Response of atmospheric rivers to aerosols treatment in regional climate simulations

Atmospheric rivers (ARs) play an essential role in extreme precipitation phenomena. To predict
such events, a correct simulation of ARs becomes crucial. Since most of the regional climate models
do not take aerosols into account in an interactive way, the main objective pursued in this work was to
analyse the role of aerosols in the intensity and behaviour of ARs on the regional scale. The identifi-
cation of ARs has always been carried out in global climate simulations applying detection algorithms
that may not be suitable in regional climate models, due to the presence of boundaries in the spatial
domain.

This work presents a new ARs identification algorithm for regional climate simulations (AIRA).
The implemented algorithm has proved to be able to properly identify the vapour structures associated
with ARs. AIRA was applied to a set of hourly data from three regional simulations (BASE, ARI and
ARCI), covering a period of 20 years. In BASE, aerosols were prescribed, while the model incorporates
aerosols dynamically in both ARI and ARCI. In ARI, aerosols are only incorporated interactively in
aerosol-radiation interactions. In ARCI, they are also included in the microphysical processes.

AIRA has identified about 250 ARs in the three simulations. Spring and autumn ARs were the
most frequent, intense and long-lasting, while they were less frequent, shorter and weaker in summer.
The identified ARs explain up to a 30% of the total precipitation in some areas of the Iberian Penin-
sula. The differences between the three simulations are significant in the spatial distribution of the
precipitation and in the trajectory and intensity of some ARs. Although the number of detected ARs
is similar, the temporal steps with ARs common to the three simulations represent only a 37% of the
total BASE steps containing ARs. This indicates that the sensitivity to the inclusion of aerosols is
relevant. The common ARs events showed that the BASE and ARI simulations generally present sim-
ilar trajectoriesk. However, important differences appear regarding ARCI, specially
when ARs are not quite intense.

A cluster analysis of the thickness field between 1000 and 850 hPa in ARI identifies three main
patterns. The comparison between the centroids in ARI and ARCI, reveals that the differences between
ARs in both simulations are mainly related to the aerosols type and concentration. The main
mechanism behind this behaviour is related to the modification of the temperature field due
to aerosol-cloud interactions (indirect effect) while aerosol-radiation effects are less relevant.