|
<br />, I
<br />
<br />inferred from damage to vegetation). Bucket data and
<br />nearby systematic gaged data a.re used to correlate
<br />rainfall data with geomorphic evidence accounting for
<br />variability such as soil type and cohesiveness,
<br />vegetation cover, and hillsiope gradient and length,
<br />Then, variations in the geomorphic evidence in the
<br />storm area are used to estimate a rainfall amount where
<br />geomorphic data are available. Finally, an isohyetal
<br />map is drawn considering available data. Sometimes,
<br />storm path can be inferred for recent storms.
<br />In the second or hydrologic method, indirect
<br />estimates of peak discharge can be obtained for many
<br />small basins in the area affected by the storm. The
<br />emphasis is obtaining peak-discharge data for small
<br />basins (<1-3 km2) where spatial and temporal rainfall
<br />variability is assumed to be small, thus, providing more
<br />reliabie paleohydrologic rainfall estimates, High-water
<br />marks (HWMs) of recent floods or paleostage indicators
<br />(essentially old HWMs) for paleofloods, channel
<br />geometry, and hydraulic data for a stream are used to
<br />estimate peak discharge such as with the critical-depth
<br />method (Jarrett, 1991; Jarrett and Tomlinson, 2000).
<br />Rainfall-runoff( (RF-RO) modeling with physical basin
<br />attributes also can be used to derive independent
<br />estimates of rainfall for each small basin. RF-RO
<br />modeling is used to back-calculate rainfall intensity and
<br />amounts from the peak discharge and a description of
<br />basins. Limited space precludes including RF-RO model
<br />results. Hydrologic rainfall estimates are used to: (1)
<br />help draw the isohyetal map with available geomorphic
<br />rainfall estimates; (2) develop isohyetal maps for
<br />historic and prehistoric rainstorms; (3) compare results
<br />with other independent sources at rainfall data or; (4)
<br />provide rainfail estimates if no other source exists.
<br />
<br />4. RESULTS
<br />
<br />No systematic precipitation or streamflow monitoring
<br />existed in the Buffalo Creek burned area in 1996, On
<br />Juiy 15-16, 1996 (before additional rainstorms), data
<br />collection consisted of obtaining rainfall bucket data,
<br />peak flow, and paleohydrologic data for the area.
<br />Eleven rainfall bucket observations were available from
<br />residents who had various types of plastic rain gages.
<br />Maximum rainfall for the July 12, 1996 storm was about
<br />80 mm in the community of Buffalo Creek and
<br />headwaters of Spring Creek (fig. 1a); residents stated
<br />most of the rain fell about 2000 to 2100 MDT.
<br />The amount and location of fresh rill and gully erosion
<br />on hillslopes generally less than 5-10 m in length was
<br />compared to nearby rainfall amounts for the July 12th
<br />storm. Hiilslopes (burned or unbumed with sparse
<br />vegetation) with less than about 25 mm rain (bucket
<br />data) had some sediment movement and minimal rill
<br />development. Hillslopes that received about 50 mm of
<br />rain typically had rills about 75 mm deep and 50 mm
<br />wide, Hiils/opes that received about 75 mm of rain
<br />typically had extensive rilling and numerous gullies up to
<br />0.5 m deep and a meter wide. HilIslope erosion in areas
<br />
<br />without any bucket data then was used to estimate
<br />rainfall in areas from these general relations, Numerous
<br />gullies up to a meter deep and 3 m wide, very extensive
<br />rilling, and large headcuts were documented in an area
<br />about 2.5 km southeast of Buffalo Creek near the
<br />headwaters of Sand Draw, Spring Creek, Shinglemiil
<br />Creek, and Spring Gulch. This area of maximum erosion
<br />was used to infer the location and area of maximum
<br />storm rainfall amount of at least 110 mm (fig, 1a).
<br />Large quantities of sediments were mobilized on
<br />hiilslopes and in channels in the bumed area during the
<br />July 12th storm. A distinct black, burn boundary (line)
<br />on rocks defined pre-flood ground surfaces and was
<br />used as a reference to estimate the general surface
<br />erosion from sheetwash (overland flow). Care was
<br />taken to estimate general erosion rather than the local
<br />erosion around a rock, In addtlion, piilars of soil were
<br />preserved under some surface rocks and metal objects
<br />on the bumed areas. The area of maximum sheetwash
<br />also was limited to the headwaters of Shinglemill Creek,
<br />Spring Gulch, Sand Draw, and Spring Creek.
<br />Sediments moved on hiilslopes ranged from silt to
<br />cobble-sized material, and 2.5-m diameter boulders
<br />were transported in some channels. A large amount of
<br />the flood-transported sediment was deposited as
<br />alluvial fans. Many new fans had dimensions of about
<br />100 m x 30 m x 1.5-2 m such as in Sand Draw, Spring
<br />Creek, Shinglemiil Creek, and Spring Gulch. Well
<br />preserved, fresh tributary fans on the Buffalo Creek
<br />floodplain were used to infer relative flood timing, and
<br />that the storm moved easterly (downstream), which
<br />likely exacerbated flooding.
<br />In unburned vegetated areas, rainfall also was
<br />inferred by the amount of duff that floated and was
<br />repositioned by sheetwash, Rainfall less than about 50
<br />mm (depending on duff thickness and composition)
<br />partially floated and reoriented needles, twigs, and other
<br />elongated debris perpendicular to the flow direction and
<br />spaced about 2 to 4 cm distance apart, These micro-
<br />scale features appear to have functioned as small dams
<br />ponding rainfall and hindering rainfall runoff. Rainfall
<br />greater than about 50 to 75 mm produced a cascading
<br />failure of these small debris dams. Small channels were
<br />preserved within the duff (similar to channel incision),
<br />which enabled peak discharge estimation. These
<br />features are similar to log jams in rivers such as the
<br />1982 Lawn Lake dam failure in Rocky Mountain National
<br />Park, located about 100 km northwest of Denver,
<br />Colorado (Jarrett and Costa, 1986).
<br />Peak-flow estimates for twenty streams, ranging in
<br />size from about 0.1 km2 to the total burned area of about
<br />50 km2, also were used to help identify areas of
<br />maximum rainfall. These basins have varied
<br />characteristics such as vegetation cover, burn intensity
<br />(including no burn), watershed aspect and slope, and
<br />sediment sizes. A number of severeiy-burned small
<br />basins in areas of maximum rainfall had unit discharges
<br />(peak discharge divided by drainage area) from about 45
<br />to 60 m3/s/km2, [For comparison, the maximum unit
<br />
|