Identifying step changes in single streamflow and evaporation records due to forest cover change

Techniques that identify forestry-induced changes to streamflow or evaporation are needed to assess available water resources. Equally, there is a growing appreciation that climate cycles may be having a profound impact on the land-surface hydrology. The ability to see forestry-induced change above the effects of climate dynamics, therefore, becomes a critical issue. Paired-catchment analyses have proved very valuable in identifying change, but cannot quantify the relative impacts of climate and land-cover change, and data from adjacent reference basins are not always available. Within this study, we examined whether step changes within single time-series of streamflow or evaporation (P-Q) could be identified without reference to those of a control catchment. The UC-DHR method was used for this analysis, and included a special routine to allow a known change-point (e.g. start of logging) to be specified or alternatively identified by the model. Data from three experimental catchments important for their seminal forestry impact studies were selected for the analyses. The study demonstrated that clear-cutting 29 % of the Hore catchment and 40 % selective felling of the Berembun basin produced a step change in the discharge trend that was clearly observable above the climate-related dynamics and uncertainty. In contrast, step changes in P-Q following the same selective felling event or following 22 % afforestation of the Upper Hodder basin were not larger than the uncertainty bands or magnitude of the inter-annual cycles produced by the climate dynamics, respectively. This demonstrates that while step changes can be observed in single hydrological time-series, errors within the observations can sometimes mask the identification of change. This masking of change is also possible where th