Research Methods for Assessing the Impacts of Forest Disturbance on Hydrology at Large-scale Watersheds

Abstract

The impact of forest disturbance on hydrology has long been an important research topic, but the majority of this research has been conducted on small-scale watersheds. Large-scale watershed studies are hampered by the difficulty of conducting paired watershed experiments, and by insufficient data, significant landscape complexity and a lack of commonly-accepted research methodologies. However, the ever-increasing demand on information at large scales to support forest and watershed planning and management highlights the need for large-scale watershed research. This paper provides a review of research methods for assessing impacts of forest disturbance on hydrology in large-scale watersheds. It focuses on definition of large-scale watersheds, quantification of cumulative forest disturbance, and research methods used to detect its impact on hydrology.

There is no a commonly-agreed definition of large scales for watersheds. Researchers have called various sizes as large scales, leading to confusion and inconsistence of comparisons. We suggest usage of a common size (≥1,000 km²) to define large-scale watersheds, which is consistent with the majority of published research reports reviewed. Forest disturbances including human (e.g. timber harvesting) and natural ones (e.g. wildfire) are typically cumulative over time and space in large-scale watersheds. How to use a single indicator to represent this cumulative disturbance is challenging. We suggest a concept named “equivalent disturbed area“ (EDA) to replace traditionally applied forest cover measures as the former considers hydrological recovery process during forest re-establishment following disturbance. Several recent studies of this kind have confirmed that EDA is a better indicator or concept for studying watershed processes and functions particularly at large scales. Because of the difficulty of applying paired watershed techniques to large-scale watershed study areas, researchers have generally used modeling and statistical approaches. In this review, we grouped statistical techniques according to the number of selected watersheds. For a single watershed, methods like time series analysis, Bayesian methods, non-parametric analyses and flow duration curves could be used whereas to multiple watersheds double mass curve and regression techniques could be applied. We conclude that the selection of research methods largely depends on data availability and the number of available watersheds experiencing a gradient of forest disturbances. As there is no commonly-accepted method, a combination of several techniques can provide a more robust assessment than just one alone. Future research directions are also discussed.