Atmospheric rivers (ARs) are long, filamentary structures that transport large quantities of moisture northward, typically from the tropics and subtropics into the mid-latitudes. The goal of ARTMIP (Atmospheric River Tracking Method Intercomparison Project) is to understand and quantify the uncertainties in atmospheric river (AR) science due to the tracking or identification methodology alone. Of equal import is to understand the implications of these uncertainties. For example, is the identification method used to count ARs from observational data appropriate for model data, including future climate projections? Algorithms that use absolute moisture thresholds often give different metrics than those with relative thresholds. What are the implications for a metric such as the seasonality of AR frequency? An algorithm using an absolute threshold values for IVT (integrated vapor transport) of 250 kg/m/s may detect a significantly different number of ARs compared to a method that uses a relative threshold, (i.e. a threshold based on spatial or temporal anomalies) because of where the storm track lies. If an individual AR tracks closer to the climatological IVT maximum value, an absolute method will naturally reach that absolute threshold more often than a relative method, which bases its detection on an anomaly that may or may not be equal to the climatological value. ARTMIP is poised to help researchers make sense of the diverse and sometime opposing nature of metrics that currently exist in the literature, and connect those metrics (and algorithms) to physical mechanisms, and, ultimately climate controls.

ARTMIP also aims to 1) provide guidance to the community as to which algorithms are most appropriate for specific science questions, 2) release data to the community (i.e., the suite of catalogues specifying the AR tracking information from participating algorithms) across the various ARTMIP phases. ARTMIP’s first phase (Tier 1) includes the comparison of algorithms applied to MERRA-2 re-analysis. Analysis goals for Tier 1 include understanding AR metrics such as frequency and seasonality, intensity, duration, and precipitation attributable to ARs. The second phase, Tier 2, is focused on specific science questions on topics such as sensitivity to re-analysis datasets and climate change questions. Here, we present results and lessons learned from Tier 1 analysis focusing on the seasonality of ARs, the precipitation attributable to ARs, and an introduction to Tier 2, with preliminary results from algorithms applied to high resolution climate change simulations.