Eungyu Park

Language-model agents can now set up and run numerical simulations from a sentence, but a reported number can be wrong in two ways: the simulator never produced it, or it came from a model that was set up badly. We built GeoAI-THM Studio, an agent harness for OpenGeoSys 6.5.8 in which the language model edits only a typed model tree through 21 tools, a configuration is computed only after its verification cases pass (27 cases; 21 within 1 % of their exact solutions, official benchmarks or consistency checks, all within 2.5 %), and an output guardrail rejects answers whose checked numbers cannot be traced to the run, the model inputs or the request. In 16 runs with gpt-6-luna and gpt-6.1-terra on eight requests in heat, groundwater flow, deformation and solute transport, every number the guardrail checks in every answer was traced in this way, and Claude Haiku 5.5, standing in for a small model during development, passed 20 of 21 test scenarios, from which the eight requests were drawn. Yet both OpenAI models answered how much a 300 W point heat source raises pore pressure wrongly, −0.507 and 0.635 MPa against 0.967 MPa from the exact solution, because both had coarsened the mesh at the source, and gpt-6-luna also around it. gpt-6.1-terra had itself found that halving the elements changed its value by 0.257 MPa and called the result mesh-sensitive, but still reported it, and the scoring script, which checks procedure and provenance, passed both answers. Recomputing every answer on the mesh of its starting example exposed the errors. A sweep of the element size at the source showed the pressure within 3 % of the exact value only when the elements there are no larger than one eighth of the distance heat has spread by the first output, and the released harness refuses coarser meshes. We propose, as a hypothesis for further tests, that each modeling choice left to the language model needs a physics-based check of this kind.

Groundwater-level forecasting with recurrent networks presupposes a multi-year record and a model trained per well. Pretrained time-series foundation models forecast without training on the well, from its record and the expected weather. Three wells of the Korean national monitoring network with twenty-year records were forecast ten days ahead throughout 2015–2024, each block re-anchored on the last observation, under one protocol for TimesFM-3, Chronos-2, Moirai-2 and per-well GRU, LSTM, gradient-boosted, feed-forward and linear models, using 2 observed weather as the forecast. Trained on ten years, tuned recurrent networks and zero-shot TimesFM-3 halve the error of persistence, TimesFM-3 being ahead on two wells and level on one; simpler models trail. The level record alone reduces the persistence error by 20–30 %, past weather adds little, and the forecast-day weather removes a further 38–44 %; a causal-window ablation ties this skill to rainfall up to the forecast day. The foundation models, alike on level alone, separate with the weather: TimesFM-3 improves by 39–49 %, Chronos-2 by 2–11 % and Moirai-2 deteriorates. With a one-year record and fixed weights, TimesFM-3 retains 86–94 % of its accuracy and the per-well networks 57–84 %. On 48 quality-filtered network wells, one fixed TimesFM-3 configuration beats persistence everywhere (median error ratio 0.54) and a fixeddesign GRU and gradient boosting on 48 of 48. For short-horizon forecasting, site-specific learning ceases to be the principal source of skill once a pretrained model can translate known future forcing into state change; in multi-year simulation the fitted network remains the better integrator.

The construction of the Three Gorges Dam (TGD) has profoundly altered the groundwater cycle downstream. The obscure spatiotemporal patterns of exchange fluxes between the Yangtze River and groundwater hinder the resolution of water resources and environmental issues in the watershed. This study investigated the spatial extent of the Yangtze River's influence on adjacent groundwater in the Four-Lake Basin, the first river-lake wetland plain downstream of the TGD, using multiple clusters of monitoring wells installed along the river. A coupled SWAT-MODFLOW model was applied to quantify period-specific surface water-groundwater exchanges. A counterfactual scenario without TGD operation, holding other conditions constant is also simulated for comparison. The results show: (1) Under the combined influence of hydrogeological conditions and distance from the TGD, the influence range of the Yangtze River on confined groundwater is significantly greater in the upper section than in the lower section of the Four-Lake Basin, with a difference of approximately one order of magnitude. (2) River and groundwater exchanges exhibit pronounced seasonal and spatial characteristics: river-to-aquifer recharge dominates during drawdown and flooding periods, while aquifer-to-river discharge dominates during impounding and dry periods. Additionally, the interaction rates between river and aquifer are consistently higher in the upper section than in the lower one. (3) Relative to natural conditions, TGD operation dramatically dampens Yangtze River-groundwater interactions overall. The effect is most pronounced during the dry period in the upper section, when the interaction rate decreases by 40.6 %. These research outcomes serve as a vital theoretical foundation for assessing the effects of the TGD's regulation on the hydrological interaction in the riparian zone of the Yangtze River.

The construction of the Three Gorges Dam (TGD) has profoundly altered the groundwater cycle downstream. The obscure spatiotemporal patterns of exchange fluxes between the Yangtze River and groundwater hinder the resolution of water resources and environmental issues in the watershed. In the Four-Lake Basin, the first river-lake wetland plain downstream of the TGD, this study investigated the spatial extent of the Yangtze River's influence on adjacent groundwater by leveraging multiple groups of monitoring wells installed along the river. A coupled SWAT-MODFLOW model was applied to quantify period-specific SW-GW exchanges. A counterfactual scenario without TGD operation-holding other conditions constant is also simulated for comparison. The results show: (1) The influence range of the Yangtze River on confined groundwater is larger in the ZJ-JLX2 section, whereas it is relatively minor on groundwater near HH1 profile and HH2 profile. The influence distance at the HH1 profile is the smallest, measuring as 1.94 km. (2) River and groundwater exchanges exhibit pronounced seasonal and spatial characteristics: river-to-aquifer recharge dominates during drawdown and flooding periods, while aquifer-to-river discharge dominates during impounding and dry periods. Using JLX2 as a divider, interaction rates are consistently higher in the upper section than in the lower one. (3) Relative to natural conditions, TGD operation dramatically dampens Yangtze River-groundwater interactions overall. The effect is most pronounced during the dry period in the upper section, when the interaction rate decreases by 40.6 %. These research outcomes serve as a vital theoretical foundation for assessing the effects of the Three Gorges Dam's regulation on the regional water cycle.

The construction of the Three Gorges Dam (TGD) has profoundly altered the groundwater cycle downstream. The obscure spatiotemporal patterns of exchange fluxes between the Yangtze River and groundwater hinder the resolution of water resources and environmental issues in the watershed. In the Four-Lake Basin, the first river-lake wetland plain downstream of the TGD, this study investigated the spatial extent of the Yangtze River's influence on adjacent groundwater by leveraging multiple groups of monitoring wells installed along the river. A coupled SWAT-MODFLOW model was applied to quantify period-specific SW-GW exchanges. A counterfactual scenario without TGD operation-holding other conditions constant is also simulated for comparison. The results show: (1) The influence range of the Yangtze River on confined groundwater is larger in the ZJ-JLX2 section, whereas it is relatively minor on groundwater near HH1 profile and HH2 profile. The influence distance at the HH1 profile is the smallest, measuring as 1.94 km. (2) River and groundwater exchanges exhibit pronounced seasonal and spatial characteristics: river-to-aquifer recharge dominates during drawdown and flooding periods, while aquifer-to-river discharge dominates during impounding and dry periods. Using JLX2 as a divider, interaction rates are consistently higher in the upper section than in the lower one. (3) Relative to natural conditions, TGD operation dramatically dampens Yangtze River-groundwater interactions overall. The effect is most pronounced during the dry period in the upper section, when the interaction rate decreases by 40.6 %. These research outcomes serve as a vital theoretical foundation for assessing the effects of the Three Gorges Dam's regulation on the regional water cycle.

The construction of the Three Gorges Dam (TGD) has profoundly altered the groundwater cycle downstream. The obscure spatiotemporal patterns of exchange fluxes between the Yangtze River and groundwater hinder the resolution of water resources and environmental issues in the watershed. In the Four-Lake Basin, the first river-lake wetland plain downstream of the TGD, this study investigated the spatial extent of the Yangtze River's influence on adjacent groundwater by leveraging multiple groups of monitoring wells installed along the river. A coupled SWAT-MODFLOW model was applied to quantify period-specific SW-GW exchanges. A counterfactual scenario without TGD operation-holding other conditions constant is also simulated for comparison. The results show: (1) The influence range of the Yangtze River on confined groundwater is larger in the ZJ-JLX2 section, whereas it is relatively minor on groundwater near HH1 profile and HH2 profile. The influence distance at the HH1 profile is the smallest, measuring as 1.94 km. (2) River and groundwater exchanges exhibit pronounced seasonal and spatial characteristics: river-to-aquifer recharge dominates during drawdown and flooding periods, while aquifer-to-river discharge dominates during impounding and dry periods. Using JLX2 as a divider, interaction rates are consistently higher in the upper section than in the lower one. (3) Relative to natural conditions, TGD operation dramatically dampens Yangtze River-groundwater interactions overall. The effect is most pronounced during the dry period in the upper section, when the interaction rate decreases by 40.6 %. These research outcomes serve as a vital theoretical foundation for assessing the effects of the Three Gorges Dam's regulation on the regional water cycle.

The construction of the Three Gorges Dam (TGD) has profoundly altered the groundwater cycle downstream. The obscure spatiotemporal patterns of exchange fluxes between the Yangtze River and groundwater hinder the resolution of water resources and environmental issues in the watershed. In the Four-Lake Basin, the first river-lake wetland plain downstream of the TGD, this study investigated the spatial extent of the Yangtze River's influence on adjacent groundwater by leveraging multiple groups of monitoring wells installed along the river. A coupled SWAT-MODFLOW model was applied to quantify period-specific SW-GW exchanges. A counterfactual scenario without TGD operation-holding other conditions constant is also simulated for comparison. The results show: (1) The influence range of the Yangtze River on confined groundwater is larger in the ZJ-JLX2 section, whereas it is relatively minor on groundwater near HH1 profile and HH2 profile. The influence distance at the HH1 profile is the smallest, measuring as 1.94 km. (2) River and groundwater exchanges exhibit pronounced seasonal and spatial characteristics: river-to-aquifer recharge dominates during drawdown and flooding periods, while aquifer-to-river discharge dominates during impounding and dry periods. Using JLX2 as a divider, interaction rates are consistently higher in the upper section than in the lower one. (3) Relative to natural conditions, TGD operation dramatically dampens Yangtze River-groundwater interactions overall. The effect is most pronounced during the dry period in the upper section, when the interaction rate decreases by 40.6 %. These research outcomes serve as a vital theoretical foundation for assessing the effects of the Three Gorges Dam's regulation on the regional water cycle.

The construction of the Three Gorges Dam (TGD) has profoundly altered the groundwater cycle downstream. The obscure spatiotemporal patterns of exchange fluxes between the Yangtze River and groundwater hinder the resolution of water resources and environmental issues in the watershed. In the Four-Lake Basin, the first river-lake wetland plain downstream of the TGD, this study investigated the spatial extent of the Yangtze River's influence on adjacent groundwater by leveraging multiple groups of monitoring wells installed along the river. A coupled SWAT-MODFLOW model was applied to quantify period-specific SW-GW exchanges. A counterfactual scenario without TGD operation-holding other conditions constant is also simulated for comparison. The results show: (1) The influence range of the Yangtze River on confined groundwater is larger in the ZJ-JLX2 section, whereas it is relatively minor on groundwater near HH1 profile and HH2 profile. The influence distance at the HH1 profile is the smallest, measuring as 1.94 km. (2) River and groundwater exchanges exhibit pronounced seasonal and spatial characteristics: river-to-aquifer recharge dominates during drawdown and flooding periods, while aquifer-to-river discharge dominates during impounding and dry periods. Using JLX2 as a divider, interaction rates are consistently higher in the upper section than in the lower one. (3) Relative to natural conditions, TGD operation dramatically dampens Yangtze River-groundwater interactions overall. The effect is most pronounced during the dry period in the upper section, when the interaction rate decreases by 40.6 %. These research outcomes serve as a vital theoretical foundation for assessing the effects of the Three Gorges Dam's regulation on the regional water cycle.