Restoration effects on the hydrological response of peatlands

Peatlands are a fundamental component of the environment, supporting biodiversity and providing essential ecosystem services, yet many have been extensively drained through networks of artificial ditches. Restoration through ditch removal is considered an effective nature-based strategy for climate adaptation and flood mitigation, but its effect on the timing and magnitude of the catchment response is still poorly quantified. Classical hydrological theories describe the transformation of effective rainfall into discharge through a static Instantaneous Unit Hydrograph (IUH), which cannot account for a drainage network that changes as ditches are removed. Within this context, the present thesis investigates how the progressive removal of ditches modifies the hydrological response of three Norwegian peatland catchments, exploiting a length dependent IUH, f(τ, L), formulated as a function of the total length L of the drainage network. The case studies differ in size, drainage density and network complexity and are analyzed using high-resolution Digital Terrain Models. The framework relies on a generalized description of the mean hillslope length <l>, which is evaluated for removal sequences proceeding from the headwaters to the outlet (Upstream–Downstream) and from the outlet to the headwaters (Downstream–Upstream). The resulting analytical envelopes are shown to reproduce the DTM-derived data satisfactorily. The hydrological response is then investigated for the Upstream–Downstream removal sequence, by solving the length dependent IUH numerically to simulate network reduction and exploring its sensitivity to the recession coefficient k, the hyporheic velocity uh and the parameter β of the impact coefficient. In all catchments the modelled response shows a monotonic reduction of the peak discharge and a progressive delay of its timing as ditches are removed. The hyporheic velocity emerges as the dominant control on the magnitude of these changes, while the delay of the peak increases with the size of the drainage network, being longest in the largest catchment. The peak reduction at maximum restoration is largely insensitive to β within its admissible range. Since only k is estimated from observations, while uh is explored over a range and β is calibrated for each catchment, these results represent a modelled sensitivity rather than a calibrated prediction. Moreover, the impact coefficient depends on the network only through its total length and is calibrated on the Upstream–Downstream sequence: it does not resolve the spatial arrangement of the removed ditches, which affects the mean hillslope length, as the difference between the two removal sequences shows, and in principle also the response. Resolving it is left for future work. This thesis links the mean hillslope length to the time-domain response of drained peatlands, providing a parsimonious framework to explore the hydrological effects of restoration.