D02
Simulating past and future responses of the terrestrial system to greenhouse gas forcing and regional anthropogenic interventions
Prof. Stefan Kollet Ph.D
Forschungszentrum Jülich | +49 2461 619593 |
Jun. Prof. Dr. techn. Michael Schindelegger
University of Bonn | +49 228 73-6345 |
Summary
In project D02, we explore whether—and to what extent—anthropogenic interventions can lead to persistent, potentially unsustainable changes in the coupled water and energy cycles of the surface/sub-surface-atmospheric system. We address these fundamental questions by performing regional forward simulations of the terrestrial system from groundwater to the top of the atmosphere under consideration of water and energy balance constraints over the European continent and the adjacent ocean. A carefully designed hierarchy of experiments, both for historical and 21st century time lines, will facilitate the identification and spatial localization of the anthropogenic impact against the background of natural variability and greenhouse gas forcing.
Graphical summary
Figure 1: Graphical summary of project D02
Contribution to the CRC
- Probing the central hypothesis of the CRC: Do anthropogenic modifications alter regional feedbacks in the terrestrial system and lead to persistent changes in the coupled water and energy cycles of the surface/sub-surface-atmospheric system?
- Identifying possible unsustainable trajectories with regard to freshwater resources and agriculture on regional scales
- Constraining signals and uncertainties in the modeled water cycle associated with moisture input from the ocean
- Providing results from historical simulations and projections to the attribution and budget analyses of other projects in the CRC
Approach
- Terrestrial System: Use of the Terrestrial System Modelling Platform, TSMP, which closes budgets across model components: Sub-surface, land surface, and atmosphere, as represented by ParFlow, CLM (Community Land Model, in different versions), and ICON or COSMO, respectively
- TSMP simulations over EURO-CORDEX slightly extended to the east, starting from control runs and then considering relevant perturbations: Greenhouse gas forcing, land use change, human water use such as groundwater abstraction and irrigation
- Simulations resolve pertinent variabilities and non-linearities in states, fluxes, and feedbacks, on time scales relevant to events and trends
- Analysis of the simulations focuses on how land-atmosphere conditions and feedbacks vary with each individual system perturbation; both linear and non-linear analyses methods are being pursued for feedback identification
- Ocean: Stand-alone (i.e., uncoupled) ICON simulations and water budget analyses to assess how biases and uncertainties in sea surface temperature (SST) impact the modeled regional atmospheric water cycle
- Using these simulations, we explore approaches (e.g., calibration of SST fields) to avoid excess oceanic evaporation and resulting excess precipitation over land
- Two-way coupling of ICON, and ultimately TSMP, to a global ocean model
Main Results in 2022 and 2023 – Terrestrial system
Figure 2. First row: Time-mean differences between HWU and CTRL runs in terms of 2-m specific humidity (left) and surface water availability (P–ET, right); Second row: Trend differences between HWU and CTL runs for the same two quantities. Red circles point out the intensively irrigated regions.
We performed a set of historical and projection simulations with TSMP at 12-km spacing, as well as a reference climatology run (1979–2021) driven by ERA5 reanalysis, one time without human water use effects (CTL run), and then with irrigation and groundwater abstraction (HWU run). We use the reference climatology run to identify the direct impact of human water use on the water cycle. Differences between HWU and CTL (Figure 2, first row) suggest a significant increase in 2-meter surface specific humidity and a decrease in water availability (precipitation P minus evaporation ET) due to enhanced ET in some irrigation hotspots, but no remote effects can be detected. In the second row in Figure 2, there are no clear changes in the trend between HWU and CTL runs in intensively-irrigated regions, indicating that the long-term trends in surface water content and water availability are mostly attributable to greenhouse gas forcing at large scale rather than the regional intervention.
Figure 3. Pearson correlation in the reference climatology run between (first row) latent and sensible heat flux, and (second row) ET and cloud water content. Columns from left to right indicate correlations from the full dataset, for cloudy days, and for rainy days.
We also analyzed the reference climatology run (1979–2021) to diagnose the land-atmosphere coupling strength over the EURO-CORDEX domain under different weather and moisture dynamic regimes. Emphasis is on feedbacks between surface flux, boundary layer processes, and cloud formation. We divide the dataset into three scenarios to test the impact of weather conditions: Full dataset (i.e., no temporal subsetting), cloudy days without precipitation, and rainy days. Figure 3 (first row) shows that the coupling between latent heat and sensible heat flux, representing the energy partition at the land surface, exhibits a north-south (positive-to-negative) dipole pattern in the full dataset. By contrast, the coupling tends to be more negative across the domain on cloudy days and very weak on rainy days. Maps of correlation between ET and cloud water content (Figure 3, second row) reveal a negative coupling between these two variables over the entire domain. The negative coupling is consistent with the strong control exerted by the atmosphere on the surface flux, but appears much less pronounced on cloudy days.
Main Results in 2022 and 2023 – Role of the ocean
We assembled multi-decadal SST fields from various sources (e.g., atmospheric reanalyses, remote sensing products) and analyzed their different spatiotemporal characteristics on a common 0.25° x 0.25° grid. A useful metric for comparison is the variance ratio of daily SST anomalies, relative to mean seasonal cycle, of one SST dataset over the other. Figure 4d shows that the variance ratio of MURSST (Multi-scale Ultra-high Resolution SST, an objectively interpolated satellite and in-situ product) vs. ERA5 for the period 2002–2021 exceeds 1.2 in many places, especially in proximity to landmasses. First ICON simulations at 12-km spacing have been conducted to examine possible impacts of these differences, especially how the choice of SST lower boundary condition in ICON affects the region’s modeled atmospheric water budget. Figures 4a and 4b illustrate the total amount of ET minus precipitation (P) over six simulation months when using either ERA5-based SST or MURSST. The differences in ET–P from the two simulations are appreciable (±100 Kg/m² in six months, Figure 4c) and highlight that accurate representation of P of oceanic origin is an important issue when analyzing Europe’s hydroclimate for changes due to anthropogenic effects.
Figure 4. Total amount of ET–P over the period June to December 2002, as simulated with ICON on a 12-km grid using either (a) ERA5 SST or (b) MURSST as lower boundary condition. Panel (c) depicts the difference (a)–(b). Daily SST anomalies from MURSST have a systematically higher variance ratio when compared to ERA5 SST anomalies (see panel d).
