The application of drones has recently revolutionised the mapping of wetlands due to their high spatial resolution and the flexibility in capturing images. In this study, the drone imagery was used to map key vegetation communities in an Irish wetland, Clara Bog, for the spring season. The mapping, carried out through image segmentation or semantic segmentation, was performed using machine learning (ML) and deep learning (DL) algorithms. With the aim of identifying the most appropriate, cost-efficient, and accurate segmentation method, multiple ML classifiers and DL models were compared. Random forest (RF) was identified as the best pixel-based ML classifier, which provided good accuracy (≈85%) when used in conjunction graph cut algorithm for image segmentation. Amongst the DL networks, a convolutional neural network (CNN) architecture in a transfer learning framework was utilised. A combination of ResNet50 and SegNet architecture gave the best semantic segmentation results (≈90%). The high accuracy of DL networks was accompanied with significantly larger labelled training dataset, computation time and hardware requirements compared to ML classifiers with slightly lower accuracy. For specific applications such as wetland mapping where networks are required to be trained for each different site, topography, season, and other atmospheric conditions, ML classifiers proved to be a more pragmatic choice.

Lowland karst aquifers can generate unique wetland ecosystems which are caused by groundwater fluctuations that result in extensive groundwater–surface water interactions (i.e. flooding). However, the complex hydrogeological attributes of these systems, linked to extremely fast aquifer recharge processes and flow through well-connected conduit networks, often present difficulty in predicting how they will respond to changing climatological conditions. This study investigates the predicted impacts of climate change on a lowland karst catchment by using a semi-distributed pipe network model of the karst aquifer populated with output from the high spatial resolution (4 km) Consortium for Small-scale Modelling Climate Lokalmodell (COSMO-CLM) regional climate model simulations for Ireland. An ensemble of projections for the future Irish climate were generated by downscaling from five different global climate models (GCMs), each based on four Representative Concentration Pathways (RCPs; RCP2.6, RCP4.5, RCP6.0 and RCP8.5) to account for the uncertainty in the estimation of future global emissions of greenhouse gases. The one-dimensional hydraulic/hydrologic karst model incorporates urban drainage software to simulate open channel and pressurised flow within the conduits, with flooding on the land surface represented by storage nodes with the same stage volume properties of the physical turlough basins. The lowland karst limestone catchment is located on the west coast of Ireland and is characterised by a well-developed conduit-dominated karst aquifer which discharges to the sea via intertidal and submarine springs. Annual above ground flooding associated with this complex karst system has led to the development of unique wetland ecosystems in the form of ephemeral lakes known as turloughs; however, extreme flooding of these features causes widespread damage and disruption in the catchment. This analysis has shown that mean, 95th and 99th percentile flood levels are expected to increase by significant proportions for all future emission scenarios. The frequency of events currently considered to be extreme is predicted to increase, indicating that more significant groundwater flooding events seem likely to become far more common. The depth and duration of flooding is of extreme importance, both from an ecological perspective in terms of wetland species distribution and for extreme flooding in terms of the disruption to homes, transport links and agricultural land inundated by flood waters. The seasonality of annual flooding is also predicted to shift later in the flooding season, which could have consequences in terms of ecology and land use in the catchment. The investigation of increasing mean sea levels, however, showed that anticipated rises would have very little impact on groundwater flooding due to the marginal impact on ebb tide outflow volumes. Overall, this study highlights the relative vulnerability of lowland karst systems to future changing climate conditions, mainly due to the extremely fast recharge which can occur in such systems. The study presents a novel and highly effective methodology for studying the impact of climate change in lowland karst systems by coupling karst hydrogeological models with the output from high-resolution climate simulations.

Conventional methods of monitoring wetlands and detecting changes over time can be time-consuming and costly. Inaccessibility and remoteness of many wetlands is also a limiting factor. Hence, there is a growing recognition of remote sensing techniques as a viable and cost-effective alternative to field-based ecosystem monitoring. Wetlands encompass a diverse array of habitats, for example, fens, bogs, marshes, and swamps. In this study, we concentrate on a natural wetland – Clara Bog, Co. Offaly, a raised bog situated in the Irish midlands. The aim of the study is to identify and monitor the environmental conditions of the bog using remote sensing techniques. Environmental conditions in this study refer to the vegetation composition of the bog and whether it is in an intact (peat-forming) or degraded state. It can be described using vegetation, the presence of water (soil moisture) and topography. Vegetation indices (VIs) derived from satellite data have been widely used to assess variations in properties of vegetation. This study uses mid-resolution data from Sentinel-2 MSI, Landsat 8 OLI for VI analysis. An initial study to delineate the boundary of the bog using the combination of edge detection and segmentation techniques namely, entropy filtering, canny edge detection, and graph-cut segmentation is performed. Once the bog boundary is defined, spectra of the delineated area are studied. VIs like NDVI, ARVI, SAVI, NDWI, derived using Sentinel-2 MSI and Landsat 8 OLI are analysed. A digital elevation model (DEM) was also used for better classification. All of these characteristics (features) serve as a basis for classifying the bog into broad vegetation communities (termed ecotopes) that indicate the quality of raised bog habitat. This analysis is validated using field derived ecotopes. The results show that, by using spectral information and vegetation index clustering, an additional linkage can be established between spectral RS signatures and wetland ecotopes. Hence, the benefit of the study is in understanding ecosystem (bog) environmental conditions and in defining appropriate metrics by which changes in the conditions can be monitored.

Peatlands are vital to the global carbon (C) cycle as they act as a significant C store and these systems in Ireland store between 1064 –1503 Gt C on ~20% of the land area. However, around 90% of this area has been drained and degraded by various anthropogenic activities and the emissions from these activities are approximately 3 million t C per year. A better understanding of the land-atmosphere C and greenhouse gas (GHG) dynamics is vital to halt these emissions and enhance the C sink strength of these ecosystems. Gross Primary Productivity (GPP) is a major part of the peatland carbon cycle and detailed knowledge of the spatial and temporal extent of GPP is imperative for improving our predictions of peatland ecology, biogeochemistry and carbon balance in response to global change. Eddy covariance (EC) techniques are widely used to measure carbon fluxes but can only account for fluxes within the flux footprint of the tower, and it is challenging to scale up data from EC towers to regional and global scales due to the limited number of towers and their geographic locations. This research assesses the relationship between remote sensing and ground-based measurements for a near-natural raised bog in Ireland using EC techniques and high-resolution Sentinel 2A satellite imagery. Vegetation indices (VIs) are one of the key input parameters for satellite-based GPP and most of the existing VIs have been developed for grassland, agriculture, and forest ecosystems. This study developed a hybrid index for raised bogs using multiple linear regression and six widely practiced conventional vegetation indices. Two approaches have been used in this study for estimating GPP using the LUE model. Initially, all the individual indices have been used to model the GPP, which was subsequently compared with the EC GPP to determine the performance of each index against the EC data. The model was run with meteorological data and satellite-derived vegetation indices. During the 2018 study period, the weather was exceptionally dry which made it challenging and rewarding at the same time as the hybrid index was developed for an exceptional year. It was crucial to test the performance of the hybrid index under more normal weather conditions with limited clear sky satellite imagery. Therefore, the hybrid index was validated for the year 2019 which had normal weather conditions. The hybrid index based modelled GPP showed a significant correlation with the EC GPP for both the years with an R2 > 0.95. Overall, this research has demonstrated the potential of combining EC techniques and the hybrid index along with satellite-derived models to better understand and monitor key drivers and patterns of GPP of raised bog ecosystems under different climate scenarios.

Alkaline fens are listed under Annex 1 of the European Union (EU) Habitats Directive (Council Directive 92/43/EEC, habitat code 7230) as habitats requiring special conservation measures, including the designation of suitable sites as Special Areas of Conservation (SACs). These fens are peat-forming wetlands predominantly fed by groundwater containing significant concentrations of calcium, magnesium and bicarbonate. The hydrogeological dynamics and hydrochemical signature supports small sedge and brown moss communities in a mosaic of different habitats. Despite fens being an important part of the natural landscape in Ireland as well as one of the most threatened wetland habitats in Europe, there is little information on the hydrology and hydrochemistry that support these habitats. As part of a three year research project (Ecometrics) on GWDTE’s (Groundwater Dependent Terrestrial Ecosystems) an intensive hydrochemical monitoring programme was established on four Irish fen sites, designated as SACs, covering an eco-hydrological gradient from intact to highly degraded conditions. Ground and surface waters monitoring started in July 2018 with spot measurements supplemented by a continuous water table time series collected every two months. Simultaneously ground and surface water samples were taken and analysed for nutrients, mayor ions and metals. Topographical surveys as well as vegetation surveys were carried out in 2019. The hydrological and hydro-chemical evidence from each fen were then collated to build conceptual eco-hydrological models to represent both temporal and spatial variability in each geological setting. In addition, remote sensing was used in order to investigate the relationship between vegetation and water levels. Utilising ground surveyed habitat map and Sentinel-2 (S2) imagery in a Random Forest model a remotely sensed vegetation map was created. This information was then correlated with water level data as follows: hydrological data were interpolated over the entire area of the fen, giving seasonal information for both surface and groundwater levels. Using K-means clustering, the data were divided into clusters which were then matched and correlated with the vegetation map produced through S2 imagery. The strength of the correlation between water level and fen vegetation can further aid the construction of conceptual models for the four research sites. These models will then be used to define appropriate metrics that characterise the environmental supporting conditions in fens, as required for the EU Water Framework Directive. Here, the preliminary conceptual models for fen functioning will be presented as they are continuously updated by the ongoing data collection.