Atmospheric River Climatology for the South Atlantic Basin Using an Image Processing‐Based Methodology

Abstract

Atmospheric Rivers (ARs) are narrow regions of extratropical water vapor transport, situated in the lower troposphere. They are responsible for a large proportion of poleward vapor transport from the tropics to the midlatitudes. Although ARs are known to cause devastating floods when encountering topography, they are a crucial source of water for many midlatitude regions. Most methods used for AR detection are based on magnitude thresholding of integrated vapor transport. Although magnitude‐thresholding approaches have been extremely beneficial in building the knowledge around ARs on a global scale, there remain some limitations with the methodology. Here, a new detection method derived from an image‐processing algorithm is applied to ERA5 data to detect ARs in the South Atlantic. A series of sensitivity tests were conducted to determine the best image‐processing parameters to detect the ARs. Consistent with the literature, a large spread in AR count and general AR characteristics are found across the different tests, highlighting that challenges remain with AR detection, even with new methods, particularly around the ambiguity in the shape criteria or geometric filtering. A full climatology for the South Atlantic over the period 1979–2023 is then developed. It was found that ARs occur all year round, with an average of 161 systems per year. They predominantly originate in the southwest Atlantic and then track in an eastward direction toward the Indian Ocean, with most occurring in the region of maximum storm activity between 40°S and 60°S, with occurrence peaking during austral summer (DJF). Less than 10% of these systems make landfall in South Africa.