The Influence of LWD and Tributary Confluences on the Local Grain Size Distributions of the H.J. Andrews Stream Network

The influence of Large Woody Debris (LWD) and tributary confluences on the grain size distributions of a mountainous stream network in the Western Cascades was analyzed. Longitudinal Trends in grain sizes were determined to be increasing with downstream distance. The influences that LWD and tributary confluences have on these trends were investigated using spatial and bivariate correlations. The median and fine fraction grain sizes were found to generally not be spatially correlated. Inconclusive correlations were determined between LWD abundance and local grain sizes, as well as for the local influences of tributaries. Considerations for future work include additional grain size sampling and log-jam type classification. Introduction The longitudinal trend of decreasing grain size with increasing downstream distance within a stream network is thought to be largely attributed to the physical phenomena of abrasion and sorting of grains during transport. There have been several predictions of the functional relationship that best describes these trends. An exponentially decreasing trend has been observed for the mean grain size of sandstone and shale (Knighton, 1980), while power functions have been observed to better describe longitudinal reduction in the grain size of coarse alluvial gravels in which the reduction mechanism was thought to be primarily sorting (Brierley and Hickin, 1985). Sorting coefficients, determined using the Folk and Ward’s 1957 method, have also been observed to decrease as downstream distance increases and hypothetically follow a sinusoidal pattern (Knighton, 1980). The irregularities in longitudinal trends have been hypothesized to be partly due to tributary confluences (Knighton, 1980). Such irregularities in the form of downstream coarsening and upstream fining have been observed to occur at tributary confluences with the main stem (Benda et al., 2004). However few studies have been performed to investigate the influences of tributaries and Large Woody Debris (LWD) on the grain size distribution throughout a stream network. Understanding the effect that LWD and tributaries have on the local grain size distribution within a stream network has great importance in regards to ecological conservation and land management. The objective of this investigation is to evaluate the longitudinal trends of, and the extent to which LWD and tributary confluences influence, the grain size distribution within a mountainous stream network in the Western Cascades. Methods Study site The H.J. Andrews experimental forest was first established in 1948, and in 1980 became a member of the National Science Foundation’s Long-Term Ecological Research program. It is located within the Western Cascades of Oregon. Within the forest lies the Lookout Creek watershed, and accompanying stream network (Figure 1). The watershed drainage area is approximately 15,800 acres, and consists of diverse landscapes due to geologic variation and land disturbance history. Figure 1: HJ Andrews stream network with designated study segments. Data Collection Longitudinal surveys of channel cross sections were conducted every 50 meters for approximately 11 kilometers of streams in the H.J. Andrews Experimental Forest. Data was collected from a total of seven segments throughout the network (Figure 1). The data collected included cross section location, grain size distributions, channel width and gradient, and an inventory of LWD. Locations were georeferenced using Avenza maps software. Grain size was recorded into geometric size classes with use of gravelometers and Wolman particle counts (sample size of 100 per cross section). Channel width was measured with a measuring tape and included mid-channel bars if side channels were thought to be part of the active channel. Channel gradient was measured every 50 meters with use of a clinometer and stadia rod. LWD was counted via estimates of size based on the classification developed by Czarnomski et al (2002). Differentiations were made between single pieces of LWD and pieces that formed log-jam accumulations. LWD was categorized as a logjam if there were at least three pieces of LWD and two points of contact between the pieces. The location of prominent features such as tributary confluences, log-jams, debris flow runout tracks, road crossings and transitions from old growth forest to new growth were noted. Grain Size Distributions Using the data collected, grain size distributions were determined for each individual cross section using a log base two interpolation between grain size categories (Equation 1, DSWC 2004). Important grain sizes (mean, d16, d50, and d84) were compared against downstream distance for individual segments, and the entire network, to determine present trends.                        P P S S P P S S 2 2 2 log log * log 2 (1) Where: S = grain size (mm) S + = grain size at the top of range (mm) S = grain size at the bottom of range (mm) P = grain size percentile of inquiry P + = percent of particles smaller than S + P = percent of particles smaller than S Spatial Autocorrelation Correlation techniques were performed on individual segments to determine if correlation was present. Correlations in the local grain size distributions between cross sections were analyzed spatially with autocorrelation and with wood volume in a bivariate analysis. . Spatial autocorrelation determines whether a particular sample of interest is correlated spatially, and the lag at which the particular variable can be estimated based on the sample taken. The autocorrelation function (ACF) (Equation 2) was used in the R programming language with a 95% confidence interval to determine if there was spatial correlation for the median and sixteen-percentile grain size (d16) of each segment.