The sparse matrix-based atmospheric moisture tracking (SMAT) model v1.0 for efficient diagnosis of global climatological moisture source–sink relationships

Abstract. Atmospheric moisture source–sink relationships characterize how moisture originating from evaporation in one region is transported through the atmosphere and ultimately contributes to precipitation in another. Quantifying these relationships is important for understanding moisture recycling, land–atmosphere interactions, and hydroclimatic teleconnections. However, existing moisture-tracking approaches remain computationally demanding for global and long-term climate applications. Here, we develop the sparse matrix-based atmospheric moisture tracking (SMAT) model that can efficiently diagnose global climatological moisture source–sink relationships. SMAT uses atmospheric moisture fluxes to describe moisture exchange among neighbouring grid cells under the local well-mixed assumption. These local transport relationships are assembled into a global sparse matrix, allowing atmospheric moisture redistribution to be solved directly without explicit step-by-step transport integration. This formulation substantially reduces the computational cost of global moisture source–sink diagnosis. Once the required transport fields have been constructed, the solution time depends primarily on spatial resolution rather than on the length of the analysis period. Global source–sink relationships can therefore be solved within hours or even minutes. We also evaluate SMAT against the multi-model-based global atmospheric moisture source–sink relationships (AMSSR) dataset and an additional Utrecht atmospheric moisture tracking (UTrack) dataset. Spatially, SMAT reproduces the major patterns of evaporation and precipitation recycling across grid, basin, and continental scales, together with the dominant land–ocean moisture source–sink pathways. Its deviations from AMSSR are generally comparable to, and in many cases smaller than, the corresponding UTrack–AMSSR deviations. Temporally, SMAT captures the climatological seasonal cycle of moisture recycling with performance comparable to that of UTrack. Overall, SMAT provides a computationally efficient framework for global atmospheric moisture source–sink diagnosis, particularly for high-resolution and large-ensemble climate applications.