Distant Influence of Kuroshio Eddies on North Pacific Weather Patterns?

Abstract

High-resolution satellite measurements of surface winds and sea-surface temperature (SST) reveal strong coupling between meso-scale ocean eddies and near-surface atmospheric flow over eddy-rich oceanic regions, such as the Kuroshio and Gulf Stream, highlighting the importance of meso-scale oceanic features in forcing the atmospheric planetary boundary layer (PBL). Here, we present high-resolution regional climate modeling results, supported by observational analyses, demonstrating that meso-scale SST variability, largely confined in the Kuroshio-Oyashio confluence region (KOCR), can further exert a significant distant influence on winter rainfall variability along the U.S. Northern Pacific coast. The presence of meso-scale SST anomalies enhances the diabatic conversion of latent heat energy to transient eddy energy, intensifying winter cyclogenesis via moist baroclinic instability, which in turn leads to an equivalent barotropic downstream anticyclone anomaly with reduced rainfall. The finding points to the potential of improving forecasts of extratropical winter cyclones and storm systems and projections of their response to future climate change, which are known to have major social and economic impacts, by improving the representation of ocean eddy-atmosphere interaction in forecast and climate models.

Figure 1. Observed and Simulated rainfall changes in response to meso-scale SST forcing.

(A) Difference of TRMM winter season (NDJFM) mean rainfall (mmd⁻¹) between Inactive Eddy Years (IEYs) and Active Eddy Years (AEY). (B) TRMM daily rainfall PDF difference between IEYs and AEYs along the Kuroshio (magenta) and over the U.S. North Pacific coast (cyan). (C) Difference of winter season (NDJFM) mean total rainfall (mmd⁻¹) between two ensembles of 10 WRF simulations, MEFS (without ocean eddies) and CTRL (with ocean eddies). Rainfall difference significant at 95% confidence level based on a two-sided Wilcoxon rank sum test is shaded by gray dots. (D) Daily rainfall PDF difference between MEFS and CTRL along Kuroshio (magenta) and over the U.S. North Pacific coast (cyan). For both TRMM3B42 and simulated rainfiall, the PDF was derived by counting the number of rainy days in different rain rate ranges at each grid point and then averaging them in the region along the Kuroshio (denoted by the magenta box in (C)) and over the U.S. North Pacific coast (denoted by the cyan box in (C)). The maps were generated using M_Map V1.4 package for Matlab (http://www.eos.ubc.ca/~rich/map.html).

Figure 2. Satellite observed and model simulated frontal- and meso-scale air-sea interactions in KOCR.

(A) 2007/8 winter season mean (NDJFM) spatially highpass-filtered SST (contour with interval of 0.5 °C) and 10 m wind speed (color in ms⁻¹) derived from MW-IR and CCMP satellite observations (B) and from the ensemble mean of 27 km uncoupled WRF CTRL simulations, (C) highpass-filtered SST (contour with interval of 0.5 °C) and PBL (color in m), (D) highpass-filtered SST (contour with interval of 0.5 °C) and CAPE (color in Jkg⁻¹) derived from the ensemble mean of 27 km uncoupled WRF CTRL simulations. The maps were generated using M_Map V1.4 package for Matlab (http://www.eos.ubc.ca/~rich/map.html).

Figure 3. Upper atmospheric response to meso-scale SST forcing.

(A) Winter season (NDJFM) mean zonal wind U at 300 hpa, U300 (ms⁻¹), in CTRL. (B) Difference of U300 (ms⁻¹), (C) Sea-Level-Pressure (mb) (contour) and geo-potential height at 500 hpa (Z500) (m) (color) and (D) transient eddy kinetic energy at 300 hpa (m²s⁻²) between MEFS and CTRL (MEFS-CTRL). The transient kinetic energy was derived using 2–8 day bandpass-filtered variables. In (B–D), the difference significant at 95% confidence level based on a two-sided Wilcoxon rank sum test is shaded by gray dots. The maps were generated using M_Map V1.4 package for Matlab (http://www.eos.ubc.ca/~rich/map.html).

Figure 4. Simulated diabatic heating difference and synoptic baroclinic waves over the Kuroshio extension region during cyclone development periods in MEFS and CTRL.

(A) Vertically integrated (from 1000 to 300 hpa) storm-day diabatic heating (Pa•K/s) in CTRL. (B) Difference of vertically integrated (from 1000 to 300 hpa) storm-day diabatic heating (Pa•K/s) (MEFS-CTRL). The difference significant at 95% confidence level based on a two-sided Wilcoxon rank sum test is shaded by gray dots. (C) Vertical structure of geopotential height (contours, m) and diabatic heating (shaded, Pa•K/s) along the storm path (denoted by yellow dotted lines in Fig. S4) for all the developing synoptic (2–8 day) storms in CTRL, composited during the simulated storm days (see SI text and Fig. S5 for details). (D) Same as (C) but for MEFS. The storm days were defined based on a 20-percentile higher threshold criteria using a surface turbulent heat flux (THF) index derived by averaging simulated daily THF over a 30°x10° area [140°E–170°E, 32°N–42°N] in the respective simulations (SI text). The maps were generated using M_Map V1.4 package for Matlab (http://www.eos.ubc.ca/~rich/map.html).