Do tropical cyclones go through phases?

Studies have shown that differences in both frequency and intensity of tropical cyclones can occur depending on the phase of the Madden-Julian Oscillation (MJO). This relationship can be seen in the tropical Atlantic, the northwest Caribbean and in the Gulf of Mexico.

What is the MJO?

The MJO is an oscillation of convection (cloud and rainfall) and winds in the tropics at intra-seasonal time scales of 30 to 60 days (Madden and Julian, 1971 and Maloney and Hartmann, 2000). Tropical convection and wind anomalies develop over the Indian Ocean and then track eastward to the Pacific Ocean. The MJO can be considered to have two key states in any location: one with enhanced convection and one with supressed convection (NOAA, 2014). The geographical location of the convection during the oscillation can then be grouped into eight ‘phases’, as shown in figure 1.

Figure 1: Rainfall anomalies for the eight MJO phases. Normal is 1970 – 2012 for November to March.
Source: NOAA/OAR/ESRL PSD, 2014

Is the MJO a trigger for tropical cyclones?

Firstly, what is a tropical cyclone?  This term is often used generically for any low pressure region with thunderstorms and a circular wind flow.  More specifically, a tropical cyclone is a tropical depression (which is allocated a number by the National Hurricane Centre, ) when sustained winds are below 39 mph.  Sustained winds between 40 – 73mph are tropical storms (and are allocated a name by NHC), with hurricanes having sustained winds of 74 mph or above, and ‘major’ hurricanes sustained winds above 111 mph.

Clustering of tropical cyclones in space and time have been observed. Some studies (e.g. Klotzbach, 2008 and Klotzbach and Oliver, 2014) suggest that the clustering can be attributed to phases of the MJO.  The study by Klotzbach (2008) was inclusive of all tropical cyclones that were observed in the period 1974 to 2007 and after categorising them according to tropical storm, hurricane and major hurricane strength, they noted that more than double the amount of hurricane days, and more than three times the amount of ‘major’ hurricane days, occurred when tropical cyclones formed in phases one and two of the MJO, as compared to tropical cyclones forming in MJO phases six and seven.

Klotzbach and Oliver (2014) investigated the influence of the MJO on tropical cyclone activity in the period 1905 to 2011. They recorded the MJO phase of the storm based on the day that a tropical cyclone was first classified as a named tropical storm. They noted that enhanced MJO-related convection over Africa and the Western Indian Ocean, typically associated with MJO phases one through three (see Figure 1), is correlated with increased tropical storm activity in the Atlantic.

Wind shear and tropical cyclones

The connection between convection over Africa and the Western Indian Ocean, and tropical cyclone activity in the Atlantic is not obvious.  It was explored by Mo (2000) and is rather complex, but at a high level: when enhanced convection is present over the Indian Ocean, the air flow anomalies in the upper troposphere (higher level of the part of the atmosphere closest to Earth) just north of the equator are positive. When these air flow anomalies are positive, there are more upper level easterly wind anomalies over the Caribbean and the Tropical Atlantic and vertical wind shear is decreased.  Low wind shear is conducive to tropical cyclone development (see figure 2 below).

Low wind shear in the North Atlantic is supported by phases one to three of the MJO, whereas phases five to seven are supportive of high wind shear (Klotzbach and Oliver, 2014).

Figure 2: In the presence of vertical wind shear, a storm’s core structure will be tilted in relationship to the wind shear. This tilting will disrupt the flow of heat and moisture which inhibits the storm from developing and becoming stronger. Source of text and figure: NOAA

Do observations in recent years support literature?

Tropical cyclones continue to be relevant, with eight of the top ten costliest hurricanes in terms of damage occurring since 2004 (JPMorgan, 2017).  Here we look to see if tropical cyclones over the last 1.5 decades continue to support the earlier findings.

All tropical cyclones of tropical depression strength and above are included.  The MJO phase for each tropical cyclone is assigned on the day it was first classified by the NHC, providing the cyclone formed when the MJO signal was sufficiently strong.  We use a Wheeler-Hendon MJO value of 1 or above to indicate ‘sufficiently strong’, which is standard practice.  (Wheeler and Hendon 2004).

Figure 3: Count of tropical cyclones and count of hurricane and major hurricane days between 2002 and 2017 according to the MJO phase on their day of formation filtered to include only tropical cyclones where MJO strength is ‘sufficiently strong’.

Figure 3 (the blue bars) shows that phases one and two of the MJO do indeed encompass more Atlantic tropical cyclones than other phases and phases seven and eight see the least.  Klozbach and Oliver’s (2014) findings are similar, though they have phases six and seven with the least.  There are three differences in methodology:  KO2014 normalise the results according to how frequent each phase of the MJO is; KO2014 categorise all tropical cyclones even if the MJO signal is weak; and  KO2014 take the MJO phase when the cyclone becomes a tropical storm, whilst we take it when the cyclone becomes a tropical depression.

When further categorising the tropical cyclones to hurricanes days and ‘major’ hurricane days, as in Klotzbach (2008), we also find that more than three times the amount of hurricane days and more than four times the amount of ‘major’ hurricane days occurred when tropical cyclones formed in phases one and two of the MJO, as compared to those forming in phases six and seven (Figure 3, orange bars for hurricane days, red bars for major hurricane days). Figure 3 also shows that both hurricane days and ‘major’ hurricane days are lower in phases seven and eight, as compared to phases six and seven.

How does the MJO influence compare to that from ENSO?

The literature widely supports the belief that an El Niño event in the Pacific is unfavourable for Atlantic tropical cyclone development (Klotzbach, 2011). Klotzbach and Oliver (2014) find that in El Niño years, MJO conditions favourable to tropical cyclone development (i.e. phases one and two) do not significantly increase the frequency of tropical cyclones, suggesting that ENSO is a more dominant mechanism than the MJO for Atlantic tropical cyclone development.