Caty F. Clifton,¹ Robin M. Harris,¹ and James K. Fitzgerald²

ABSTRACT: The northern Blue Mountains of southeastern Washington and northeastern Oregon experienced a series of unusual storm and flood events during the winters of 1995-96 and 1996-97. Flood magnitude and frequency, estimated by the indirect slope-area method, varied between watersheds from 25-year to 100-year events. Flood frequency was related to watershed orientation with west-facing watersheds experiencing 100-year events, and east-facing watersheds experiencing 25-year to 50-year events. Post-flood investigations have been ongoing since 1996, including assessments of mass wasting features, surveys of stream channel measurement sites, and inventories of roads and instream habitat structures. Channel adjustments in cross section area, volume of stored sediment, and particle size distributions appear to be more related to reach-level controls such as large wood jams and local mass wasting sources than to overall flood magnitude. Results from this monitoring are being used in flood recovery efforts and land management planning.

KEY TERMS: Floods; channel morphology; watershed hydrology.

INTRODUCTION

Major floods present an opportunity to observe the way in which mountain watersheds respond to unusual weather and streamflow conditions, and how land uses are affected by and contribute to flood impacts. Accelerated upland erosion, rapid delivery of sediment and wood into streams, and high rates of sediment transport alter watershed and stream form and may have serious consequences for aquatic habitat and downstream communities. Flood assessments improve our predictive capability and damage assessment techniques, and contribute to the understanding of watershed response and recovery processes. Mountain landscapes are defined by complex climate, topography, and stream channels where floods play an important role influencing water quality and aquatic habitat, which are ongoing resource management issues in the Pacific Northwest.

Above-average autumn rainfall, and an early winter storm caused localized flooding in the foothills of the northern Blue Mountains in November of 1995 (Figure 1). Late development of the winter snowpack coupled with low temperatures led to frozen soil

conditions. Storms in early February 1996, brought the snow level down in elevation with accumulations of 6 to 12 inches in the interior valleys. A series of warm-moist subtropical surges delivered warm rain during the week of February 5th, rapidly melting snow over frozen soils below 4000 feet. Rivers and streams reached flood stage by the end of the week, affecting many communities downstream of the National Forest. The Umatilla River above Meacham gage peaked at a record 9.2’ flood stage and 6200 cubic feet per second flow. A third snowmelt-generated flood occurred in April (Figure 1).

Figure 1. Floods on the Umatilla River above Meacham,
November 1995 - April 1996

(U. S. Geological Survey gage #14020000)

Above-average precipitation occurred again in the fall of 1996. The snowpack developed gradually through December. During Christmas week an arctic cold system moved south into Washington and Oregon producing snowfalls in the interior valleys. In late December, a warm Pacific storm moved inland overriding the cold system. By New Year’s eve, most precipitation was falling as rain and the warm system had scoured out cold air remaining in the valleys. Snowmelt occurred rapidly with strong rises in streamflows. Walla Walla, Washington reported record temperatures of 67° F. The Umatilla River above Meacham peaked again on January 1, 1997 at 8.5’ and 5120 cfs.

These flood events triggered shallow, rapid landslides and debris flows between about 2000 and 4000 feet, mobilizing sediment and debris into the channel network. High stream flows uprooted riparian vegetation, scoured channel banks, caused lateral channel migration, and moved large volumes of sediment and debris. Damage to roads, trails, and other investments exceeded 2 million dollars on the National Forest. Six counties in SE Washington and NE Oregon were declared federal disaster areas after the 1996 floods.

This report presents results from our ongoing efforts to understand flood effects and watershed response in the northern Blue Mountains of southeastern Washington and northeastern Oregon. The objectives of our flood investigation were to: estimate flood magnitude, frequency, and geographic distribution; evaluate mass erosion features and stream channel response to flooding; and, address land management implications.

METHODS

Field Surveys

We identified mass wasting features using aerial and ground reconnaissance. Features were mapped and described following methods described by Costa (1984). We assessed flood impacts on stream channels using methods described by Harrelson et al (1994). Channel measurement sites were established at representative stream reaches. At each site we surveyed two or three channel cross sections spanning the entire flooded area, measured the longitudinal profile, collected pebble counts, and took discharge measurements and representative photographs. Permanent benchmarks and stakes were installed for resurvey. Twenty-nine sites were initially established after the 1996 floods for the purpose of estimating flood magnitude. Sixteen sites were resurveyed over the next two years for analysis of stream channel change.

Road-stream crossing inventories followed methods developed by the Six Rivers National Forest (Furniss et al, 1998). Assessment of instream aquatic habitat structures followed methods developed by the Pacific Northwest Region (U. S. Forest Service).

Estimating Flood Magnitude and Frequency

We estimated flood discharge using the indirect slope-area method (Williams and Costa, 1988). Flood stage was identified from field indicators which included sediment or debris deposits and scour lines. Flood discharge (Q) was approximated using cross-section area (A), hydraulic radius (R), slope (S), and Manning’s equation,

Q = 1.49 (AR^2/3S^1/2), where n = roughness coefficient

Flood frequency, or recurrence interval, was determined by comparing field-estimated flood discharge with flood discharge data recorded at nearby gages, and published values for flood frequencies (Harris and Hubbard, 1983). Data from field sites were then grouped by watershed and landscape position (windward and leeward side of the mountains) and analyzed to determine relationships between discharge, flood frequency, watershed area, and landscape position.

Stream Channel Response

We analyzed post-flood channel response by using the 1996 initial post-flood survey as a baseline and comparing following year (1997 and 1998) resurvey data to the baseline year. Since 1997 was also a flood year we expected to see channel changes from 1996 as a result. Indices of channel response include change in channel cross section area and pebble distributions (Olson-Rutz and Marlow, 1992, and Harrelson et al, 1994). Change in cross section area at a site was determined by averaging cross sections and comparing flood area in 1996 to the following year’s survey. Multiple cross sections allowed an approximation of reach-level response. Change in the volume of sediment stored at a site was estimated by the average area change over the length of the reach surveyed. Sources or controls on channel response were identified during field surveys. Sites that show a net reduction in cross section area and increase in stored channel material are considered to have aggraded, and sites that show a net increase in cross section area and decrease in stored sediment show evidence of degradation.

RESULTS

Compared to published magnitude-frequency data, the flood at the South Fork Walla Walla gage appeared to be on the order of at least a 100-year, or Q_.01 event (Table 1). Flood frequency at ungaged National Forest streams varied from 100-year events in the Umatilla and Walla Walla watersheds, to less than 25-year events in the Tucannon and Wenaha watersheds.

Table 1. Comparison of Slope-Area,
gaged, and USGS 100-year flood discharge on
the South Fork of the Walla Walla River.

Overall, flood magnitude increased with increasing drainage area, however, windward watersheds (Umatilla and Walla Walla) had significantly higher flood discharge (P=0.001) for the same drainage area compared to leeward watersheds (Tucannon and Wenaha) (Figure 2). West-facing watersheds appear to have sustained higher flood discharge in part because of watershed orientation to winter storms, which generally track from southwest to northeast.

Figure 2. Estimated Flood Discharge vs Drainage Area for field sites in the
Umatilla, Walla Walla, Tucannon, and Wenaha watersheds.

Stream Channel Change

Post-flooding channel adjustments were anticipated because of increases in sediment sources, influx of large woody debris, and direct effects of high flows on channels and floodplains. Channel adjustments within two years of the initial baseline survey year were determined using channel area comparisons and estimated change in the volume of stored sediment (Table 2). Percent change in channel area ranged from less than 5 percent (low to no change) at 11 sites, to a 5 to 7 percent change at the remaining 5 sites. Change was most apparent at SF Umatilla-3 and NF Touchet-1, where lateral channel migration occurred in association with debris jams entering or leaving the site (Figure 2).

Table 2. Channel Changes 1996-1998: Change in Cross Section Area
and Estimate of Sediment Aggraded or Degraded.

Changes in channel substrate were also observed in three years of post-flood monitoring (Table 3). The median particle diameter (d₅₀) decreased in 1997 followed by an increase in 1998. The level of fine material (% < 6 mm) increased in 1997, decreasing in 1998. Changes in percent fines were statistically significant (P<=0.05) from 1996 to 1997 at 3 sites. Both measures of streambed material size indicate an initial increase of fines into the channel, which is attributed to the influx of hillslope and channel-derived sediment, followed by a flushing of fines as sources were depleted by subsequent flows.

Figure 3. North Fork Touchet River (NF Touchet-1) channel cross-section changes 1996 - 1997.

Management Implications

Results from road-stream crossing and aquatic habitat structure inventories are being used in flood repair and maintenance activities (Fitzgerald and Clifton, 1998). Much of the damage to roads occurred at road-stream crossings where culverts typically failed because of plugging by sediment and debris. In general, culvert sizing is based on a design flow, not on passage of sediment and debris. As a result, new design criteria are now used at high risk sites on the National Forest. These include installing a "sag" at the crossing (lowering the fill over the culvert) to allow for plugging and over-topping, and preventing diversion onto the roadway. Instream structures, typically log-rock weirs designed for late summer pool habitat, were found to have a high rate of retention during and after flood events. Inventory findings have been used to identify and prioritize maintenance and repairs.

DISCUSSION

Our analysis of flood magnitude and frequency in the northern Blue Mountains shows the influence of watershed characteristics on flooding. Watershed position relative to prevailing storms contributed to overall flood magnitude and return period with windward watersheds experiencing 100-year floods compared to 25-year and 50-year events in leeward watersheds. Changes in channel morphology and sediment size appear to be more related to local or reach-level controls than to overall flood magnitude. Factors influencing post-flood channel adjustment include sediment source, large wood, and valley widths. Streams are adjusting to increased sediment and debris through complex interactions between hillslopes, floodplains, and channels.

Table 3. Post-flood changes in median particle size and percent fines.

All resurveyed channel sites showed evidence of post-1996 flood response. Change in sediment storage (aggradation and degradation) and volume estimates are indicative of the general magnitude and direction of channel change. For example, the SF Umatilla-3 site lost approximately 243 yd³/100 ft of channel because of lateral channel migration caused by flow deflection around an upstream log jam. Particle size distributions showed significant increases in fines at 3 sites from 1996 to 1997 indicating a flushing of fines through the channel system. The channel area index did not appear to be a highly sensitive indicator of change in these large, relatively complex cross-sections in part due to inherent variability, problems with replicating cross-sections, and computational problems associated with averaging multiple, highly variable cross sections. Over time, flood-derived sediment will continue to move episodically downstream, with some reaches aggrading and others degrading. Future channel surveys may show the continued downstream movement of sediment and gradual recovery of channel stability (Madej and Ozaki, 1996).

Recent flooding in the Blue Mountains provided a unique opportunity to make direct measurements of flood effects and watershed response in a dynamic mountain environment. Results from initial assessments have been used to redesign road-stream crossings and reconfigure instream habitat structures. Ongoing investigations of stream channel response show complex interactions between hillslope and floodplain-derived sediment and channel morphology. With continued regional emphasis on restoring native fish runs, understanding how large flood events affect watershed and stream processes will contribute to developing reasonable goals for water quality and restoration of aquatic habitat.

ACKNOWLEDGEMENTS

This work was conducted as part of the Region Six, USDA-Forest Service flood assessment program.

REFERENCES