Atmospheric blocking and intercomparison of objective detection methods: flow field characteristics

Abstract

Objective methods for identifying and quantifying atmospheric blocking have been developed over recent decades, primarily targeting North Atlantic blocks. Differences arise from these methods, leading to changes in the resultant blocking climatology. To understand these differences, and better inform future assessments built on quantitative detection of blocks, this paper examines blocking properties produced by three different objective detection algorithms over the global extratropics. Blocking criteria examined include 500 hPa geopotential height anomaly ( $Z^{\ast }$ ), column-averaged potential vorticity anomaly ( $P{V}^{\ast }$ ), and 500 hPa geopotential height gradient (AGP). Results are analyzed for blocking climatologies and for instantaneous blocking patterns, as well as distributions of block size, speed, duration, and distance traveled. The results emphasize physical characteristics of the flow field and the subsequent blocking regions that emerge; overall, $P{V}^{\ast }$ and $Z^{\ast }$ blocked regions often have higher pattern correlation and spatial similarity, though these two methods also display high agreement with AGP in some instances. $Z^{\ast }$ finds the largest (and greatest number of) blocked regions, while $P{V}^{\ast }$ -detected regions are smallest in all instances except Southern Hemisphere winter. In some cases, $P{V}^{\ast }$ tracks a nearby jet streak, leading to differences with height-based algorithms. All three algorithms detect some questionable low-latitude blocks that are stationary and persist but do not impair zonal flow, although at different times. Therefore, careful consideration of the algorithm biases is important in future blocking studies. For example, linking extreme weather to detected blocking could vary substantially depending on the algorithm used.

Keywords: Blocking; Climate variability; Climatology; Objective detection.

Fig. 1

Schematic of workflow for blocking calculations using StitchBlobs

Fig. 2

Seasonal averages of long term daily mean Z500 (top) and vertically averaged PV (bottom) values for (left) JJA, and (right) DJF, ERA-Interim 1979–2018. Each seasonal average contains 39 years’ worth of data. Red rectangles denote the study regions as outlined in Table 1; from left to right, the NH regions are NC, NP, and NA, and the SH regions are SI, SP, and SA. The contours are in intervals of 100 m for Z500 and 0.5 PVU for PV

Fig. 3

Seasonal averages of (top) Z500 anomaly ($Z{500}^{\ast }$) and (bottom) vertically averaged PV anomaly ($VP{V}^{\ast }$) threshold values, as described in Sect. 2.2.4, for (left) JJA and (right) DJF. Red rectangles denote same regions described in Fig. 2 caption. The contours are in intervals of 10 m for $Z{500}^{\ast }$ and 0.05 PVU for $VP{V}^{\ast }$

Fig. 4

Blocking frequency, averaged over winters from 2012–2016, for the AGP method (left), $Z^{\ast }$ method (center), and $P{V}^{\ast }$ method (right)

Fig. 5

a, b Examples of ridges that were detected by only the anomaly methods ($Z^{\ast }$, blue, and $P{V}^{\ast }$, green) in January 2013. c, d Examples of ridges that were detected by all three methods (including AGP, purple) in December of 2013

Fig. 6

Long term seasonally averaged blocking frequency for (left) JJA and (right) DJF, (top row) AGP method, (center row), $Z^{\ast }$ method, (bottom row) $P{V}^{\ast }$ method. Frequency values represent the fraction of blocked days per season as averaged over the 39 years of the study, with frequencies here ranging from 0.01 (less than 1 day per season) to 0.40 (about 37 days per season). Contour lines have intervals of 0.03

Fig. 7

Boxplots of block duration values for NH (top) and SH (bottom), in days. The upper and lower bounds of the box correspond to the 25th and 75th percentile values; the ends of the whiskers correspond to 1.5 times the 25th and 75th percentiles. Dots signify outliers beyond the whiskers. The brackets indicate pairs with statistically significant differences in the median values, with a “*” denoting $0.01<p<0.05$ and a “**” denoting $p<0.01$. The colors of the asterisks indicate which method’s median value is larger (i.e. a purple asterisk indicates that the median value for AGP is larger)

Fig. 8

Similar to Fig. 7 except with distance values in km

Fig. 9

Similar to Fig. 7 except with zonal speed in km/hr

Fig. 10

Similar to Fig. 7 except with area in ${10}^6{\text{km}}^2$

Fig. 11

Example, 12 h apart in 1995 MAM, of instances in which there is a less and b more agreement between the AGP method (purple) and the two anomaly methods (blue and green) in SP

Fig. 12

Example, in 24-h increments, of omega block detection in 1989 NA SON. The $P{V}^{\ast }$ method is denoted by the green contour, the $Z^{\ast }$ method is blue, and the AGP method is purple

Fig. 13

500 hPa vector wind field corresponding to previous figure (September 28th–October 1st), showing location of jet streaks. Wind speeds upwards of 45 m/s are visualized as the red areas. The thick black contour corresponds to the blocked region detected by $P{V}^{\ast }$

Fig. 14

Example, in 48-h increments, from NP JJA 2014, of a low-latitude block detected by the AGP (the other two methods do not detect a block here). Thin contours are Z500 in 50m intervals, and the thick purple contour denotes the detected feature. The blue box spans [230E–234W, 25N–43N] and outlines the extent of the detected block

Fig. 15

Averaged a 850 hPa temperature, b 500 hPa meridional, and c 500 hPa zonal wind anomalies for June 8th–17th 1984. The temperature contour intervals are 1K and the wind contour intervals are 2 m/s. The blue box corresponds to the one seen in Fig. 14

Fig. 16

Example, in 24-h increments, from NP DJF 2006, of a low-latitude block detected by the AGP (purple), $Z^{\ast }$ (blue), and $P{V}^{\ast }$ (green) methods. The blue box spans [165E–227E, 26N–47N] and outlines the extent of the detected block

Fig. 17

Averaged a 850 hPa temperature, b 500 hPa meridional, and c 500 hPa zonal wind anomalies for January 6th–11th 2006. The temperature contour intervals are 1K and the wind contour intervals are 2 m/s. The blue box corresponds to the one seen in Fig. 16

Fig. 18

500 hPa total wind fields and vectors for a the JJA blocking case in Fig. 14 and b the DJF blocking case in Fig. 16. The vectors indicate the wind direction, and the colors indicate the wind magnitude