Sorting Atmospheric Rivers in a Future Climate by Pattern

The antecedent environment of an atmospheric river (AR) determines its defining characteristics and the type of forecast challenges it may present. Prior work using neural networks and a customized, high-resolution 32 water-year reforecast (West-WRF) elucidated common synoptic-setups that were associated with specific forecast errors relating to the AR’s translation speed and integrated vapor intensity (IVT) intensity, along with the AR’s predictive skill. We also identified how certain ENSO and MJO states via defining upper-level wind and height patterns corresponded to these forecast errors (i.e. El Nino was associated with negative IVT intensity biases while La Nina was associated with positive IVT intensity biases; MJO phases 5 and 3 were associated with positive and negative translation speed biases, respectively). Now, training climate-forced ARs on historically-trained self-organizing maps (SOMs), we can see how past and present error patterns and their associated meteorology may change in a future environment. Additionally, we can identify and contrast how these flow patterns differ from present-day ARs, along with their subseasonal-to-seasonal influences, as each climate-forced AR is representative of a different ENSO state. While our climate-forced dataset (MPAS; Michaelis et. al 2019) and AR detection scheme (Guan and Waliser, 2015) differ from our prior research, enough similarities exist to draw meaningful linkages. Differences in future IVT intensity and average AR-centroid location per node are already evident. This work aims to illuminate further meteorological differences between present and future ARs.

References:

Guan, B. and Waliser, D.E., 2015. Detection of atmospheric rivers: Evaluation and application of an algorithm for global studies. Journal of Geophysical Research: Atmospheres, 120(24), Pp.12514-12535.

Michaelis, A. C., G. M. Lackmann, and W. A. Robinson, 2019: Evaluation of a unique approach to high-resolution climate modeling using the Model for Prediction Across Scales – Atmosphere (MPAS-A) version 5.1, Geosci. Model Dev., 12, 3725–3743.