Novel validation and performance assessment framework for atmospheric moisture tracking approaches: the case of UTrack

Abstract. Moisture tracking models are the primary tools for investigating atmospheric water vapour transport, the provenance of moisture fuelling extreme precipitation events, the hydrological impacts of land-use changes, and for supporting water isotopic composition analyses. Despite the numerous models and moisture identification schemes, their results often exhibit substantial disagreement. Because the exact evaporative sources of precipitation cannot be measured directly, there is a critical need for shared validation and performance assessment methodologies. Here, we propose a novel validation framework for evaluating estimated moisture sources and sinks against reference reanalysis data. We demonstrate our validation approach through its application to the Lagrangian atmospheric moisture tracking model UTrack. The global simulations performed show that UTrack achieves very high spatial accuracy in reconstructing monthly and instantaneous evaporative sources, precipitation sinks, and total atmospheric moisture content. Crucially, our framework elucidates performance asymmetries between forward and backward tracking modes, showing how the latter is inherently more vulnerable to trajectories errors, due to the localized and intermittent nature of input precipitation. Furthermore, validation experiments with different model configurations reveal that sub-grid vertical mixing parameterizations and the representation of vertical velocity heavily affect the accuracy of the diagnosed moisture patterns. In contrast, within-grid spatial interpolation, trajectory advection time step, upper-atmospheric vertical resolution, and the inclusion of condensed hydrometeors exert a relatively minor influence at the global scale, revealing significant opportunities for computational optimization. Alongside this validation and model-sensitivity analysis, we introduce an updated, open-source Python version of UTrack, designed for computational efficiency and broader scientific applicability. Ultimately, the validation framework and the results presented in this study establish a rigorous foundation for future model development, parameter calibration, and standardized intercomparison projects across the moisture tracking community.