Springtime ENSO phase evolution and U.S. tornado activity Sang-Ki Lee,1,2 David Enfield1,2, Chunzai Wang2, Scott Weaver3, Brian Mapes1 and Robert Atlas2 1Univ. of MIAMI-RSMAS-CIMAS, 2NOAA-AOML, 3NOAA-CPC Outline Motivation and Background ENSO phase evolution in spring Springtime U.S. tornado activity linked to ENSO Hi, I am Sang-Ki Lee from the university of Miami CIMAS and NOAA AOML. My talk today is about springtime ENSO phase evolution and its relationship to U.S. tornado activity.

Motivation April and May of 2011, a record breaking 1,084 tornadoes occurred in the U.S. with 541 fatalities 2011, one of the four deadliest tornado years in the U.S history along with 1925, 1936 and 1917 April 27, 2011 Tuscaloosa, Alabama Can we predict extreme tornado outbreaks beyond the “weather” time scale? In April and May of 2011, we had a record breaking number of tornadoes and tornado-related fatalities in the U.S. That makes 2011 one of the four deadliest tornado years in the U.S. history. Shortly afterwards, one question raised was about whether we could predict extreme tornado outbreaks beyond the weather time scale. April 29, 2011 Tuscaloosa, Alabama

Background: U.S. atmospheric conditions in spring A diagram of the location of tornado alley and the related weather systems from NOAA SPC (Art work by Dan Craggs) Little bit of background here. Over the central U.S. in spring, cold and dry upper-level air form the high latitudes collides with warm and moist lower-level air from the Gulf of Mexico at different altitudes. Due to this so-called large-scale differential advection, CAPE and lower-level wind shear are increased and that provides favorable environment to form a supercell, which is known to be linked to tornado genesis Over the central U.S. in spring, cold & dry upper-level air collides with warm & moist lower-level air from the Gulf of Mexico (large-scale differential advection) CAPE & lower-level shear provide favorable environment to form a supercell, linked to tornado genesis

Background Brooks et al. (2003): derived a tornadic condition based on CAPE and low-level shear (0 ~ 3km) threshold values Tippett et al. (2012): used a similar criteria to reasonably reproduce the number of U.S. tornado for 1971-2010 Question: Is there a slowly varying climate process that is linked to the tornadic condition and is potentially predictable? Lee, S.-K., R. Atlas, D. B. Enfield, C. Wang and H. Liu, 2013: Is there an optimal ENSO pattern that enhances large-scale atmospheric processes conducive to major tornado outbreaks in the U.S.? J. Climate, 26, 1626-1642. Brooks et al in 2003 used CAPE and low-level shear threshold values to derive the so-called tornadic condition. And more recently, Tippett et al. (2012) used a similar criteria to reasonably reproduce the number of U.S. tornadoes since 1971. However, another important question is if there is a slowly varying climate process that is linked to the tornadic condition and is potentially predictable. To answer thin question, we published this paper last year in journal of climate. In the next few slides, I will give a brief summary of what we found in this paper.

Lee et al. (2013, JCLI) The number of intense (F3 - F5) tornadoes nearly doubles during a (+) Trans-Niño phase 7 out of 10 extreme U.S. tornado outbreaks (including the top 3) during 1951 - 2010 associated with a (+) Trans-Niño phase These dots indicate the incidents of intense tornadoes during the 10 positive Trans-Niño years, and for the 10 neutral Trans-Niño years here. As you can see, the number of intense tornadoes is nearly double during a (+) phase of Trans-Niño . We also found that 7 out of 10 extreme tornado outbreaks,inlcuding the top 3, during 1951-2010 were all associated with a (+) phase of Trans-Niño.

Lee et al. (2013, JCLI) (+) Trans-Niño: zonal gradient of SST anomalies along the equatorial Pacific between CP & EP > 0 (+) Trans-Niño typically occurs during the decay phase of La Niña The positive Trans-Niño means that the zonal gradient of SST anomalies along the equatorial Pacific between CP and EP is greater than zero. The positive Trans-Niño this typically occurs during the decay phase of La Niña. And, we found that five historical tornado outbreak years, including 1917, 25, 36, 74 and 2011, were (+) Trans-Niño years as you can see here for 1974 and 2011. Five historic tornado outbreak years (1917, 1925, 1936, 1974 and 2011) were all (+) Trans-Niño years

Lee et al. (2013, JCLI) 10 Most Active Yrs 10 (+) Trans-Niño Yrs CAM3-SOM This is the composites of atmospheric conditions during the 10 most active tornado years. This is geopotential height and winds at 500mb. You can see an anomalous cyclone that bring more dry and cold upper-level air to the central US., increased moisture transport and increased lower-level shear. When we make composites for the ten positive Trans-Niño years, we see very similar conditions. And, we performed an AGCM simulation with the tropical Pacific SST here prescribed with that from the 10 positive Trans-Niño years. The model’s response to the positive Trans-Niño is very similar to the observations.

Springtime ENSO phase evolution Equatorial Pacific ENSO SST anomalies weaker and less coherent in spring than in winter SST anomalies in the central Pacific (CP) relatively persistent and consistent between ENSO events in spring However, in Lee et al. (2013), we looked at only one particular pattern of springtime ENSO. So, now we want to explore what happen during different phases of springtime ENSO evolution. This picture shows the equatorial Pacific SST anomalies during ENSO. And, this shows inter-El Niño standard deviation, and this is inter-La Niña standard deviation. Obviously, the equatorial Pacific ENSO SST anomalies are much weaker and less coherent in spring compared to winter. However, the SST anomalies in the central pacific is relatively persistent and consistent between ENSO events in spring. So, we think that the coherent springtime ENSO SST anomalies in the central Pacific can excite ENSO teleconnection to the U.S. Coherent springtime ENSO SST anomalies in CP excite ENSO teleconnection to the U.S. (Lee et al. 2013) Lee et al. (2014)

Probability of regional tornado outbreak # of tornadoes is not an effective index of tornado activity - a handful of outbreak years dominates the time series data Probability that the number of tornadoes in a predefined region exceeds the regional climatological mean + STD Count # of F1-F5 tornadoes for 1×1 grid boxes for each month and year, then For each point, month and year, find the regional maximum # within a circle of 3° radius For each point, month and year, identify whether the regional maximum # exceeds the regional mean + STD For a subset of data, count the number of outbreak years and perform Chi-square test of 90% significance Before we move on, we need to decide the way to measure tornado activity. The number of tornadoes is typically used in previous studies. However, it is not an effective measure of tornado activity because a handful of outbreak years dominate the time series data. So, it is difficult to perform any kind of statistical analysis. So, we propose a new index, probability of regional tornado outbreak. This is the probability that the number of tornadoes in a predefined region exceeds the regional climatological mean plus one standard deviation. And, here are the steps to calculate that.

El Niño onset and U.S. tornado activity in spring First, the El Niño onset phase. These are SST anomalies for March, April and May, and these are corresponding probability of tornado outbreak. Here this black dots mean that this signal is statistically significant. And, there is some increase here and there, but they are limited to very small regions.

El Niño decay and U.S. tornado activity in spring This is for El Niño onset phase. I don’t see any increase except in this small region in May.

La Niña onset and U.S. tornado activity in spring This is La Niña onset phase. SST anomalies are pretty small. But, this wide region is strongly affected. And, tornado activity is also increased in this region in May.

La Niña decay and U.S. tornado activity in spring This is for La Niña decay phase. We see some increase here in March. But, the main signal is in April over this region.

La Niña and U.S. tornado activity in April La Niña onset phase: probability of tornado outbreak increases from 9% to 20% south of the Ohio River (KY, TN, MS, AL and GA) in April La Niña decay phase: probability of tornado outbreak increases from 12% to 24% over the central U.S. (IL and IN) in April So, let’s take a closer look at La Niña onset and decay phases in April. During La Niña onset phase, the probability of tornado outbreak increases From 9&% to 29% south of the Ohio River, especially in Kentucky, Tennessee, Mississippi, Alabama and Georgia. Here, when I say 9%, it means that in climatological sense, this boxed region has about 9% of chance to have tornado outbreak. And, that chance increase to 20% during the onset phase of La Niña. During La Niña decay phase, the probability also increases from 12% to 24% over the central U.S. in these two states, Illinois and Indiana.

Low-level vertical wind shear and lifted index Now, let’s take a look at atmospheric conditions during the La Niña onset phase in the upper panel, and for La Niña decay phase in the lower panel. You can see that the low-level shear is much increased over this region. Some region is more unstable, but this region is more stable. So, it appears that the wind shear is more critical factor here. During the La Niña decay phase, this region becomes slightly more unstable, but not so much. And, you can see that the wind shear is increased over here. Low-level wind shear increased over the regions of increased probability of tornado outbreak

Large-scale differential advection This is geopotential height and wind anomalies at 500mb. There is a anomalous cyclone that bring more cold and dry upper-level air to this region. We also see increased moisture transport form the Gulf of Mexico over this region. And, we have similar conditions during the decay phase of La Niña. This means that the large-scale differential advection is enhanced in both cases. Enhanced large-scale differential advection to the regions of increased probability of tornado outbreak

Atmospheric jet stream and synoptic activity This is the last one. This shows that the zonal wind at 300mb is much increased over here and also here in this case. And the synoptic activity, which is computed by using the variance of 5-day high-pass filtered meridional wind at 300mb, is increased in both cases. Atmospheric jet and synoptic activity increased over the regions of increased probability of tornado outbreak

Summary Probability of regional tornado outbreak in the U.S. proposed as a new index of tornado activity La Niña onset phase: probability of regional tornado outbreak over KY, TN, MS, AL and GA doubles (from 9% to 20%) in April La Niña decay phase: probability of tornado outbreak over IL and ID doubles (from 12% to 24%) in April Atmospheric conditions during La Niña onset and decay phases favorable for tornado outbreaks in the Central and Southeast U.S. Here is the summary of my talk.

Publications and funding Lee, S.-K., R. Atlas, D. B. Enfield, C. Wang and H. Liu, 2013: Is there an optimal ENSO pattern that enhances large-scale atmospheric processes conducive to major tornado outbreaks in the U.S.? J. Climate, 26, 1626-1642. Lee, S.-K., B. E. Mapes, C. Wang, D. B. Enfield and S. J. Weaver, 2014: Springtime ENSO phase evolution and its relation to rainfall in the continental U.S. under review in Geophys. Res. Lett. Supported by funding from NOAA/CPO/MAPP and AOML And, my presentation today is based on these two papers. And, this work is supported by funding from NOAA/CPO/MAPP and AOML

Future work AGCM experiments to demonstrate the link between springtime ENSO phase evolution and large-scale atmosphere processes conducive to U.S. tornado outbreak Extend the analysis for ENSO transition (ex: El Niño decay + La Niña onset) and resurgence phases (ex: La Nina decay + onset) Ultimate goal is to develop and evaluate a seasonal tornado outlook for the U.S. Here is the future work. Explore and build theoretical and observational basis for springtime ENSO phase evolution and its teleconnection to the U.S. Evaluate and improve the currently existing seasonal forecast systems