	2				1	3	36	16	31
RH3		1	1				1	1			17	60	1				2	3	60	27	23
UG		1				1		2			2	4	23					2	23	12	34
OW1														10					10	0	0
OW2															5				5	0	0
OW3																1			1	0	0
NV																	3		3	0	0
X																		6	6	0	0
TOTAL CORRECT	0	26	18	9	0	4	26	6	0	9	36	60	23	10	5	1	3	6	242
Commission Error	0	4	2	0	0	2	4	3	0	0	27	19	7	1	0	1	3	12
Percent Commission	0	13	10	0	0	33	17	33	0	0	43	24	23	9	0	50	50	67
Overall Percent Correct Classification (PCC) = 242 correct/327 total = 74.0% Combine RH2 & RH3 - PCC = 81.7% Combine RH1, RH2, & RH3 - PCC = 83.8% Combine all major classes (RT, RS, RH, OW, and NV) - PCC = 85.3% Percent Correct Classification of Riparian/Non-Riparian = 315 correct/327 total = 96.3%

Figure 2 - Accuracy Assessment Data Sheet for Transect 57a

Field Transect #: 57a Field Crew: CU5

No. of sample points on transect: 3 Date: 8/12/97

Transect azimuth: N to S

Site Description
 

	Tree (RT)			Shrub (RS)			Herbaceous(RS)
Point #	Type	* Cover	# Size	Type	* Cover	# Size	Type	* Cover	# Size
1							RH2	D
2	RT	< A	< A	RS1	C		RH2	A
3							RH2	D

Location:                                                        Photo Information:                      Quad Name:
Township & Range: T23S-R71W                                Camera #: CU5-1                                            ROSITA
Section/Quarter #: 14                                                    Exposure #: 8
UTM-X: 47400                                                               Direction: SOUTH
UTM-Y: 4210850                                                           Aerial Photo #: ?

*            %Group Cover:                                #               Size:
A         0% - 25%                                                 A             Sapling (emergent º 8' & 3"dbh)
B         26% - 50%                                               B             Pole (8' º 20' & 8"dbh)
C         51% - 75%                                               C            Mature (> 20' & > 8"dbh)
D         76% - 100%

Remarks/Comments:
Pt #2: RT is either small cottonwood or small aspen that are sapling height. Probably weren=t present 8 years ago.