TL;DR
This paper presents a new dataset of global wetland extent maps from 1980-2020, generated using a hydrological model (TOPMODEL) and a combination of soil moisture reanalysis data and observational wetland datasets. This long-term, high-resolution dataset addresses a critical gap in understanding wetland dynamics, which are crucial for biodiversity, hydrology, and greenhouse gas emissions.
What Problem Does This Paper Address?
Wetlands play a vital role in Earth's systems, influencing biodiversity, water cycles, and greenhouse gas emissions (especially methane). However, a significant challenge has been the lack of comprehensive, long-term datasets that capture the dynamic nature of wetland extents. Existing wetland inventories are often static, providing only a snapshot of maximum extent, or are limited in their spatial or temporal coverage. This makes it difficult to study how wetland areas change over time due to climate variability, land-use change, or other factors, hindering our ability to accurately assess their impact on global processes.
Methods
The researchers employed a diagnostic model based on the TOPography-based hydrological MODEL (TOPMODEL). This model uses topographic information to simulate water flow and saturation within a landscape. To create the wetland maps, they adapted TOPMODEL to estimate wetland fraction based on simulated water table depth, which is derived from soil moisture reanalysis data. This approach allows for the simulation of wetland dynamics at a fine spatial resolution (0.25° x 0.25°). The model's parameters were calibrated using a combination of four different observation-based wetland datasets to ensure robustness and capture different definitions of wetlands. An ensemble of 28 different simulations was generated by using seven different soil moisture reanalysis datasets, providing a measure of uncertainty.
Datasets
The study utilized the following datasets:
- Soil Moisture Reanalysis Data (7 datasets): NCEP-DOE, MERRA-Land, MERRA-2, ERA5, ERA5-Land, GLDAS-Noah v2.0, and GLDAS-Noah v2.1. These datasets provided the input for the TOPMODEL simulations.
- Observation-based Wetland Data (4 datasets for calibration):
- GIEMS-2 (Global Inundation Extent from Multi-satellites version 2)
- RFW (Regularly Flooded Wetland map)
- WAD2M (Wetland Area and Dynamics for Methane Modeling)
- G2017 (pantropical wetland extent from an expert system model)
- Independent Datasets for Evaluation:
- JRC (Global Surface Water dataset)
- H2015 (Japanese Earth Resources Satellite L-band SAR imagery)
- BAWLD (Boreal-Arctic Wetland and Lake Dataset)
- GRACE (Gravity Recovery and Climate Experiment) for terrestrial water storage (TWS).
The study did not introduce new benchmark datasets but rather leveraged existing ones for calibration and validation.
Key Insights and Ideas
- Bridging the Temporal Gap: The most significant contribution is the creation of a long-term (1980-2020) monthly wetland dataset, extending beyond the typical satellite observation era. This allows for historical analysis and understanding of long-term trends.
- Ensemble Approach for Uncertainty: By using an ensemble of 28 simulations derived from multiple soil moisture datasets, the study provides a valuable assessment of the uncertainty associated with wetland extent estimates.
- TOPMODEL's Adaptability: The successful application of the TOPMODEL framework, originally designed for hydrological modeling, to simulate wetland dynamics highlights its versatility and potential for broader use in Earth system science.
- Calibration Data Matters: The study emphasizes that the choice of calibration wetland data significantly influences the resulting simulated wetland extents, underscoring the importance of understanding the definitions and limitations of different observational datasets.
Key Results
- The study produced an ensemble of 28 gridded maps of monthly global/regional wetland extents at 0.25° × 0.25° spatial resolution for the period 1980–2020.
- The generated wetland maps effectively capture the spatial distributions, seasonal cycles, and interannual variabilities of observed wetland extent.
- The simulated wetland extents showed good agreement with independent satellite-based terrestrial water storage (TWS) estimates.
- The dataset is particularly reliable for mid-low latitudes, with more uncertainty noted in high latitudes due to the exclusion of frozen soil layers in the water table depth calculation.
Limitations and Future Work
- High Latitudes: The exclusion of frozen soil layers in the water table depth calculation introduces potential uncertainties in wetland estimation at high latitudes. The authors suggest that this could lead to an overestimation of wetland fraction in these regions.
- Model Simplifications: The use of a simplified TOPMODEL approach and the reliance on soil moisture reanalysis data inherently carry uncertainties. The paper acknowledges that the model might not fully capture complex lateral flow and overland flow dynamics.
- Calibration Data Influence: The study highlights that differences in calibration wetland data lead to substantial variations in the simulated wetland extents, suggesting that users should carefully consider the wetland definitions of the chosen product.
- Future Work: The authors suggest that the dataset can be used for various applications, including analyzing wetland-related methane emissions. The ability to provide real-time updates from ongoing soil moisture data also opens avenues for near-real-time monitoring. Further refinement of the model to better incorporate frozen soil dynamics and explore more advanced hydrological modeling techniques could improve accuracy, especially in polar regions.
Summary updated