Atlantic Tropical Action 2013

[Ed Lizard]

Looks like there was some sassycane action while I was gone...

 

 

How many Cat 5 hurricanes were consumed during that vacay?

[icebreaker5221]

I am beginning to think that the STR is going to be so strong heading into July, that any

tropical development will be pushed further south like Barry was. The Euro is showing

this scenario continuing. I think that this is a result of the more positive NAO this year.

 

500.gif

 

 

 

This hurricane season and large-scale pattern is starting off a lot like 1979. 

 

~ Persistent +NAO that develops during June (which stayed + till Oct that season)

~ A huge ridge in the west / -PNA combined with a trough in the east (this broke down in July that year)

~ A PDO that started the year - but trended + before the start of hurricane season

~ Neutral ENSO, but trending +

 

And particular to the tropical Atlantic:

~ 2 TS in June, one of which threatened the US, the other an early easterly wave

~ High pressure in the W. Gulf keeping any threats early on far from TX

 

For completeness, the first hurricane that year, Bob, formed on July 9 and became a hurricane on July 11.

 

MSLP anomalies 1979:

[Srain]

This hurricane season and large-scale pattern is starting off a lot like 1979. 

 

~ Persistent +NAO that develops during June (which stayed + till Oct that season)

~ A huge ridge in the west / -PNA combined with a trough in the east (this broke down in July that year)

~ A PDO that started the year - but trended + before the start of hurricane season

~ Neutral ENSO, but trending +

 

And particular to the tropical Atlantic:

~ 2 TS in June, one of which threatened the US, the other an early easterly wave

~ High pressure in the W. Gulf keeping any threats early on far from TX

 

For completeness, the first hurricane that year, Bob, formed on July 9 and became a hurricane on July 11.

 

MSLP anomalies 1979:

compday.5EeUrOmc45.gif

TS Claudette still holds the record for the most rainfall in a 24 hour period recorded in the US I believe.

 

http://www.hpc.ncep.noaa.gov/tropical/rain/claudette1979.html

[TropicalAnalystwx13]

It's consistent [with development] if nothing else. Has gradually been moving the timeframe up. 192 hours now. Track has been shifting a bit, but it's been steady with intensity as well.

 

We'll see.

 

[Ed Lizard]

It's consistent [with development] if nothing else. Has gradually been moving the timeframe up. 192 hours now. Track has been shifting a bit, but it's been steady with intensity as well.

 

We'll see.

 

 

 

Assuming the Canadian is also bad with WestPac tropics, would this mean even if it dealt with weather in the Westerlies fairly well, out past a week badly forecast or phantom West Pac recurvers into the Westerlies would ruin the Canadian as a long range global model anywhere in the world?  Does someone have one of those model comparison graphs for someplace outside the tropics?

 

 

One small tribute to the Canadian- perhaps because the older lower res Canadian spun up so many ersatz cyclones, it was bound to be first, but the non-tropical development of Edouard was first hinted by the Canadian.  Been five years, it is about due to be right.

 

 

 

 

Edit to Add-  there isn't a single (out of 20 displayed) 12Z GEFS member at Hour 192 with any low 1004 mb or lower in the Gulf or Caribbean.  By Day 10, GEFS like the EasPac, but not a one in the Gulf or Caribbean still.

[Ed Lizard]

This hurricane season and large-scale pattern is starting off a lot like 1979. 

 

~ Persistent +NAO that develops during June (which stayed + till Oct that season)

~ A huge ridge in the west / -PNA combined with a trough in the east (this broke down in July that year)

~ A PDO that started the year - but trended + before the start of hurricane season

~ Neutral ENSO, but trending +

 

And particular to the tropical Atlantic:

~ 2 TS in June, one of which threatened the US, the other an early easterly wave

~ High pressure in the W. Gulf keeping any threats early on far from TX

 

For completeness, the first hurricane that year, Bob, formed on July 9 and became a hurricane on July 11.

 

MSLP anomalies 1979:

compday.5EeUrOmc45.gif

David and Frederick, I was still in high school then up North and I still remember them.  David was going post-tropical, but after Belle and 1978, I was interested enough to remember the wind and rain.  Frederic below, a Josh dream storm...

 

[icebreaker5221]

TS Claudette still holds the record for the most rainfall in a 24 hour period recorded in the US I believe.

 

http://www.hpc.ncep.noaa.gov/tropical/rain/claudette1979.html

 

 

Yeah, by that time the big western -PNA had broken down, and the westward extension of the Atlantic subtropical high was less robust.  TX was only "protected" early on from June 1 - July 15. 

 

Still, strong high pressure was still in place, it just retreated slightly inland and northward.  In fact, the fact that Claudette was essentially punching right into a ridge with little overall steering flow likely contributed to the very slow storm motion and incredible rainfall totals. 

 

[icebreaker5221]

David and Frederick, I was still in high school then up North and I still remember them.  David was going post-tropical, but after Belle and 1978, I was interested enough to remember the wind and rain.  Frederic below, a Josh dream storm...

 

 

Yeah, both of those storms could have been much worse than they were for the U.S. if Hispaniola had not demolished David's core and if Cuba had not delayed the intensification of Frederick.  Still both very impressive storms at their peaks. 

[phil882]

OK, OK, sorry for suggesting the CMC was improved. I thought it was because it's not cranking out six Cat 5s in every run, but whatevz. Forget I said it.

I certainly wasn't trying to call you out Josh... just needed to be the wet blanket before too many people get excited about the CMC's "new" outlook in the tropics.

Speaking of new... the ECMWF had an upgrade today, improving its vertical resolution.

[GaWx]

Yeah, both of those storms could have been much worse than they were for the U.S. if Hispaniola had not demolished David's core and if Cuba had not delayed the intensification of Frederick.  Still both very impressive storms at their peaks. 

 

 Thank goodness for Hisp. for Savannah's sake since it was hit head on by David. It was bad enough as it was. Wow, what a major headache! I'll never forget it! We had power out for four days and were some of the lucky ones. Some had it out for two weeks. Still, those four days were miserable with no A/C and lots of heat and humidity to follow the storm. Many trees were down including four at one neighbor's house.

[HurricaneJosh]

Yeah, both of those storms could have been much worse than they were for the U.S. if Hispaniola had not demolished David's core and if Cuba had not delayed the intensification of Frederick.  Still both very impressive storms at their peaks.

I don't see the two as analogous in terms of how the Greater Antilles affected them. Yes, Hispaniola wrecked David completely and protected the USA, but Frederic had plenty of time to intensify after Cuba and it did. With what we know now about eyewall replacement cycles, etc., I doubt Frederic would have been much stronger at landfall in the USA, even if Cuba weren't there. It came ashore as a strong Cat 3-- that in itself is unusual on the N Gulf Coast.

[HurricaneJosh]

Speaking of new... the ECMWF had an upgrade today, improving its vertical resolution.

Cool-- will be interesting to see what happens. Will it get better at predicting genesis, perhaps?

[dtk]

Cool-- will be interesting to see what happens. Will it get better at predicting genesis, perhaps?

Not sure, but from what I understand they struggled like heck to get this implementation done (it's been in the works for a while).  So many parameters can be sensitive to vertical resolution and layer distribution, particularly those related to boundary layer and convective parameterizations.  It looks like their retrospective testing yielded a mixed bag in the tropics:

www.ecmwf.int/products/changes/ifs_cycle_38r2/scorecard.html

 

Given that they retuned the previously used parameterizations, I wouldn't expect too much of a difference in terms of predicting genesis...but I have nothing other than a gut feeling to back that up.

[icebreaker5221]

I don't see the two as analogous in terms of how the Greater Antilles affected them. Yes, Hispaniola wrecked David completely and protected the USA, but Frederic had plenty of time to intensify after Cuba and it did. With what we know now about eyewall replacement cycles, etc., I doubt Frederic would have been much stronger at landfall in the USA, even if Cuba weren't there. It came ashore as a strong Cat 3-- that in itself is unusual on the N Gulf Coast.

 

 

Yeah, hard to say.  From what I can tell via reanalysis, conditions we're extremely favorable in the Gulf, so the absence of delayed intensification over the warmest waters may have prevented something more along the lines of Katrina or Camile, although we'll never know for sure.  Yes I know Katrina was a cat 3 at landfall just like Frederic, but the extended period of time as a cat 4-5 likely contributed to the massive storm surge.  Perhaps with more time as a major Frederic could have built up an even greater storm surge than it did produce. 

 

But yes, I'll concede the fact that David was a much more clear-cut case of weakening due to Hispaniola. 

[bluewave]

Yeah, hard to say.  From what I can tell via reanalysis, conditions we're extremely favorable in the Gulf, so the absence of delayed intensification over the warmest waters may have prevented something more along the lines of Katrina or Camile, although we'll never know for sure.  Yes I know Katrina was a cat 3 at landfall just like Frederic, but the extended period of time as a cat 4-5 likely contributed to the massive storm surge.  Perhaps with more time as a major Frederic could have built up an even greater storm surge than it did produce. 

 

But yes, I'll concede the fact that David was a much more clear-cut case of weakening due to Hispaniola. 

 

David could have been more like Hugo in the SE if it had  passed closer to Puerto Rico instead. The surprising aspect

of David was the unexpected ET intensification over the Northeast when it interacted with a vort over the Midwest.

It was quite a surprise for me here on Long Island to wake up to 50-60 mph gusts and a tornado watch.

[Ed Lizard]

David could have been more like Hugo in the SE if it had  passed closer to Puerto Rico instead. The surprising aspect

of David was the unexpected ET intensification over the Northeast when it interacted with a vort over the Midwest.

It was quite a surprise for me here on Long Island to wake up to 50-60 mph gusts and a tornado watch.

 

 

Like I said, because of Belle and the Blizzard of '78, I was into watching weather, as best one can pre-internet and TWC, and the amount of wind from ex-tropical cyclone David was amazing to behold.

 

 

From KHOU weather on Facebook.  I looked at GFS ensemble spaghetti with 1004 mb as a discriminator, and I didn't see anything that looked like this.  Not sure if NAEFS (Canadian members included) or a much lower bar or what.  But it is an interesting looking graphic Steve found.  Ensembles have a couple of cyclones, but nothing crazy.  Could use a higher pressure or NAEFS members I guess.

 

[Sunny and Warm]

David could have been more like Hugo in the SE if it had  passed closer to Puerto Rico instead. The surprising aspect

of David was the unexpected ET intensification over the Northeast when it interacted with a vort over the Midwest.

It was quite a surprise for me here on Long Island to wake up to 50-60 mph gusts and a tornado watch.

This is where improvements in modeling can best be seen.  What was once a surprise is now a known outcome several days in advance, or at least depicted on at least one model as a possible outcome.

[phil882]

Not sure, but from what I understand they struggled like heck to get this implementation done (it's been in the works for a while).  So many parameters can be sensitive to vertical resolution and layer distribution, particularly those related to boundary layer and convective parameterizations.  It looks like their retrospective testing yielded a mixed bag in the tropics:

www.ecmwf.int/products/changes/ifs_cycle_38r2/scorecard.html

 

Given that they retuned the previously used parameterizations, I wouldn't expect too much of a difference in terms of predicting genesis...but I have nothing other than a gut feeling to back that up.

 

Very interesting and thank you for the link! I am a bit perplexed by the conflicting signals in the tropics. For example, what can we deduce from the fact that the 850 hPa temp using RMS error was significantly better in the analysis, yet the 850 hPa temp using RMS error was significantly worse than the actual observations.

 

Perhaps this is a problem more related to the initial conditions made via data assimilation? If we are looking at the same forecasat and we note an improvement compared to the analysis and a degradation in comparison to the observations, how much can we trust the analysis... to play devils advocate, maybe the (in-situ) observations in the tropics are still too sparse to really get a good error assessment when comparing to the forecast model (e.g. maybe comparing the forecast grid-points to the analysis grid-points are more of an apples to apples comparison rather than trying to match up the forecast grid-points with non-uniform observations). Not sure if that makes complete sense, but the scorecard might come off as confusing in the tropics without really acknowledging the different datasets and the things that need to be done to get a direct comparison of data. 

[icebreaker5221]

Very interesting and thank you for the link! I am a bit perplexed by the conflicting signals in the tropics. For example, what can we deduce from the fact that the 850 hPa temp using RMS error was significantly better in the analysis, yet the 850 hPa temp using RMS error was significantly worse than the actual observations.

 

Perhaps this is a problem more related to the initial conditions made via data assimilation? If we are looking at the same forecasat and we note an improvement compared to the analysis and a degradation in comparison to the observations, how much can we trust the analysis... to play devils advocate, maybe the (in-situ) observations in the tropics are still too sparse to really get a good error assessment when comparing to the forecast model (e.g. maybe comparing the forecast grid-points to the analysis grid-points are more of an apples to apples comparison rather than trying to match up the forecast grid-points with non-uniform observations). Not sure if that makes complete sense, but the scorecard might come off as confusing in the tropics without really acknowledging the different datasets and the things that need to be done to get a direct comparison of data. 

 

Yeah, that is a bit confusing.  I'm wondering if this is due to unfortunate differences between the model's phase space and the real world's phase space.  Especially note that results for 100 mb and 850 mb temp is flipped vs analysis and vs obs.  Maybe when the model balances the analysis, to fit the obs at 100 mb it is also warming/cooling the temps throughout a layer of atmosphere (e.g. at 850 mb) but then makes the analysis at other levels further from reality. 

 

I've seen something similar with GFS 3d-var experiments where there was a TC with a warm-over-cold core.  When they tried to assimilate a warmer temp at the upper levels, the DA warmed the whole column of atmosphere down to the surface, rather than warming the top but cooling the bottom. 

 

It would be disturbing though if a problem like this was happening on the large scale rather than special cases (i.e. TCs). 

[Superstorm93]

Just some cautious optimism while things are dead 

 

Overall signals from the CFSv2 seem to suggest that we could actually start seeing some MDR activity or at least some pretty impressive easterly waves towards the middle of July. Overall message from the GFS and its ensembles also point to lowing trade winds across much of the MDR within five days or so. Waters may respond rather quickly this time around and warm up quite a bit, putting us decently above average. This could potentially cause upward motion to remain focused in the CATL instead of it's typical progressive pattern thus far. 

 

Something to keep a weary eye on as the EPAC mojo propagates into our region. 

 

 

"~70-80% of last 48 CFS forecasts had very strong tropical wave in central Atlantic with lows attached Days 12-15."

 

-Levi Cowan 

 

[dtk]

Very interesting and thank you for the link! I am a bit perplexed by the conflicting signals in the tropics. For example, what can we deduce from the fact that the 850 hPa temp using RMS error was significantly better in the analysis, yet the 850 hPa temp using RMS error was significantly worse than the actual observations.

 

Perhaps this is a problem more related to the initial conditions made via data assimilation? If we are looking at the same forecasat and we note an improvement compared to the analysis and a degradation in comparison to the observations, how much can we trust the analysis... to play devils advocate, maybe the (in-situ) observations in the tropics are still too sparse to really get a good error assessment when comparing to the forecast model (e.g. maybe comparing the forecast grid-points to the analysis grid-points are more of an apples to apples comparison rather than trying to match up the forecast grid-points with non-uniform observations). Not sure if that makes complete sense, but the scorecard might come off as confusing in the tropics without really acknowledging the different datasets and the things that need to be done to get a direct comparison of data. 

 

 

Yeah, that is a bit confusing.  I'm wondering if this is due to unfortunate differences between the model's phase space and the real world's phase space.  Especially note that results for 100 mb and 850 mb temp is flipped vs analysis and vs obs.  Maybe when the model balances the analysis, to fit the obs at 100 mb it is also warming/cooling the temps throughout a layer of atmosphere (e.g. at 850 mb) but then makes the analysis at other levels further from reality. 

 

I've seen something similar with GFS 3d-var experiments where there was a TC with a warm-over-cold core.  When they tried to assimilate a warmer temp at the upper levels, the DA warmed the whole column of atmosphere down to the surface, rather than warming the top but cooling the bottom. 

 

It would be disturbing though if a problem like this was happening on the large scale rather than special cases (i.e. TCs). 

 

We see this quite often in verification (i.e. analysis-based metrics say one thing wheras the observations say another).  True verification is impossible, as observations and analyses have errors.  Analysis based RMSE is especially difficult to interpret because it is possible for "stuff", say higher frequency spatial noise as one example, in the analysis to show up as error if it disagrees with the forecast in terms of those characteristics.  There is a good power point on the topic that was put together by a colleague of mine at EMC:

http://www.emc.ncep.noaa.gov/gmb/wx24fy/doc/RMSE_decomposition.pdf

 

We observed similar things when we first implemented the 3d hybrid, especially when trying to evaluate tropical winds.  Fits to analyses were quite a bit worse at certain levels, especially at one/two days lead time, wheras the forecast fits to observations were substantially improved.  Similarly, the Met Office actually got an opposite sign in their NWP index between analysis- and observation- based verification when they were first evaluating the hybrid. 

 

I guess another way to think about analysis-based verification:  If we never assimilated any observations, our forecasts would always verify perfectly if using self analysis as the metric.....

 

In terms of icebreaker's last point regarding 3DVAR increments....the good news is that we make much better use of observational information now that we have more sophisticated background error estimates (including the use of ensembles) that are flow dependent and situation-aware.

[TropicalAnalystwx13]

Just some cautious optimism while things are dead 

 

Overall signals from the CFSv2 seem to suggest that we could actually start seeing some MDR activity or at least some pretty impressive easterly waves towards the middle of July. Overall message from the GFS and its ensembles also point to lowing trade winds across much of the MDR within five days or so. Waters may respond rather quickly this time around and warm up quite a bit, putting us decently above average. This could potentially cause upward motion to remain focused in the CATL instead of it's typical progressive pattern thus far. 

 

Something to keep a weary eye on as the EPAC mojo propagates into our region. 

 

 

 

ScreenHunter_52 Jun. 26 18.27.png

Just so everybody knows how to read the maps Levi made, the colors in the background indicate significant disagreement in the pressure field between the CFS ensemble members. The blue numbers indicate high pressure points, and the red number indicate low pressure points.

 

The CFS week 3 precipitation anomaly maps show 'something', whether it be a series of strong AEWs or tropical cyclones. We'll just have to wait and see.

 

 

The GFS does have a major downtick in wind shear across the MDR over the next week that continues through the end of its forecast period.

 

Could be interesting. 

[icebreaker5221]

We see this quite often in verification (i.e. analysis-based metrics say one thing wheras the observations say another).  True verification is impossible, as observations and analyses have errors.  Analysis based RMSE is especially difficult to interpret because it is possible for "stuff", say higher frequency spatial noise as one example, in the analysis to show up as error if it disagrees with the forecast in terms of those characteristics.  There is a good power point on the topic that was put together by a colleague of mine at EMC:

http://www.emc.ncep.noaa.gov/gmb/wx24fy/doc/RMSE_decomposition.pdf

 

We observed similar things when we first implemented the 3d hybrid, especially when trying to evaluate tropical winds.  Fits to analyses were quite a bit worse at certain levels, especially at one/two days lead time, wheras the forecast fits to observations were substantially improved.  Similarly, the Met Office actually got an opposite sign in their NWP index between analysis- and observation- based verification when they were first evaluating the hybrid. 

 

I guess another way to think about analysis-based verification:  If we never assimilated any observations, our forecasts would always verify perfectly if using self analysis as the metric.....

 

In terms of icebreaker's last point regarding 3DVAR increments....the good news is that we make much better use of observational information now that we have more sophisticated background error estimates (including the use of ensembles) that are flow dependent and situation-aware.

 

 

Thanks for the info!  I 'm sure that the same TC experiment repeated on the GFS-EnKF or ECMWF would give much better results. 

 

And I hear you on RMSE error scores.  They still have their uses for e.g. 500 mb heights, but tell you almost nothing for verifying something like convective-scale precipitation forecasts, where the forecast could be almost perfect but just slightly off on space/time and get a terrible verification score.  That's where object-based verification comes in... (but I'll stop now before we get too OT.)

[bluewave]

The season is starting with a similar instability reduction that we saw in the last several years.

So maybe some early calls for for higher ACE than 2011 and 2012 could be harder

to achieve. We will have to see how things go once we approach the peak of the season to know.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[Srain]

Random question for the smart weather people and/or the history nerds.

 

 

Was Hurricane Audrey, the early season big Louisiana hurricane, (which, if I recall, people on this very forum in earlier years say Audrey may have been weakening to a Cat 1 at landfall) an analog of any type (TC possibly going to a warm seclusion XTC maybe) to Hurricane Sandy and New York/New Jersey?

 

 

 

Radar imagery loop of Audrey from Ellington AFB (Houston)

 

http://docs.lib.noaa.gov/rescue/mwr/085/mwr-085-06-0221.pdf

Today marks the 56th Anniversary of Hurricane Audrey.

[Ed Lizard]

Today marks the 56th Anniversary of Hurricane Audrey.

 

I-Cyclone has it on FB with an entire picture collection.  Looks like the eye opened up before landfall..

 

 

959 mb in Cameron with eye passing to the West isn't to shabby.

 

 

docs.lib.noaa.gov/rescue/mwr/085/mwr-085-06-0221.pdf

 

http://www.youtube.com/watch?feature=player_embedded&v=6ivf-yWyww8

[Ground Scouring]

The season is starting with a similar instability reduction that we saw in the last several years.

So maybe some early calls for for higher ACE than 2011 and 2012 could be harder

to achieve. We will have to see how things go once we approach the peak of the season to know.

But why? I thought that the SIOD was more favorable for instability this year. May HM et al. chime in on this?

[Ed Lizard]

But why? I thought that the SIOD was more favorable for instability this year. May HM et al. chime in on this?

 

 

The Caribbean plot looks close to normal.  If we're talking the preferred Josh Yucatan landfalls, I don't see a problem,

[bluewave]

But why? I thought that the SIOD was more favorable for instability this year. May HM et al. chime in on this?

 

We probably need more research to know for sure.

 

http://www.wunderground.com/blog/JeffMasters/archive.html?year=2012&month=11

 

Unusually stable air over the Tropical Atlantic in 2012

For the third consecutive hurricane season, 2012 featured an unusual amount of dry, sinking air over the tropical Atlantic and Caribbean Sea. Due to warmer-than-average sea surface temperatures and an active African Monsoon that generated plenty of African waves, a remarkably high number of tropical storms managed to form, but the unusually stable air in the hurricane genesis regions made it difficult for the storms to become strong. When we did see storms undergo significant intensification, it tended to occur outside of the tropics, north of 25°N, where there was not as much dry, sinking air (Sandy's intensification as it approached landfall in Cuba was an exception to this rule.) If we look at the last nine hurricane seasons (Figure 2), we can see that the hurricane seasons of 2010, 2011, and 2012 all featured similar levels of highly stable air over the tropical Atlantic. This is in marked contrast to what occurred the previous six years. The past three seasons all featured a near-record number of named storms (nineteen each year), but an unusually low ratio of strong hurricanes. Steering patterns the past three years also acted to keep most of the storms out to sea. Is this strange pattern something we'll see more of, due to climate change? Or is it mostly due to natural cycles in hurricane activity? I don't have any answers at this point, but the past three hurricane seasons have definitely been highly unusual in a historical context. I expect the steering currents to shift and bring more landfalling hurricanes to the U.S. at some point this decade, though.

[TropicalAnalystwx13]

Vertical instability across the tropical Atlantic reached normal values a few weeks ago, before the MJO left and the large SAL outbreak occurred. The unusual strength of the Bermuda High has pushed the ITCZ south of its normal position as well, drying the area out.

 

I think we'll see it trend back towards normal over the next few weeks. The Gulf of Guinea is trying to turn cold, we've got the negative IOD, and Atlantic SSTs are warmer relative to normal which should help focus the MJO in the region.