HydroBlocks-MSSUBv0.1: a multiscale approach for simulating lateral subsurface flow dynamics in Land Surface Models

Abstract. Groundwater is critical in the hydrological cycle, impacting water supply, agriculture, and climate regulation. However, current Land Surface Models (LSMs) often struggle to accurately represent the multiple spatial scales of subsurface flow primarily due to the complexity of incorporating sufficient and yet efficiently surface heterogeneity, which significantly influences subsurface dynamics. Accurately modeling this heterogeneity requires substantial computational resources, often making it challenging to achieve in practice. This study introduces a multiscale approach to address this limitation. The approach leverages the hierarchical clustering scheme of the HydroBlocks model to define hydrologically similar areas that the model uses to capture local, intermediate, and regional flow dynamics within regional units, which interact laterally based on hydraulic gradients and soil properties. The proposed method is compared against a benchmark simulation with 1.4 million modeling units – 34 times the number of tiles in the multiscale experiment. The results show consistency in spatial distribution and a Pearson coefficient of correlation above 0.80 for the temporal variability of hydrological variables such as latent and sensible heat flux, surface runoff, and effective saturation at the root zone, demonstrating its ability to represent subsurface flow patterns adequately. The scheme, however, struggles to adequately represent volumetric water content at the bottom of the soil column, as evidenced by lower correlation coefficients, where the misrepresentation of elevation heterogeneity may play a larger role. This multiscale approach offers a computationally efficient way to incorporate detailed subsurface processes into large-scale hydrological simulations, improving our understanding of water cycle dynamics and supporting informed water resource management.