Peakflow responses to forest practices in Oregon, USA - A modelling approach

Table of Contents

Peakflow Responses to Forest Practices in Oregon, USA - a Modelling Approach

Peak flow responses to clear cutting and roads?

Paired watershed approach

Modelling approach to quantify land-use-change effects

Simulation vs. observation

Simulation vs. simulation

The HBV model

Model application

Model residuals

Summary of simulation-comparison for watershed 1

Summary of simulation-comparison for watershed 2

Lookout Creek

Summary of simulation-comparison for Lookout Creek

Conclusions

Consistent data?

Simulation of 1996-flood in watershed 1

Largest events in watershed 1

Model approaches

Model-approach rollercoaster

Model parameter values

Model parameter values

Model approaches

Home Page: http://www.cof.orst.edu/cof/fe/watershd/

Abstract
The issue of peakflow responses to clear-cutting and forest-road building is a highly charged issue in the western Cascades of Oregon, USA. To date, the effects of forest practices on streamflow response has mostly been studied using paired-catchment approaches. This paper presents model approaches as an alternative to paired-catchment methods. Model approaches are of particular interest for evaluating land-use-change effects in larger catchments, for which a suitable control catchment might not exist. The Swedish HBV model was used for the study of peak flow changes in hopes of contributing positively to the dialogue about peakflow response to land-use change. These effects were analyzed in different ways: a) Calibration for a period before (or after) change and simulation of runoff, which would have been observed without land-use change (reconstruction of runoff series), b) Comparison of calibrated parameter values for periods before and after a land-use change, and c) Comparison of runoff predicted with parameter sets calibrated for periods before and after a land-use change. While the first method was seemingly straightforward, we found that it is impeded by parameter uncertainty and problems of parameter identifiability. Since the model was not perfect for the pre-change conditions, it also had to be acknowledged that deviations between observed runoff and simulated runoff (=runoff with unchanged conditions) could be caused by both the land-use change and model inaccuracies. Monte Carlo techniques were used to address the problem of parameter uncertainty and to allow the computation of confidence intervals in model results and parameter ranges. We compared these different methods in the context of peakflow response to forest practices in the western Cascades for three 1-km2 catchments in the HJ Andrews Experimental Forest (watersheds 1 to 3) and two 100-km2 watersheds (Lookout and Blue River).