Peatlands store about one third of the global soil carbon. Field-based research indicates a tight coupling between hydrological conditions in peatlands and their vegetation and soil properties and greenhouse gas emissions, suggesting that only small changes in groundwater table depth in response to human disturbance or climate change can lead to profound changes in peatland functioning. This can turn peatlands from carbon sinks into global hotspots of emissions of soil carbon. In EO4PEAT, we will push the boundaries of the satellite-based estimation of hydrological conditions in peatlands using earth observation (EO) data at multiple scales and from multiple sensors. We hypothesize that the integration of multiple satellite observations of hydrological variables, and land use and land cover (LULC) into peatland-specific models will produce estimates of hydrological conditions, including groundwater table depth, that are superior to estimates from EO data or modeling alone. EO4PEAT is focused on tropical peatlands which are understudied compared to northern peatlands and also present the highest rates of carbon dioxide emissions after being disturbed. A specific focus is furthermore given to peatlands that are dependent on surface water hydrology and currently poorly addressed in peatland-specific remote sensing and land surface modeling research.

Available satellite soil moisture retrievals are known to be biased but the dynamical information in anomaly time series alone is sufficient for many applications such as for drought monitoring or weather forecasting. However, unlike for mineral soils, a bias in moisture conditions is highly problematic over peatlands in which the absolute values of e.g., groundwater table depth is crucial to estimate ecosystem functions and flammability, i.e., fire risk. Over the last decade, a number of global land surface models have integrated peatland-specific modules to reduce bias over peatlands but they suffer from the lack of information about spatial variability of model parameters. Two of the major challenges are the representation of the various degrees of human disturbances to peatlands and the quantification of the influence of adjacent surface water on peatland hydrology because only about half of the global peatlands is supposed to be purely fed by rain. The potential of EO data for constraining the spatial variability of model parameters is huge but largely unexploited. However, crucial peatland variables such as groundwater table depth cannot be directly monitored with satellites. A promising solution to this problem is provided by data assimilation techniques that optimally combine EO data with modeling to estimate unobserved variables from EO.

The two overarching scientific objectives of EO4PEAT are (1) to enhance the accuracy of peatland hydrological estimates through innovative EO-based monitoring and EO-informed modeling techniques, and (2) to improve the process understanding in peatlands facing different types of human and climate disturbance, in support of climate change research as well as peatland conservation, restoration and management. The project will be structured into seven work packages (WP). WP1 is focused on the use of advanced radar modeling for the monitoring of inundation dynamics based on L-band active microwave data (PALSAR-1/2, Global Navigation Satellite System Reflectometry with CYGNSS). In WP2, a peatland-specific machine learning based classification and monitoring of LULC change (LULCC) will be developed using optical- and active microwave data (Landsat 5-9, Sentinel-1/2, PALSAR-1/2). In WP3, river-peatland interaction will be studied by making use of data of WP1 together with surface water elevation from SWOT and from existing EO-based products, and where available, ground elevation derived from ICESat-2 LiDAR data. The observations will be used for the development of regional and global modeling approaches of river-peatland interaction. WP4 aims at employing data assimilation approaches that include a parameter updating scheme to address model biases. Peatland models of different levels of complexity will be used for the assimilation of inundation fraction data from WP1 and observations of terrestrial water storage (TWS from Gravity Recovery and Climate Experiment; GRACE). In WP5, monitoring results from WP1 and WP2 will be interpreted together with the time series of parameters from the data assimilation in WP4 to eventually improve our understanding of processes in changing peatlands. To isolate climate change impacts, scenarios with counterfactual climate mimicking a world without climate change will be conducted. Results of WP1 through WP5 will be validated in WP6 with ground-based reference data of peatland water level monitoring stations, and river stage and streamflow gauges. In WP7, results of EO4PEAT will be disseminated to both the scientific community and various stakeholders.