Dendroclimatological Reconstruction of Streamflow Variability in a Small, Semi-Arid Mountain Catchment

Publication Date


Type of Culminating Activity


Degree Title

Master of Science in Geology



Major Advisor

David E. Wilkins


Tree rings—the expression of annual cambial growth—from moisture-sensitive species can be used as non-linear biological proxies for streamflow variability as both streamflow and a tree's growth respond to or co-vary with precipitation inputs to a watershed hydrologic system. Many researchers have reconstructed streamflow using tree-rings with varying degrees of success, explaining ~ 70 to 80 percent of streamflow variability in sub-continental (> 10,000 km2) to sub-regional (260 — 10,000 km2) basins, particularly in the drought-prone areas of western North America.

The objective of this study is to reconstruct streamflow with statistical validity in the Reynolds Creek Experimental Watershed (RCEW), a 239 km2, second- or third-order stream catchment located in the snowmelt-dominated Owyhee Mountains of southwest Idaho.

Tree-ring chronologies developed from Douglas-fir (Pseudotsuga menziesii {Mirb.} Franco), subalpine fir (Abies lasiocarpa {Hook.} Nutt.), western juniper (Juniperus occidentalis Hook.), and a set of aspect-limited series were tested for significant correlations with streamflow time series for the years 1966 to 1997 and three-site-averaged precipitation time series for the years 1962 to 1996, using correlation procedures in SAS. Significant correlations found between streamflow time series and tree-ring chronologies were investigated further using different curve fits for improvement of the correlations.

Transfer functions, from significant but moderate correlations between a one-year-lagged aspect-limited standard tree-ring chronology and gaged February cumulative streamflow (r = 0.5201, p < 0.01) and between gaged water year cumulative streamflow (r = 0.4036, p < 0.05), were used to reconstruct 110 years of streamflow. Precipitation, an independent dataset that was highly correlated with cumulative water year streamflow (r = 0.956, p < 0.0001), was used for model validation and resulted in weak but statistically significant correlations (r = 0.3575, P < 0.05 for the February reconstruction and r = 0.3870, P < 0.05 for the water year reconstruction).

Validation results explain only about a third of streamflow variability: This weak explanation of variability can be attributed to many factors, including a lack of sample depth in the chronologies as mature (older than 100 years) trees were scarce, a lack of sensitivity to soil moisture in the samples, the use of precipitation (a second order approximation) as a proxy for streamflow, and the possible presence of more prevalent limiting growth factors at sampling sites, such as temperature extremes in the high elevations, and site specific, non-climatic effects from logging practices and disease and insect attacks.

The scale of the study is also believed to be a contributing factor to the lack of variability explained in the tree-ring signal. Site specific local noise in the tree-ring data is assumed to be amplified in smaller catchments such as the RCEW as a result of less spatial distribution between trees sampled and a lack of sample depth.

Files over 30MB may be slow to open. For best results, right-click and select "save as..."