|
<br />in Buffalo Creek. Because the area of maximum rainfall was within the burned area, rainfall-runoff
<br />modeling is necessary to estimate potential flood runoff, without the fire.
<br />
<br />It is more difficult to assess whether the fire, which dramatically changed the local surface-energy
<br />balance, exacerbated the severity of the July 12,1996 rainstorm. In general, a burned area will
<br />heat up more quickly from solar energy than adjacent unburned areas, and would therefore be a
<br />site for initiating convective processes (N. Doesken, Colorado Climate Center, written commun.
<br />1998). Doesken noted that once convection is initiated, there is little evidence that burned areas
<br />would serve to enhance the continuation or intensity of storms. Henz (1998) suggested that a
<br />gust front provided the meteorological mechanism that caused the Buffalo Creek thunderstorm to
<br />go stationary over the watershed. He further noted that the storm was the only one that went
<br />stationary during severe weather in much of eastern Colorado on July 12, 1996. It may be
<br />coincidental that rainfall intensified once the storm passed over the burned area and qUickly
<br />dissipated after it moved eastward over unburned basins. In 1996 and 1997, there were five
<br />rainstorms that had recurrence intervais greater than 10 years. Although there is a small chance
<br />this many large storms would occur in two summers, it may suggest the burned area is having an
<br />effect on storm meteorology. Another possibility is that the rainfall-frequency relations for Buffalo
<br />Creek (fig. 5) are questionable because there are no gaged data near Buffalo Creek; these
<br />relations were based on two independent regional analysis (Miller et aI., 1973; Diller, 1 997) that
<br />produced similar results. Henz (1974,1998) noted that Buffalo Creek is one of several preferred
<br />thunderstorm generation areas or hotspots in eastern Colorado. Thus, rainfall frequency for
<br />Buffalo Creek may not be adequately represented by generalized rainfall-frequency relations for
<br />the region. Research is needed to determine the effect of hotspots on rainfall intensity, amount,
<br />and frequency, and the meteorology of fire-caused modifications to the surface-energy balance on
<br />precipitation.
<br />
<br />Effects of Watershed Rehabilitation
<br />
<br />Watershed-rehabilitation efforts utilized to help restore the Buffalo Creek burned area include
<br />aerial and ground seeding; bonded-fiber matrix; soil tilling; contour tree felling; log and strawbale
<br />check dams; and untreated natural recovery (Casey Clapsaddle, USFS, wntten commun., 1996).
<br />Extensive efforts to mechanically break up hydrophobic soils and slow water and sediment runolf
<br />began very soon after the fire. Most efforts were in basins posing greatest risk to the public:
<br />Sand Draw, Spring Gulch, and Shinglemill Creek. A moderate flood on June 12, 1996 (Casey
<br />Clapsaddle, USFS, written commun., 1996) and the severe flash flood on July 12, 1996 washed
<br />out most of the initial rehabilitation efforts. Small amounts of water were applied to burned areas
<br />(simple infiltration tests) in 1996 and 1997. Generally, no water infiltrated and small droplets of
<br />water formed, indicating hydrophobic soils in 1996. During the wet spring of 1997, applied water
<br />infiltrated rapidly, even on very steep hillslopes (>30 %). However, after the soils dried,
<br />infiltration was low due to reformed hydrophobic conditions or other factors. Data monitoring in this
<br />study complements an instrumented, paired-basin analysis being conducted by Casey
<br />Clapsaddle to assess the rehabilitation efforts used in Shinglemill Creek and Morrison Creek
<br />basins (burned area was left untreated).
<br />
<br />Despite extensive rehabilitation efforts in bumed area, smaller rainstorms after July 12 in 1996
<br />and in 1997 produced similar rates of runoff (fig. 10). On August 31, 1997, a 63 mm rainfall in
<br />about 30 minutes over the headwaters of Sand Draw, produced a flash flood of about 34 m3/s
<br />from about 1.5 km2, which reflect persistent conditions that exacerbate flash-flood potential and
<br />minimal watershed recovery. Natural debris (trees and sediment) present in many channels
<br />appeared to help slow water and sediment runoff for many events. In addition, small runoff
<br />events (as much as -3 m3/s) from burned hills/opes that reached channels having thick (>3 rn)
<br />pea-gravel sediments such as in Sand Draw often infiltrated in a short distance. Long-time
<br />residents indicated that since the fire, streams in the burned area have more flow that usual.
<br />
<br />Base flows as small as 0.2 m3/s after the July 12, 1996, flood were competent to incise and
<br />erode much of the new alluvial fans. Many fans had several agradation-degradation cycles for
<br />small events (peak flows of -1-2 m3/s) since the fire, which reflects channel instability. Lower
<br />tributaries reaches that agraded on July 12, 1996, generally had degraded by about a meter by
<br />
<br />Draft 3/30/98
<br />
<br />7
<br />
|