HTML
-
In this study, we use fifth-generation European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric reanalysis data (ERA5) (Copernicus Climate Change Service (C3S), 2017). ERA5 is produced using ECMWF’s Integrated Forecast System with a horizontal spatial resolution of 31 km and 137 hybrid sigma-pressure (model) levels in the vertical direction, up to a top level of 0.01 hPa. The pressure-level variables used are hourly geopotential, air temperature and horizontal winds on 2.5° × 2.5° and 1.5° × 1.5° latitude/longitude grids. The near-surface variables include hourly 2-m air temperature, 10-m horizontal winds and precipitation on a 1° longitude × 1° latitude grid. We employ vertically integrated eastward and westward water vapor fluxes on a 1° longitude × 1° latitude grid. We also use hourly potential temperature on the dynamical tropopause (2 PVU, where 1 PVU = 10−6 K m2 kg−1 s−1) with 2.5° × 2.5°, 2° × 2° and 1.5° × 1.5° latitude–longitude grids. The daily mean data are obtained by averaging the hourly reanalysis data. The period examined here is from 1 July to 31 August over the years 1979–2018.
The daily gridded precipitation data are from APHRODITE (Asian Precipitation—Highly Resolved Observational Data Integration Toward Evaluation of Water Resources) (Yatagai et al., 2012). This suite of precipitation data is derived based on rain gauge precipitation measurements in monsoonal Asia (15°S–55°N, 60°–150°E), the Middle East (15°–45°N, 20°–65°E) and Russia (34°–84°N, 15°–165°W), and provided on a 0.5° × 0.5° longitude–latitude grid. The data are available from 1951 to 2007 over three areas in version 1101 and are additionally extended to 2015 in version 1101 extend in monsoonal Asia.
-
Blocking events are identified on the basis of the reversal of potential temperature on the dynamic tropopause surface (e.g., 2 PVU) (Pelly and Hoskins, 2003), similar to Masato et al. (2012) and Shi et al. (2016). Unlike these studies, a low-pass filter with an 8-day cutoff period is applied to the potential temperature and pressure-level fields to remove the high-frequency disturbance. The reversal of potential temperature, denoted as B index, is calculated at each grid (λ0,
$\phi_0 $ ) within the 40°–70°N zonal belt for each day according towhere θ is the potential temperature at the dynamic tropopause and Δ
$\phi$ is 30°. If the B index is positive, a transient local blocking occurs. The local maximum positive grid of the B index is referred to as the central location of the blocking. A blocking event is identified if the following three criteria are met (Masato et al., 2013; Shi et al., 2016):(1) If there is a maximum positive B index at day n + 1 within a 36° × 27° box (longitude–latitude) centered at the blocking center at day n. Such a maximum is considered as a continuation of that blocking event; otherwise, it is considered as a new event.
(2) If there is no maximum positive B index within a box 1.5 times greater in both latitude and longitude than the box used in (1) centered at the blocking center on the onset day, it is considered to be the end of an event.
(3) The duration of the blocking event is at least four days.
Parallel blocking identifications are applied to 2.5° × 2.5°, 2° × 2° and 1.5° × 1.5° latitude–longitude grids. Blocking distributions are generally consistent with each other among the three different data resolutions, but with sharper features in the higher resolution data (not shown). Given that a blocking is a large-scale atmospheric circulation with a longitudinal scale between 12° (Tibaldi and Molteni, 1990) and 45° (Rex, 1950), it is sufficient to resolve the blocking occurrence using the 2.5° × 2.5° grid.
As seen from Figs. 1 and 2a, an increased blocking occurrence over central Siberia (45°–70°N, 75°–115°E) is highly related to the variability of areal-average precipitation over southeastern Lake Baikal (40°–55°N, 105°–125°E). Meanwhile, a blocking maximum extends from the Ural Mountains to Lake Baikal (Fig. 2b). Therefore, we primarily focus on blockings over central Siberia where the blocking center should be in this region on the peak day (i.e., maximum B index). A total of 60 blocking events are identified over central Siberia during July and August. Since blocking events identified in terms of B index exhibit various RWB features (Tyrlis and Hoskins, 2008; Masato et al., 2012, 2013), we use the direction of breaking (DB) index to separate blockings into AWB and CWB blocking groups. The daily DB index is calculated in terms of zonal gradients of the areal average θ around the blocking center, as derived by Masato et al. (2012) and expressed as
Figure 1. (a) Spatial distributions of linear trends of precipitation and (b) trends and time series of areal-averaged precipitation within a box encompassing (40°–55°N, 105°-125°E) based on APHRODITE data. Dark (Light) shading in (a) indicates statistical significance above the 95% (90%) confidence level. The red lines in (a) represent the linear regressions, and the blue dashed line is the Theil–Sen estimate of the linear trend. (c, d) As in (a, b), but for ERA5.
Figure 2. (a) Map of B index anomalies regressed against the standardized areal-averaged precipitation within the box in Figs. 1a and b. Dark (Light) shading indicates statistical significance above the 95% (90%) confidence level. (b) Distribution of RWB blocking centers in July and August (1979–2018). A nine-point smoothing is used and the polygon designates the region of central Siberia. (c, d) The distribution of blocking centers on the peak day for (c) AWB and (d) CWB blocking events over central Siberia.
where λc and
$\phi$ c represent the longitude and latitude of the blocking center, respectively, and Δλ is 2.5°. The instantaneous DB index is normalized by the standard deviation of all DB values when the B index is positive (e.g., during the blocking event). It has been noted that the orientation of a blocking event can change during its lifetime, particularly over Asia (Masato et al., 2012; Shi et al., 2016). Here, we use the temporal average of the DB index during each blocking event to represent the primary RWB feature of the blocking event. A blocking event is considered to be an AWB feature with an average DB index of no less than 0.2, while a CWB signature has an average DB index of no more than −0.2. Otherwise, it is defined as a neutral blocking. We adopt this threshold from Masato et al. (2012), who showed that the classification result is less sensitive to the threshold within a few decimals of variation. The numbers of AWB, CWB and neutral blockings are 18, 29 and 13, respectively. The present study mainly discusses the AWB and CWB blocking groups shown in Tables 1 and 2.NO. Onset date Peak date Duration (days) Latitude Longitude DB index 1 8 August 1979 8 August 1979 5 67.5°N 100°E 0.6 2 7 July 1981 11 July 1981 6 65°N 90°E 0.9 3 10 July 1982 13 July 1982 7 60°N 105°E 1.3 4 27 August 1984 28 August 1984 4 55°N 100°E 0.2 5 4 July 1987 6 July 1987 8 62.5°N 95°E 1.3 6 13 August 1988 15 August 1988 4 60°N 100°E 2.8 7 1 July 1989 3 July 1989 7 60°N 85°E 1.4 8 19 July 1989 20 July 1989 5 62.5°N 85°E 3.3 9 13 July 1990 14 July 1990 5 62.5°N 87.5°E 1.3 10 26 July 2002 28 July 2002 5 60°N 82.5°E 1.6 11 2 July 2004 5 July 2004 10 65°N 80°E 1.6 12 6 August 2006 6 August 2006 4 67.5°N 82.5°E 1.0 13 15 August 2007 17 August 2007 5 60°N 82.5°E 0.8 14 9 August 2008 10 August 2008 4 67.5°N 115°E 0.3 15 15 August 2009 17 August 2009 4 62.5°N 92.5°E 2.6 16 22 August 2009 24 August 2009 6 65°N 90°E 1.8 17 18 July 2012 21 July 2012 6 60°N 102.5°E 3.2 18 2 July 2014 4 July 2014 4 57.5°N 75°E 3.2 Table 1. Event dates and center positions on peak days along with the average DB index during each event for AWB blocking events over central Siberia.
NO. Onset date Peak date Duration (days) Latitude Longitude DB index 1 23 August 1979 23 August 1979 4 67.5°N 107.5°E −2.0 2 30 July 1980 31 July 1980 9 60°N 105°E −1.6 3 7 August 1981 10 August 1981 12 60°N 110°E −2.8 4 20 July 1982 23 July 1982 6 65°N 80°E −1.4 5 10 August 1982 11 August 1982 5 65°N 75°E −2.6 6 1 August 1983 5 August 1983 5 70°N 92.5°E −3.4 7 26 August 1983 28 August 1983 4 62.5°N 95°E −1.1 8 13 July 1984 17 July 1984 19 65°N 90°E −1.7 9 16 July 1985 18 July 1985 5 62.5°N 85°E −4.2 10 4 July 1986 9 July 1986 8 62.5°N 92.5°E −3.1 11 25 July 1987 27 July 1987 4 70°N 105°E −0.4 12 1 July 1988 1 July 1988 5 62.5°N 115°E −1.9 13 1 July 1991 13 July 1991 13 62.5°N 77.5°E −1.4 14 21 July 1992 22 July 1992 4 60°N 102.5°E −2.7 15 12 August 1992 17 August 1992 7 62.5°N 82.5°E −2.5 16 6 August 1993 8 August 1993 6 57.5°N 112.5°E −0.5 17 7 July 1994 19 July 1994 15 62.5°N 82.5°E −2.6 18 20 July 1995 21 July 1995 9 62.5°N 97.5°E −1.9 19 18 August 1998 20 August 1998 6 67.5°N 80°E −0.9 20 10 July 2000 12 July 2000 4 70°N 100°E −1.5 21 22 July 2000 25 July 2000 7 67.5°N 82.5°E −2.3 22 1 August 2000 4 August 2000 4 65°N 92.5°E −0.8 23 18 August 2000 20 August 2000 5 65°N 75°E −0.7 24 3 July 2001 5 July 2001 4 70°N 87.5°E −1.2 25 8 August 2003 10 August 2003 7 65°N 95°E −0.9 26 7 July 2006 9 July 2006 5 57.5°N 95°E −2.7 27 9 July 2008 13 July 2008 8 67.5°N 107.5°E −3.7 28 6 August 2011 9 August 2011 7 70°N 95°E −4.7 29 23 July 2014 26 July 2014 6 67.5°N 77.5°E −4.1 Table 2. As in Table 1, but for CWB blocking events.
To test the sensitivity of the choice of blocking events, we adopt the blocking index defined by Tibaldi and Molteni (1990), which is characterized by reversed meridional 500-hPa geopotential height gradients around 60°N. As pointed out by Pelly and Hoskins (2003) and Jin et al. (2009), this blocking index has a deficiency in identifying blockings due to the fixed reference latitude. Therefore, we extend the reference latitude to 45°–70°N for every 5° to calculate the southern and northern eight-day low-pass filtered 500-hPa geopotential height gradients (GHGS and GHGN) for each longitude as follows:
where ϕn=75°N+Δ, ϕr=60°N+Δ, ϕs=45°N+Δ, and Δ=−15°, −10°, −5°, 0°, 5° or 10°. A given longitude is considered to be blocked and assigned to be 1 if GHGS > 0 and GHGN < −10 for any 5° of latitude between 45°N and 70°N.
-
Rossby wave energy propagation is described by the wave activity flux (W) defined by Takaya and Nakamura (2001). The expression of lateral W is
where p is the pressure normalized by 1000 (hPa); U = (U, V) is the horizontal basic flow, which is the 31-day running average of the annual cycle (1979–2018); ψ′ is the streamfunction anomaly calculated from the geopotential according to ψ=ϕ/f; and u′ and v′ are geostrophic wind velocity perturbations. For each blocking event, daily perturbations are defined by subtracting the corresponding 31-day running mean annual cycle. The wave activity flux is calculated using the composite of the anomalous (u, v) and geostrophic streamfunction. The wave activity flux of the composite assumes that the composite fields represent the primary features of interest. The advantage of W is that it can depict the evolution of large-scale disturbances embedded in a zonally varying basic flow.