The Arctic has undergone dramatic changes in recent decades, warming at a rate nearly four times faster than the global average, a phenomenon known as Arctic amplification (AA). Among all the poleward energy transport components contributing to AA, atmospheric moisture transport is especially effective in inducing Arctic warming. It has long been known that atmospheric rivers (ARs), long and narrow corridors of intense moisture transport in the atmosphere, are responsible for most of the poleward atmospheric moisture transport over mid-latitudes, suggesting their potential contribution to AA. Arctic ARs can also trigger a wide range of weather extremes. Based on hourly reanalysis dataset, we first characterize wintertime high Arctic warm extremes during 1980-2021. Atmospheric blocking patterns over the northern Eurasia are identified as a key ingredient in driving these events, as they can effectively deflect the eastward propagating cyclones poleward, facilitating intense moisture and heat intrusions into the high Arctic. Using an AR detection algorithm modified for high latitude, the characteristics of Arctic ARs and their role in driving these warm events are explicitly quantified. ARs are identified as the direct driver for most of these historical Arctic warm extreme events. Given the importance of ARs in driving the Arctic extreme weather, we also investigate the Arctic AR trends in recent decades. Using reanalysis data and CMIP6 large ensemble simulations, we show that the observed Arctic AR frequency has increased twice as much over the Atlantic sector than the Pacific sector. In contrast, the Arctic AR trends in CMIP6 models, mostly externally forced, are spatially more uniform. This discrepancy between the observed and the simulated trends can be explained by the phase shift of the Interdecadal Pacific Oscillation (IPO) and Atlantic Multidecadal Oscillation (AMO) in recent decades. More specifically, the observed negative phase shift of IPO and positive phase shift of AMO both favor the increase of ARs over the Atlantic sector and the reduction over the Pacific sector. We will also demonstrate how the E3SM Arctic regionally refined meshes impact the model performance in simulating historical Arctic ARs.