Global methane emissions from freshwater offsets 25% of terrestrial greenhouse gas sink (Bastviken et al., 2011). Global rivers contribute roughly the same as lakes to this emission (Stanely et al., 2016; Rocher-Ros et al., 2023). Global warming is set to exacerbate this further as warmer water leads to lower levels of dissolved O₂, reduced CO₂ capture, and increase in methane production and eutrophication. Understanding the thermal inflow from rivers to lakes and reservoirs is, therefore, essential to monitor (and forecast) the exceedance of emission critical values.

Large-scale hyper-resolution modeling allows for locally relevant analyses, a feature recently achieved for hydrological modeling at continental-scale and global-scale (Hoch et al,. 2023; van Jaarsveld et al., 2025). While global river and lake temperature models exist (van Vliet et al., 2011; Wanders et al., 2019), hyper-resolution modeling of river thermal content at the global scale is yet to be demonstrated.

Here, we analyze the changes in thermal inflows to surface water bodies (lakes and reservoirs) globally at 1 km based on the river temperature routing module implemented within the mesoscale hydrological model (mHM, https://mhm-ufz.org). Surface water temperature in rivers is modeled by balancing the heat exchange between the atmosphere and river water while accounting for energy sources from the sub-surface systems. The experimental setup is based on Shrestha et al. (under review), where 62 domains cover the global land-surface and the simulation period is set to 1961-2020. The setup includes major global lakes and reservoirs enlisted in the HydroLAKES and GRanD v1.3, respectively, the process representation of which follows Shrestha et al. (2024).

We have evaluated the model at 5,000 streamflow observations from the Global Runoff Data Centre and 500 river temperature observations from Global Environment Monitoring System. We have also carried out sensitivity analysis of water temperature simulations to the surface albedo, where space-time varying albedo is expected to result in a closer match to the observations, than with a constant albedo, besides analyzing the trends and clusters of water temperature and derived indicators. For reproducibility, the experiment backend is powered by ecFlow, the workflow management tool developed by ECMWF. Our modeling framework on the analysis of water and energy inflows, covering surface water systems – lakes and reservoirs – globally forms a basis for timely warning of critical events with high thermal inflows, and such systems could see far-reaching applications, e.g., insurance underwriting for the fisheries industry.