Klotzbach on: Atlantic Hurricane Season Analysis

Guest post by: Dr. Philip Klotzbach,  Research Scientist, Department of Atmospheric Science, Colorado State University

As an author on the Colorado State University (CSU) seasonal hurricane forecast, I read with interest the blog regarding “Global Tropical Cyclone Activity still at 30 year low” posted yesterday. I have started to receive questions from the media asking where the hurricanes in the Atlantic are. We forecast a very active season, calling for a total of 18 named storms, 10 hurricanes and 5 major hurricanes (compared with the climatological average of 11 named storms, 6 hurricanes and 2 major hurricanes). Before I go into more detail describing why I think it is too early to think that this is a seasonal forecast bust, I wanted to briefly address the global storm component.

I completely agree that tropical cyclone (TC) activity is very quiet so far for this year’s Northern Hemisphere season. The Northeast Pacific had no named storms during the month of July, which is the first time that this has happened since 1966. The Joint Typhoon Warning Center did not name its fifth storm in the Northwest Pacific until August 8, which is also a record. The North Atlantic has also been very quiet since Hurricane Alex in late June. Alex was the strongest storm in terms of wind speed in the month of June in the North Atlantic since Alma (1966).

With a moderate La Niña event, it is typical to expect reduced activity in the Northwest Pacific and the Northeast Pacific. It has been well-documented that storm formations in the Northwest Pacific shift northwestward in La Niña years (Camargo et al. 2007). Consequently, these storms have less time to track over warm ocean water before making landfall and therefore have less time to reach their maximum potential intensity.

Northeast Pacific storm activity is also typically reduced in La Niña years, due to anomalous upper-level easterly winds that develop at upper levels associated with the strengthening and westward-shifting of the Walker Circulation (Figure 1). From a climatological point of view, upper-level winds in the Northeast Pacific blow out of the east (Figure 2), so stronger upper-level easterly winds increases vertical wind shear, which is detrimental for storm formation. Upper-level winds in the North Atlantic’s Main Development Region (MDR) (defined as 10-20°N, 20-70°W) blow out of the west in a climatological average (Figure 3), so anomalous upper-level easterlies reduces vertical wind shear (Wang and Lee 2009).

Figure 1: Correlation between the August-October Nino 3.4 index and 200 mb zonal winds. These correlations imply that a La Niña event increases vertical shear in the Northeast Pacific while reducing vertical shear in the North Atlantic.

Figure 2: Climatological upper-level winds in the Northeast Pacific during the months of August-October. Note that the climatological upper-level winds are easterly (so upper-level easterly anomalies associated with La Niña increases vertical wind shear).

Figure 3: Climatological upper-level winds in the MDR of the North Atlantic during the months of August-October. Note that the climatological upper-level winds throughout most of the MDR are westerly (so upper-level easterly anomalies associated with La Niña reduce vertical wind shear).

I want to begin addressing the North Atlantic component of the TC activity by examining historical hurricane seasons in La Niña years. I selected years that had an August-October averaged Nino 3.4 index less than -0.5°C since 1950. I calculated August-October averages from the Climate Prediction Center’s dataset available here:

http://www.cpc.ncep.noaa.gov/data/indices/sstoi.indices

I thought that an easy way to examine the typical progression of these seasons was to see when the 2^nd hurricane formed. So far in 2010, the North Atlantic has had only one hurricane (Alex). Table 1 displays the La Niña years since 1950 along with the date of 2^nd hurricane formation and the seasonal Accumulated Cyclone Energy (ACE) index for that year. ACE is defined as the sum of the square of a named storm’s maximum wind speed (in 10⁴ knots²) divided by 10000. The 1950-2000 average of this index was 96, and for the 2010 season, we are predicting a value of 185.

Table 1: La Niña years since 1950 along with the date of 2^nd hurricane formation and the seasonal ACE accumulated in each year.

The average date of 2^nd hurricane formation for all of these years is August 21, and you will note that five years with very high ACE values of 170 or greater did not have their 2^nd hurricane formation until August 20^th or later. The 2^nd storm in 1961 did not form until September, and that September went on to have four major hurricanes, a record for the month. So, from a climatological perspective, it is not time to write off the TC season yet.

With regards to sea surface temperature (SST) anomalies, they are still running at record levels across the MDR, based on data from the NCEP/NCAR Reanalysis. I calculated the July SST over the MDR and have plotted the timeseries from 1948-2010 below (Figure 4). July 2010’s value was at record levels, approximately 0.1°C greater than it was in 2005. Calculations were made from the following website:

http://www.esrl.noaa.gov/psd/cgi-bin/data/timeseries/timeseries1.pl

Figure 4: July SST averaged over the MDR. The value of 27.5°C reached in 2010 is the warmest on record, beating out 2005 and 1958 by approximately 0.1°C.

I tend to disagree with the SST analysis given by Steve Goddard yesterday. Other SST datasets that I look at in real-time tend to agree with the fact that the MDR is running at record or near-record levels right now. Here’s an additional analysis from NOAA (Figure 5):

Figure 5: Real-time SST anomaly analysis from NOAA.

In addition, analysis from the Advanced Very High Resolution Radiometer of the difference in SST between 2010 and 2005 indicates comparable SSTs throughout the MDR (Figure 6).

Figure 6: SST difference between 2010 and 2005. Note that there are only small differences between the two years.

The sea level pressure anomaly and low-level wind pattern in July would also tend to reinforce the very warm SST anomalies that were already in place from the spring. Figure 7 displays the SLP anomaly pattern in July, while Figure 8 displays the 925-mb wind anomalies. The trades were very weak in July, which is to be expected from the pressure gradient pattern observed in Figure 7. Very weak trades were observed over the MDR, which feeds back into continued warmth due to reductions in mixing and upwelling.

Figure 7: Anomalous sea level pressure in July. This pressure gradient pattern drives anomalous low-level westerly flow, thereby weakening the trades across most of the MDR.

Figure 8: Anomalous 925-mb winds in July. Note the anomalous westerly flow across most of the MDR, implying weaker trade winds (which feed back into warmer SSTs).

With that being said, it does appear that TC activity in the Atlantic should increase over the next couple of weeks. There are a couple of systems that currently have a high chance of formation into TCs in the next 48 hours according to the National Hurricane Center’s website. In addition, we should be heading into a more favorable large-scale regime for TC formation according to the latest Madden-Julian Oscillation forecasts. I showed in a paper published earlier this year that when the MJO is located in Phases 1 and 2 (convectively active over the Indian Ocean), it reduces vertical wind shear in the tropical Atlantic, thereby providing a more conducive environment for formation on a shorter time-scale basis (Klotzbach 2010). The GFS ensemble is hinting that the MJO may be amplifying in the Indian Ocean in the next couple of weeks (Figure 9).

Figure 9: Ensemble GFS forecast for the MJO over the next two weeks.

To summarize, I would say that it is too early to discount seasonal forecasts issued by CSU, NOAA and other agencies. Our August forecast has shown significant skill over the period from 1984-2009, with our average real-time forecast error over that time period being ± 2.2 named storms, ± 1.7 hurricanes and ± 1.1 major hurricanes. Correlations between our early August predictions and post-31 July TC activity are approximately 0.60 for most predictands over that same period. Full forecast verifications from CSU are available here:

http://tropical.atmos.colostate.edu/

NOAA’s forecasts show similar levels of skill. While seasonal forecasts do bust on occasion, these forecasts show moderate skill in real-time and should not be dismissed this early in the TC season.

References:

Camargo, S. J., A. W. Robertson, S. J. Gaffney, P. Smyth, and M. Ghil, 2007: Cluster analysis of typhoon tracks. Part II: Large-scale circulation and ENSO. J. Climate, 20, 3654-3676.

Klotzbach, P. J., 2010: On the Madden-Julian oscillation-Atlantic hurricane relationship. J. Climate, 23, 282-293.

Wang, C. and S.-K. Lee, 2009: Co-variability of tropical cyclones in the North Atlantic and the Eastern North Pacific. J. Geophys. Res., 36, L24702,doi:10.1029/2009GL041469.