2.3 Atmospheric River Detection

2.3.1 Automated global algorithm
We utilize the Guan and Waliser (2019), GW19, global AR detection algorithm which first specifies a threshold of (vertically) Integrated Vapor Transport (IVT), and then identifies ARs from the subset of IVT that conforms to a 3:1 length to width. Here, we use the 85th% IVT for 2018-2022, during which changes in baseline temperature (and IVT) are minimal (Guan et al, 2023). GW19 is one of multiple automated global AR detection algorithms, which do not all agree (ARTMIP: Shields et al., 2018). Algorithm choice can introduce uncertainty on AR seasonality and changes over time (Payne et al., 2020).

2.3.2 Atmospheric River Genesis and Transport Distances

A strength of the GW19 algorithm is that it tracks ARs through their lifetime including the genesis timestep. We partition each grid box and time into time categories of all AR (AllAR), non-Atmospheric River (nonAR), and AR genesis (ARgenesis), defined as the appearance of new ARs that persist for at least 12 hours. These ARgenesis footprints are used in the VSD processing.
Each of the VSD tracers are vertically integrated such that tracer 1 (equivalent to q) is identical to IVT. In post processing, we calculate moisture provenance for each ARgenesis, and we calculate the attendant AllAR and nonAR moisture provenance (over each ARgenesis footprint) given a monthly climatological distribution of VSDs over 2018-2022. We calculate the centroid of the ARgenesis mask and the centroid of the moisture provenance for each ARgenesis, AllARs, and nonARs to determine source latitude and distance.