Atmospheric rivers (ARs) are filaments of enhanced water vapor transport in the atmosphere. Globally, ARs play a key role in the meridional moisture transport. Regionally,ARs can either serve as freshwater suppliers or culprits behind many of the weather hazards.
ARs and their impacts have been studied extensively. However, what controls the variability of ARs on different time scales remains largely unknown. In particular, further studies
are especially needed to better understand the relative role of circulation (dynamic) variability versus moisture (thermodynamic) variability, internal variability versus sea surface
temperature (SST)/ sea ice variability and anthropogenic forcing versus internal variability
originated from SST/sea ice variability in controlling the variability of ARs and their associated precipitation on different time scales. In addition, reanalyses have long been used as
proxies of observations in AR studies. Yet, the representation of ARs and their associated
precipitation in reanalyses remain unknown. Although satellite observations have also been
used to study ARs, previous satellite-based AR studies used only the moisture component
(integrated water vapor or IWV) to detect ARs. While ARs have been defined as filaments
of enhanced moisture transport in the atmosphere, detecting ARs with only the moisture
field would inevitably run the risk of detecting those filamentary features with high moisture
content, but relatively weak transport component.
In this dissertation, we will address the research gaps above from five different angles.
First, we investigate the relative role of SST/sea ice versus internal variability in driving the
interannual variability of winter AR activities over the North Hemisphere. We show that,
while both SST/sea ice and internal variability play roles in driving the interannual AR variability, their roles differ across ocean basins. Over the North Pacific, SST/sea ice variability
exerts substantially stronger control on the AR variability compared to interval variability.
However, both SST/sea ice and internal variability play comparable roles in modulating the
AR variability over the North Atlantic. Second, on longer time scale, we discover that ARs
over the Southern Hemisphere have been shifting poleward in the past four decades. Using
a simple scaling method, we find that this poleward shift in the ARs is mostly driven by the
poleward shift of the westerly jet (dynamic) while the contribution from the changes in the
moisture field is relatively minor. Using two ensembles from the Community Earth System
Model (CESM), one with fully coupled oceans and another one driven by observed SST/sea
ice, we show that anthropogenic forcing is mainly responsible for the observed poleward shift.
However, the negative phase of Interdecadal Pacific Oscillation (IPO) in recent decades also
further drives the poleward shift in ARs. Third, ARs are expected to change under a warmer
climate. However, the AR response to warming is determined by numerous factors. Two of
the most prominent factors are the warming of the tropical upper troposphere, which can
drive the poleward shift of the westerly jet, and the amplified warming of the polar regions,
which can drive the equatorward shift of the westerly jet. Using nine models participated in
the Polar Amplification Model Intercomparison Project (PAMIP), we investigate how Arctic
Amplification and its associated sea ice loss would affect the boreal winter AR activities
over the Northern Hemisphere. We find that, in response to Arctic sea ice loss, ARs extend
northeastward over North Pacific and shift equatorward over North Atlantic. We further
demonstrate that these response patterns are mostly determined by the responses in the
circulation. Fourth, using a moisture budget approach, the relative contribution of dynamic
change versus thermodynamic change to the intensification of extreme precipitation along the North American West Coast (predominantly driven by ARs) is also quantified. We show
the intensification of the extreme precipitation along the North American West Coast is
mostly driven by the increase in moisture while the contribution from the dynamic change
is minor. Lastly, we develop a novel method to detect ARs in satellite observations using
both IWV and wind information based on the geostrophic winds. Using this method, we
create satellite-based AR statistics and use these statistics to evaluate the performance of
seven commonly used reanalyses in representing ARs and their associated precipitation. We
show that both satellite observation and reanalyses show high agreement with each other in
representing the AR frequency distribution. In terms of AR precipitation, ARs in reanalyses
tend to precipitate too lightly and too often.
Our studies shed light on the mechanisms driving the variability of ARs across different
time scales. These findings have important implications. First, given the more important
role SST/sea ice variability in controlling the AR interannual variability, ARs in the North
Pacific are likely more predictable than those over the North Atlantic. Better SST forecast
can thus likely lead to better AR forecast along the North American West Coast. Second,
on the interdecadal time scale, internal variability related to ocean processes still plays an
important role in modulating AR variability. Internal variability should thus be taken into
consideration when studying future AR response under warming. Third, our results indicate
that the thermodynamic aspect of the AR and AR precipitation response is quite robust. We
thus need to constrain the response in circulation to reduce the AR response uncertainty.
Lastly, the finding that ARs in reanalyses tend to precipitate too often and too lightly
directly questions the use of reanalysis-based precipitation in AR studies.