Revisiting Land-Atmosphere Coupling Across Spatial Scales: From Coarse to Kilometer-Scale Simulations

Soil moisture–precipitation (SM–P) coupling is a key component of land–atmosphere interactions, but its strength and sign remain highly uncertain in large-scale models. While high-resolution models that explicitly resolve convection offer a way to reduce these uncertainties, their impact on SM–P coupling is not yet fully understood. Here, we investigate global SM–P coupling across different spatial resolutions using the Nonhydrostatic Icosahedral Atmospheric Model (NICAM). We find that SM–P coupling strongly depends on model resolution. As resolution increases, precipitation becomes more localized, leading to a smaller rainy area and a more heterogeneous spatial structure of the coupling. These changes involve significant regional variations in both coupling strength and sign. At high resolution, coupling is strengthened in major land–atmosphere hotspots, driven by enhanced convection that produces higher precipitation and more active moisture exchange. Meanwhile, high-resolution simulations exhibit widespread sign reversals in SM–P coupling. These reversals are caused by the convection-driven redistribution of precipitation, where localized moisture convergence and divergence reshape the coupling relationships. Compared to FLUXNET and ERA5 data, increasing model resolution systematically reduces negative biases in SM–P coupling, bringing the simulation closer to observations. Our results show that high-resolution modeling helps reconcile simulations with observations and emphasize the importance of using high-resolution frameworks to represent land–atmosphere interactions accurately.