Emerging technology can guide ecosystem restoration for future water security

Abstract

Water is the primary medium through which society will experience the effects of climate change. Altered precipitation patterns, enhanced evapotranspiration rates, loss of snow and ice, declines in groundwater storage, and increased risk of flooding and droughts will all have important implications for our water resources. In fact, we are already experiencing climate-related perturbations to the water cycle, as manifested through increases in temperatures and frequency of weather extremes across the globe (IPCC, 2021). However, a critical but often neglected aspect of predicting future hydrological change is recognizing the implications of historical land-use decisions that can act to either enhance or reduce the resilience of both the quality and quantity of water. Such land use decisions may create 'locked-in' effects reflecting the policies and path dependencies that underpin management inertia (Unruh & Carrillo-Hermosilla, 2006), as well as physical disturbances that caused hydrological changes that now are difficult to reverse (Lindenmayer et al., 2011). While water challenges are global, the water cycle in boreal regions is perhaps uniquely vulnerable due to cumulative effects of land use and climate warming, with the latter proceeding at faster rates at high latitudes (Buntgen et al., 2021). Climate change in the north is predicted to give rise to increasingly snow-free winters and warmer summers in boreal landscapes, which collectively will reduce water storage. Superimposed on these changes are barriers or inertia due to past land-use decisions, that may exacerbate pressures on ground- and surface waters. The most widespread management activity in many boreal countries is the historical ditching of wet soils. These ditches are constructed waterways that were dug, often by hand, to remove excess water affecting several million hectares of land, especially in Finland and Sweden, but also in many other northern countries (Strack, 2008; Figure 1). This resulted in improved forest growth and transformed wet soils and peatlands to productive forests, but also came with unwanted environmental consequences. Drainage ditches are sources of enhanced methane emissions (Peacock et al., 2021) and elevated dissolved organic carbon concentrations (Nieminen et al., 2021). Low water tables also lead to enhanced soil carbon mineralization and increased terrestrial greenhouse gas (GHG) emissions (Evans et al., 2021). The biogeochemical properties of these extensions to the drainage network dramatically alter our understanding of running waters as sources of GHGs to the atmosphere and the overall provisions of ecosystem services. Most importantly, in an era increasingly affected by climate change, we now face conditions that create the opposite obstacle that the ditches aimed to solve—insufficient amounts of water—due to the expected increase in frequency, duration and severity of drought (Spinoni et al., 2018). Historical ditching may thus create enhanced vulnerability to drought in the future, which would require massive efforts to reverse. Documenting the cumulative effects of such local-scale management activities raises complex challenges that new technologies can help solve. In the case of ditching, the full scope of the problem has largely been hidden because existing models fail to map landscape hydrography when natural processes no longer have primacy over channel initiation. Nevertheless, with recent modelling advances, small-scale waterways can now be identified and mapped using high-resolution airborne laser scanning (ALS), combined with tools based on artificial intelligence (AI; Sit et al., 2020). We mapped small channels across all of Sweden with a deep neural network (Lidberg, Shekhar, et al., 2022). The deep learning model was based on ALS data and 1607 km of manually digitized ditches (Ågren et al., 2022). The ALS point cloud had a last return density of 0.5–1 points m−2 and a surface model with 1 m resolution was created by the Swedish land use and cadastral registration authority. A high pass median filter (HPMF) was applied to the surface model to emphasize short-range variability in the topography. The HPMF-algorithm was implemented in Whitebox tools (Lindsay, 2014) and operates by subtracting the value at the grid cell at the centre of the window from the median value in the surrounding neighbourhood with a kernel of five cells. Negative values indicate depressions while positive values indicate ridges. The digitized vector lines have no width so we utilized average ditch width from a field inventory where 2188 ditch channels were surveyed across Sweden. The average ditch width was 2 m with a standard deviation of 1.3 m. Instead of flagging all pixels within ~3 m of a vector line as ditch, we utilized the HPMF to create more natural ditch labels. Pixels within 3 m of a vector line and with a HPMF value less than −0.075 were flagged as ditch pixels. TensorFlow 2.6 was used to build an encoder–dec