1cross-sections taken at a location where the channel flowed on either side of an island

All cross-sections, except 49a, indicated a C type stream. Cross-sections 49a and 49b were taken at a location where the stream flowed on either side of an island. The entrenchment ratio of 49a was lower than for the other cross-sections, and at this point, one side of the channel (49a) was classified as an F type. Sinuosity was calculated from the aerial photography for the whole study area, while the slope was determined in the field by measuring a longitudinal profile over a distance of 637 meters.

The field measurements were taken in October during low flow. It became apparent that in some short sections of the stream the channel was braided, with 3 to 4 different small channels. There was a large amount of mid-channel deposition of gravel and cobbles with the channels flowing at different elevations. This caused the water to flow perpendicular to the main channel in some places, moving from the higher to the lower elevation. Although no measurements were taken in those areas, it was expected that these sections would have been classified as a D type channel, indicating braiding.

Rosgen (1996) described the progressive stages of a stream channel adjusting to changes in driving variables. These may include changes in sediment and flow regime, which in this case may be related to the large landslide that occurred in the watershed in the mid-70s. A C 4-type stream often progresses to a C 4 (bar 6), and then to a D 4 type. The difference between C 4 and C 3 is the channel material: cobbles for C 3, and gravel for C 4 type stream channels.

Field observations at the cross-sections 49, 35 and 34 (Table 3) and comparison with older aerial photos showed increased bar depositions and lowered sediment transport capability, resulting in larger islands and mid-channel bars in 1998 compared to 1979. It is possible that a progression from a C to a D type channel, as described in Rosgen (1996), is occurring in this portion of the stream. This section is located below the straight stream section, and it is assumed that the increased sediment load resulting from the mid-70s landslide was at least partly responsible for this aggradation.

In the future, we plan to develop a detailed 3-dimensional model of the stream channel, using a laser range finder and directional compass. These measurements will be repeated over the years to measure and possibly predict channel changes at a large scale.

Bankfull measurements conducted in the field showed that bankfull width had been underestimated somewhat from the aerial photos (Table 4). Due to time constraints, these field measurements were conducted after the GIS work had been completed, and no adjustments to the channel outline were done. Even in the field, it is not easy to determine bankfull, and it is common to underestimate bankfull width (Rosgen 1996). Two different

observers may come up with slightly different locations for bankfull. It is even more difficult to estimate this location from an aerial photo. However, bankfull is a consistent morphological index for comparing the stream in 2 different years since it is the

location of the flow with a recurrence interval of 1.5 years (Dunne and Leopold 1978). Ideally, the field measurements should have been done before digitizing the stream.

Changes in water level could potentially change the placement of the stream channel while digitizing it. To get an idea of how accurate the digitization was done (in other words, unaffected by fluctuations in stream flow), the relationship between discharge and stream surface area was determined. If the channel had been outlined too close to the wetted width, a change in discharge would have resulted in a large change in surface area. If, however, the channel was digitized close to bankfull, then a change in discharge should not have changed the surface area very much. Table 5 shows the relationship between discharge and surface area.

Since the stream gauging station was discontinued in 1996, the discharge for 1998 was estimated by averaging all stream flow values for August 3 from 1978 to 1996. The resulting value (1.27 m³ sec^-1) closely resembled discharge measurements for 14 days on either side from the date the aerial photo was taken in 1998. In 1982, Catherine Creek had an above average discharge. While the average mean flow was 3.32 m³ sec^-1 (1979-1995), the peak flow for that year was recorded at 18.29 m³ sec^-1 (on 25 May 1982). This means that the aerial photo of 1982 was taken close to peak flow, which would give a reference point for bankfull width. The numbers in Table 5 show that a 3-fold increase in flow from 1979 to 1982 resulted in only a 6.7 % increase in surface area, while a 4.4 fold decrease in discharge from 1979 to 1998 resulted in a 2.8 % decrease in surface area. This small change in surface area caused by a large change in discharge means that the stream was able to absorb that increase by becoming deeper, and not much wider. Only a small amount of the width change would have occurred due to changes in discharge and water level. Therefore, observed changes in stream width were attributed to actual change in the stream channel as opposed to changes due to higher or lower discharge.

The results of the change analysis for the stream and islands are shown in Table 6. This cross-tabulation matrix shows the areas of change and no change occurring from 1979 to 1998. It can be seen that the 1979 stream area was 5.07 ha, compared to 4.93 ha in 1998, representing a 2.76 % decrease in stream area. Areas of change include 1.74 ha (1979 land to 1998 stream) and 1.88 ha (1979 stream to 1998 land), for a total of 3.62 ha of change. The area of 1979 stream that remained stream in 1998 is 3.19 ha (no change). This shows that the area of change is nearly the same as that of the area of no change. The size of the area of land in 1979 that remained land in 1998 (37.82 ha) is large, since the whole study area is included. Although this represents an area of no change, our interest lies in the change detection within the stream channel boundaries.

The cross-tabulation image (Figure 3) shows areas of change colored blue (stream to land) and green (land to stream), while the stream channel area experiencing no change (stream to stream) is colored red. The processes of erosion and deposition can be seen clearly from this image; the stream erodes bank material in 1 area and deposits it downstream. The green area in the center of the study area shows how the outside of the stream meander is pushed outward, increasing the stream’s sinuosity. In the lower right of the image in the upstream portion of the stream, large changes are visible. This is an area of the stream with several larger islands that have undergone changes, or where the stream channel changed from one side of an island to another.

Changes in the islands were also analyzed with cross-classification (Table 7). The interest was in determining how much of the island area of 1979 had remained the same within the stream channel, and how much had changed. Areas of change included areas of stream and land outside the stream channel moving to island and vice versa (islands moving to stream or land). Due to the nature of the cross-classification, areas being compared were island areas in 1979 and 1998, but the stream channel was not included. Hence, the area of ‘not island’ was left blank, since it consisted of the stream channel and the entire background of the image.

In this case, island area increased from 1.16 ha in 1979 to 2 ha in 1998, an increase of 72 %. The area of no change (0.45 ha) is very small compared to the areas of change (2.26 ha). The cross-tabulation image (Fig. 4) shows areas of change in red and areas of no change in blue. It becomes obvious how much change occurred in island area. The nature of this change is further explained by the summary statistics derived from GIS layers and shown in Table 8.

While the total island area increased from 1979 to 1998 from 1.16 ha to 2 ha, the number of islands decreased from 50 to 23. Mean island perimeter increased from 62 m to 124 m, while minimum perimeter remained the same. Similar change was observed for mean island area, which increased dramatically (from 232 m² to 868 m²), while minimum island area remained relatively the same.

These numbers indicate that although fewer islands occurred in 1998, they were much larger than they had been in 1979. This is obvious from studying the images, which also show more vegetation occurring on the islands. Observations of the images showed that what was a small island in 1979 had grown to a much larger, densely vegetated island in 1998. Figure 5 shows one such area. It appears that the small island began to capture sediments carried by the stream, and the island gradually grew in size by increasing this sediment capture, coupled with vegetation that was able to take hold on the island. Aerial photos from the years between 1979 and 1998 confirmed the gradual growth of this island.

Fewer, larger and more vegetated islands suggest an increase in stability occurring over the 19 years. The original increase in deposition may be related to the mid-70s landslide that resulted in large amounts of sediment being deposited over the years.

The length of right and left stream bank changed very little from 1979 to 1998, increasing 1.75 % from 2500 m to 2544 m for the left bank, and decreasing 1.08 % from 2400 m to 2375 m for the right bank (Table 8). Likewise, the stream perimeter remained almost the same (decrease of 0.16 %). The larger change in the wetted area (-24.88 %) can be explained by the changes observed in the islands. Although the stream area did not change much, the increase in island area within the channel reduced the size of the area actually covered by water.

In some areas, lateral movement of the channel was as much as 20 m, independent of island area change (Fig 5). In this particular location the stream eroded the outside of the meander, cutting into the bank and depositing the material downstream. We observed deposition at downstream meanders and islands.

Another parameter of interest was length of thalweg and change in sinuosity, which is the ratio of stream length to valley length. Thalweg length was measured for both small and large-scale images; this allowed for a comparison of thalweg length and sinuosity over a period of 61 years. Figures 6 and 7 show how these parameters changed over time.

The graph illustrates variability in thalweg length, with the highest value occurring in 1998; however, the increase from the lowest to highest value is relatively small: an increase of 6.3 % from 2287 m (1960) to 2430 m (1998). Likewise, an increase in sinuosity from 1.099 to 1.168 is not large. Rosgen (1996) uses sinuosity as one of his level II inventory criteria for determining stream type. Sinuosity carries the least weight of these criteria. However, a general guideline for high sinuosity in a C type stream is >1.2; very high sinuosity in an E type stream is >1.5. In those calculations, sinuosity ratios can vary by +/- 0.2 units. Rosgen’s (1996) work implies that the sinuosity change we observed is within normal range for C type streams.

The location of woody debris was digitized from the 1979 and 1998 images, but due to increased vegetation near the stream bank in the 1998 images, it was difficult to see the portion of the woody debris covered by bankside shrubs. It was concluded that although woody debris larger than 20 cm in diameter was clearly visible and could be digitized, the concealment by overhanging vegetation prevented an accurate measurement of its size. Therefore, woody debris size was not analyzed.

Stream width decreased in all exclosed and all grazed areas from 1979 to 1998 (Fig. 8). Values for stream widths are displayed in Table 9. Mean stream width in exclosed (E) sites decreased 36 % from 18.67 m to 11.85 m, compared to a 22 % width decrease in grazed areas (G) from 16.62 m to 12.96 m.

This decrease in stream width had to be studied in conjunction with the change in island area. Since the stream width was calculated by excluding all islands, thereby measuring only wetted width, the decrease in stream width was related to the change in island area. In fact, when one looks at the overlay of island and stream outlines of both years, the large influence of the islands on stream width measurement becomes obvious.

However, it is not easy to discern a pattern at a larger scale, such as for grazed or exclosed sites. In one area (E3, Fig. 9), a relatively large width decrease occurred concurrent with a large change in island area, while in E4 (Fig. 10), a small change in width and a small change in island area occurred together.

On the other hand, the site of smallest change in stream width (G3) appears to have areas of little change, where the stream has remained straight, as well as areas of large increase in island area. The site G4 shows large changes in island area; however, these changes did not affect stream width to a high degree since the channel split on either side of the island. In the area of largest stream width decrease (E5), there was also a large decrease in island area, causing the stream to become narrower. In E1, the stream tried to erode its banks near the road and was straightened by the Highway Department in 1990. This would account for the stream width decrease in that area.

These results show that over the whole length of the stream, one could observe an increase in island area and a narrowing of the channel, but locally (such as in E5), a decrease in island area was observed concurrently with narrowing of the channel in that section. This demonstrates the importance of scale when studying time change in a stream system. What occurs in one small area does not reflect the response for 2.5 km of study area. The response of one exclosure did not reflect the response of all others. Since islands straddle fence lines in exclosed and grazed areas, the interaction of island change and stream width is difficult to interpret within these boundaries, and the assessment of the whole study area seems more appropriate for island change and its effect on stream width.

The stream width was highly affected by island morphology: if a stream section had a large island in the channel, with water flowing on either side, the stream width was narrower than it would have been without an island in that location. There was a large difference between a section of stream with one channel and another section where it flowed on either side of an island. Stream width as well as width to depth ratio are affected by these changes in stream morphology. When a stream becomes narrower, it also deepens. The narrowing of Catherine Creek, coupled with increased vegetation on islands and stream bank, demonstrates a trend towards increased stability for this stream.

We did not find large differences between grazed and exclosed areas. Results of others comparing grazed and exclosed sites vary widely, from no differences to significant differences. Magilligan and McDowell (1997) studied 4 eastern Oregon streams for their responses to the removal of cattle grazing. In all 4 areas, cattle grazing had not occurred for at least 10 years. The authors found significant changes between grazed and ungrazed reaches, such as increased pool areas and channel narrowing in the exclosures. These changes were mostly attributed to increased vegetation and, therefore, increased channel roughness following removal of cattle grazing. However, the authors also determined that local effects, such as considerable flow reduction in the summer and particle size, influenced the magnitude of channel adjustments.

Other authors reported no differences between exclosures and grazed sites. Medina and Martin (1988) observed channel width and depth increases in both exclosed and grazed areas in their study in southwestern New Mexico. The changes were attributed to a previous wildfire and hydrological processes of stream equilibrium. Kondolf (1993) measured cross-sections in exclosed and grazed areas in California and observed no significant differences between the 2 treatments. He concluded that this could possibly be attributed to a lag in adjustment of the stream channel after livestock had been excluded for 24 years from the site.

At Catherine Creek, it appears that the topography, stream dynamics and the road have a larger impact on stream morphology than the grazing regime. The influence of local topography on channel morphology was also observed by Clifton (1989) in a central Oregon study relating vegetation and land use to channel morphology over a 50-year period. The author determined that temporal variability was related to exclusion of grazing, while spatial variability (between different stream reaches) was due to the prevailing riparian vegetation, input of large organic debris and local physiography. Specifically, channel width, wetted perimeter and channel shape were mostly correlated with local vegetation variability (Clifton 1989).

In streams bordered by trees, the input of large woody debris plays a considerable role in channel change. Large logs in the stream divert the flow of water and become sources of deposition and pools; bank erosion may be reduced or enhanced locally, and stream energy is dissipated (Keller and Swanson 1979). We observed similar channel changes in aerial photography from 1990, although those images had not been scanned and digitally analyzed at the same fine resolution of this study. A large Ponderosa pine tree fell in the stream in 1990, and in the photos of the following years, the diversion of water flow could be observed, leading to a knickpoint that moved downstream over the following years.

The impact of the road also has to be taken into consideration. In the 1937 aerial photo, the road is visible in the same location as it is today. Visits to the local museum in the town of Union confirmed that the hot springs upstream from the study area were a popular destination at the turn of the century. It is, therefore, assumed that the road has been in the same location for a long time. There is also evidence that the stream was moved to one side of the valley, possibly to accommodate the road. Old meanders visible on the aerial photography cross the road in several places. Large, old cottonwoods still remain near these meanders and are additional indicators of old stream channels. The remnants of these old channels were digitized from the images and overlaid on the map of the study area (Fig. 11). Before Catherine creek enters exclosure E 5, at the southeast corner of the study area, the stream encounters the road and makes a sharp turn from flowing west to north-northwest. Rip-rap has been placed there by the highway department to reduce erosion. If the road were not present, the stream would continue on its westerly course, flowing in the old channel visible from the aerial photo.

For these reasons, it has to be assumed that the stream was moved over to accommodate the road and actions were taken over the years to reduce erosion at the point where road and stream met at a sharp angle. Further downstream, the road prevented the stream from flowing through its old channel, and the stream had to adjust to the lower sinuosity. This increased bank erosion, bar deposition and channel aggradation. The channel changes observed in portions of the stream may be the result of the stream’s attempt to deal with these changes.

Conclusions

Over the 19-year time period from 1979 to 1998, a variety of changes in stream morphology have taken place at the Catherine Creek study area. Some measured parameters changed to a large degree, others did not, and the observations can be categorized into those of small or large change. Parameters of small change included: length of thalweg, left and right streambank, sinuosity, stream perimeter and stream area (including islands). Parameters of larger change were: island area, island number, stream width and wetted area (excluding islands).

Although the stream area itself changed little, the results from the cross-classification showed that the area of change was almost equal to the area of no change, in terms of water changing to land, vice versa or remaining the same. It appears that at a small scale (i.e. the whole study area) the stream remains largely where it has been since 1979. On a larger scale, however, localized changes become visible. These include lateral movement of the stream bank or the increase in amplitude of some meanders, although this did not result in increased sinuosity. The topography of the area and the road prevent a large increase in sinuosity. Any large change observed seemed to be connected with the islands: their area increase or decrease affected the wetted area and, therefore, the stream width.

Several signs of increased stability have become apparent in this area over the time period studied. The decrease in island number and increase in island area, coupled with increased vegetation on the islands, demonstrates this trend. Narrowing of a stream is usually associated with greater stream depth and better fish habitat, although the lack of stream depth data for 1979 prevents us from making any detailed conclusions about changing stream depth.

Stream width decreased in grazed and exclosed area, but was smaller in grazed sites. However, the reason for width decrease was tied to the island changes, and it was observed that both an island area increase and decrease could result in a narrower stream.

The lack of distinct responses for grazed or exclosed areas suggests that the topography, stream dynamics (erosion, deposition and especially island formation) and the road have a larger impact on stream morphology than the grazing regime in this particular system.

 Catherine Creek is confined by steep hills on one side and a road on the other. This is especially true in the upper 1/3 of the study area. In the lower portion (E1), the stream tried to erode its banks near the road and was straightened by the Highway Department. Rip-rap was added at the upper end near E 5. If allowed to run its course without the road, the stream would have had more meanders and increased sinuosity. Today, areas for sinuosity increase or change in meanders appear to be limited due to topography and the road.

The use of remote sensing, GIS/GPS and ground truthing provided a definite advantage over only ground data collection. Many of the statistics are quickly extracted from the GIS database for analysis. Other parameters would be difficult or nearly impossible to attain without these techniques. For example, a cross-classification is valuable in time change analysis since it yields areas that have changed from land to water and vice versa. Without the GIS, it would have been time consuming to produce. The ability to overlay many features and measure distances on-screen is helpful in analyzing change. The visual aspect of GIS allows the user to perceive changes where they may not be obvious otherwise. Not many time change analysis studies have been done at such a large scale (1:4000), but this project demonstrates that it can be done successfully.

Anderson, G.L. 1996. The application of spatial technologies for rangeland research and management: state of the art. Geocarto 11(3):5-11.

Clifton, C. 1989. Effects of vegetation and landuse on channel morphology. In:Greswell, R.E., B.B. Barton and J.L. Kershner (ed.) Practical approaches to riparian resource management, an educational workshop. May 8-11, Billings, Mont. US BLM. 121- 29.

Dunne, T. and L.B. Leopold. 1978. Water in Environmental Planning. W.H. Freeman and Co., San Francisco, Calif. 818 pp.

Eastman, J.R. and J.E. McKendry. 1991. Change and time series analysis. Explorations in Geographic Information Systems Technology, Vol. 1. United Nations Institute for Training and Research, Geneva, Switzerland. 78 pp.

ERDAS^®, Inc. 1997. ERDAS^® Field Guide^TM, Version 8.3.1, Atlanta, Ga. 655 pp.

ESRI (Environmental Systems Research Institute). 1996. Getting to know ArcView. Environmental Systems Research Institute, Redlands, Calif.

Everitt, J.H. and C.J. Deloach. 1990. Remote sensing of Chinese tamarisk (Tamarix chinensis) and associated vegetation. Weed Sci. 38:273-278.

Everitt, J.H., D.E. Escobar, R. Villareal, M.A. Alaniz and M.R. Davis. 1993. Integration of airborne video, global positioning system and geographic information system technologies for detecting and mapping two woody legumes on rangelands. Weed Tech. 7:981-987.

Ferguson, R.L., L.L. Wood and D.B. Graham. 1993. Monitoring spatial change in seagrass habitat with aerial photography. Photogramm. Eng. Rem. Sens. 59(6):1033- 1038.

Golden, M., J. Johnson, V. Landrum, J. Powell, V. Varner and T. Wirth. 1996. Guidelines for the use of digital imagery for vegetation mapping. EM-7140-25. USDA Forest Service. Engineering Staff. Washington, D.C. 125 pp.

Hinckley, T.K. and J.W. Walker. 1993. Obtaining and using low altitude large scale imagery. Photogramm. Eng. Rem. Sens. 59(3):310-318.

Johnson, D.E, N.R. Harris, S. du Plessis and T.M. Tibbs. 1995. Mapping and analysis of Catherine Creek using remote sensing and geographic information systems (GIS). In: Eastern Oregon Agricultural Research Center Field Day Annual Report, Special Report 951, July 1995:10-26.

Keller, E.A. and F.J. Swanson. 1979. Effects of large organic material on channel form and fluvial processes. Earth Surface Processes 4:361-380.

Kondolf, G.M. 1993. Lag in stream channel adjustment to livestock exclosure, White Mountains, California. Restor. Ecol. 1:226-230.

Laliberte, A.S. 2000. The use of remote sensing and Geographic Information Systems(GIS) in assessing changes in stream morphology and vegetation. Masters thesis.Oregon State University, Corvallis, Ore. 158 pp.

Magilligan, F.J. and P.F. McDowell. 1997. Stream channel adjustments following elimination of cattle grazing. J. Amer. Wat. Res. Assoc. 33(4):867-878.

Medina, A.L. and S.C. Martin. 1988. Stream channel and vegetation changes in sections of McKnight Creek, New Mexico. Gr. Bas. Natur. 48(3):373-381.

Miller, J.R., T.T. Schulz, N.T. Hobbs, K.R. Wilson, D.L. Schrupp and W.L. Baker. 1995. Changes in the landscape structure of a southeastern Wyoming zone following shifts in stream dynamics. Biol. Conserv. 72:371-379.

Rosgen, D.L. 1996. Applied River Morphology. Wildland Hydrology, Pagosa Springs, Colo. 8-43.

Roth, N.E., J.D. Allan, and D.L. Erickson. 1996. Landscape influences on stream biotic integrity assessed at multiple spatial scales. Landscape Ecology 11(3):141- 156.

Strong, L.L., R.W. Dana and L.H. Carpenter. 1985. Estimating phytomass of sagebrush habitat types from microdensitometer data. Photogramm. Eng. Rem. Sens. 51:467-474.

Trimble Navigation. 1996. Trimble Pathfinder Office^TM, Software Reference Guide. Trimble Navigation Ltd., Sunnyvale, Calif.

Tueller, P.T. 1996. Near-earth monitoring of range condition and trend. Geocarto Int. 11(3):53-62.

Tueller, P.T., P.C. Lent, R.D. Stager, E. A. Jacobson and K.A. Platou. 1988. Rangeland vegetation changes measured from helicopter borne 35 mm aerial photography. Photogramm. Eng. Rem. Sens. 54(5):609-614.

United States Department of Agriculture Forest Service Global Positioning Systems page. 1999. Online. http://www.fs.fed.us/database/gps/ (Accessed Sept.12, 1999).

United States Geological Survey page. 1999. Online. National Water Information System (NWIS) data retrieval. Water resources data for Oregon. http://waterdata.usgs.gov/nwis-w/OR/index.cgi?statnum=13320000 (Accessed Mar.6, 1999).

Warner, W.S., R.W. Graham and R.E. Read. 1996. Small Format Aerial Photography. American Society for Photogrammetry and Remote Sensing. Whittles Publishing, Scotland, UK. 348 pp.

Warren, P. L. and C. Dunford. 1986. Sampling semiarid vegetation with large-scale aerial photography. ITC Journal 4:273-279