A key part of the global hydrological cycle is comprised of the moisture flows through the atmosphere, which connect locations where the moisture evaporates (upwind locations) with the locations where it subsequently precipitates (downwind locations). 

Moisture recycling, the phenomenon for which the moisture that origins from land vegetation re-preciptates over land, connects land- and atmospheric conditions up to thousands of kilometres away. Globally, moisture recycling is so important for global precipitation patterns that roughly half of all terrestrial precipitation has come from evapotranspiration from land, the other half being from the ocean.

Today, high attention is posed on the effects of land use and vegetation changes on moisture recycling. Indeed, land-use changes that affect evapotranspiration flows, such as deforestation, can affect precipitation regimes, the severity of droughts and hydrological flows in downwind regions.

Therefore, reconstructing evapotranspiration-to-precipitation connections has involved a broad range of researchers who has developed the so-called atmospheric moisture tracking models to  develop these vapour flows reconstructions.

These models typically use atmospheric reanalysis data to simulate the atmospheric branch of the hydrological cycle. However, despite the large interest, it is often not feasible for broad researchers to perform these simulations. As with all methods, becoming familiar with them requires significant time investment, but an additional constraint on widespread use of atmospheric moisture tracking is its heavy data demand. This data demand has increased considerably with the largest generation of atmospheric reanalysis data, ERA5, which allows for obtaining detailed global moisture flows. Models steadily advanced as cutting-edge data becomes increasingly available. This is the case with the UTrack Lagrangian model, which takes advantage of state-of-art climate reanalysis data and, by testing several combinations of model assumptions, optimally generates highly detailed evaporation footprints while avoiding unnecessary complexity. Despite these cutting-edge model advancements and the wide applications already achieved, less attention has been given to ensuring the consistency of tracked moisture volumes with reanalysis data of precipitation and evaporation simultaneously to ensure the closure of the annual hydrological cycle. Model error and assumptions, as well as possible discrepancies in ERA5 data, may lead to inconsistencies that could impede internally consistent descriptions of the global hydrological cycle.  Indeed, uncertainty related to a set of modelling assumptions and data resolution still poses an issue for the moisture tracking community.

To address this gap, our study -openly accessible at Scientific Data (https://doi.org/10.1038/s41597-025-04964-3)- proposes a reconciliation framework based on the Iterative Proportional Fitting (IPF) procedure, a rigorous mathematical framework for refining tracking model outputs, thereby reducing uncertainties arising from modeling assumptions and data resolution constraints.  The study further includes a pre-processing of the ERA5 reanalysis data to address the existing annual unbalance between ERA5 precipitation and evaporation. Overall, the entire reconciliation framework ensures that the total tracked atmospheric moisture matches the total precipitation at the sink and the total evaporation at the source on an annual basis and in each cell.

The outcome is a new dataset of moisture flow volumes from sources of evaporation to fates of precipitation at 0.5°, named RECON, with global coverage and centred over 2008-2017, which aligns coherently with annual precipitation and evaporation volumes from ERA5 reanalysis. The reconciled dataset is available at https://zenodo.org/records/14191920.

The reconciled cell-grid dataset provided within this study offers post-processed atmospheric moisture portions of evaporation at the source precipitating at the sink which closes the atmospheric hydrological balance on the annual basis. This marks a significant advancement in enhancing the reliability of the UTrack dataset and paves the way for future applications across multiple tracking models and forcing data.

A first application of the reconciliation approach is  a country-ocean and subcontinental analysis of transboundary atmospheric  water flows which reveals that 45% of total terrestrial precipitation  originates from land evaporation. The country-ocean and subcontinental datasets are available at 10.5281/zenodo.10400695 and published in Communications Earth & Environment (https://doi.org/10.1038/s43247-025-02289-y).