Weatherinfo

For fun (and to learn more Java) I plan to write a program (the “Hurricane Game”) to model inter-seasonal variation in Atlantic storm count. This will be a first-order approximation and will in no way pretend to catch the nuances of a real season. However, I do hope that, despite the crude simplifications, the model will generate realistic inter-seasonal variability. We’ll see!

Below is a general outline and statements of the rules that will govern the program. First, to help explain things, I offer a map (double-click to expand it):

The game begins at the starting line, with a seedling. A seedling moves across the map one square at a time. The seedling starts with an arbitrary initial “score” (a number from 1 to 19) and has the opportunity to pick up additional points as it crosses the map. If it can raise its score to 20 before hitting land then it becomes a named storm! 

As mentioned, a seedling moves one square at a time. For simplicity in this exercise, the seedling moves due west (left). The latitude at which a seedling launches varies randomly along the heavy black line. (Later, I’ll add a feature to allow seedlings to gain latitude as they progress across the Atlantic.)

A new seedling is launched from the starting line with every eighth move. This way there are usually multiple seedlings on the map. (Why every eighth move? Well, in this model a “move” corresponds to about 12 hours, so a decent seedling is launched from Africa every four days, which is about right.)

Now, how can a seedling earn points so as to be crowned as a named storm? Well, some of the squares on the map award points to a seedling, which the seedling accumulates. To help the explanation, here’s a second map (double-click to expand):

The map contains “green blobs”. Green blobs award points: one point for landing on a light green square, two points for a dark green square. As a seedling moves through a blob it likely lands on green squares and is awarded points, with the aim of accumulating 20 points.

An additional wrinkle: the green blobs are not stationary but instead they move within the red boxes. For every move of the seedlings, the blobs also move. A blob can move one up, one down, one left or one right, at random. It always retains its odd shape, though and stays within its red box.

There are 360 turns in a game (which more-or-less corresponds to the number of 12-hour periods in a hurricane season). Those 360 turns create a total of  45 or so seedlings ( one every eighth turn).

The above approximates the likelihood of a seedling becoming a named storm at a given time in the season, but doesn’t reflect how conditions change during a season, or among seasons. To accomplish that I’ll introduce a few more rules:

*  the initial score assigned to a new seedling is 5 during the first 90 turns, which increases to 9 during the next next 90 turns, then increases to 12 during the next 90 turns, then decreases to 9 during the final 90 turns.

* Blob #1 starts with the point totals (1 point for light green, 2 for dark green) shown above for the first 90 turns, then the point totals grow (1.5 light green, 3 dark green) for the next 180 days, then the point totals decrease (0.5 light green, 1 dark green) for the final 90 turns.

* Blob #2 starts with point totals shown (1 light green, 2 dark green) for the first 180 turns, then increase (1.5 light green, 3 dark green) for the final 180 turns.

* If a seedling reaches 20 points and thus becomes a named storm, it “poisons” all the green boxes in a blob square while it travels through that box. The poison is that any green blob square assumes a point value of -1, which is deducted from any other seedling’s score which lands in that square.

How does all of this relate to reality? Well, seedlings are produced at roughly equal frequencies ( every three or four days) and are stronger during midseason than at the beginning or end. The seedlings travel through regions of favorable conditions and regions of neutral (or unfavorable) conditions. Those regions move over time. Those regions strengthen towards midseason and then weaken late in the season. The timing of strengthening and weakening varies by region location.

Later I’ll introduce “red blobs” (regions of unfavorable conditions which subtract points from the seedlings if encountered) and allow seedlings to change latitude as they cross the ocean. I’ll also make the rules governing the strength of seedlings and blobs more spohisticated.

 And, I may create an AMO effect which strengthens the green blobs midway through a 100-year run.

My plan is to make 100-year runs using certain assumptions and rules and see how storm count distribution appears. I’ll then vary the assumptions and see how that impacts things.

Storm count will be a function of defined variables (seedling number, seedling strengths, intensification rules) and of random events (latitude of seedling launches, exact locations of encounters with green intensification squares, proximity to named storms). It’ll be interesting to see the shapes of the 100-year storm distributions, especially as rules are changed.

As always, comments are welcome.

One of the topics vigorously discussed is whether Atlantic tropical cyclones (hurricanes and tropical storms) have increased in number. It’s a complicated topic.

A recent paper by Mann Emanuel Holland and Webster is titled, “Atlantic Tropical Cyclones Revisited”. Dr. Mann makes it available here (second bullet):

http://www.meteo.psu.edu/~mann/Mann/articles/articles.html

A key part of the data used in the paper is the seasonal count of tropical cyclones. Since the paper is titled “tropical cyclones” and the key figure (#1) is also titled “tropical cyclones”, it seems reasonable to assume that the tropical cyclone data is what is used. Reasonable, but that would be wrong.

I looked at Figure 1 in some detail, as shown on the rather messy chart below. Mann’s “tropical cyclone” count is the green line while the actual count is my hand-drawn red line. (Click on the graphs to expand them.)

Since the chart above is cluttered, I extracted the data and placed it on the following graph:

 The blue line shows the actual count of tropical cyclones while the red line shows the greater number reported in Mann et al. The biggest differences are in the 1970s but occasional differences are also found in later years. Mann et al shows 22 more tropical cyclones over a 38 year period (0.57 extra storms per season) than are shown in the official records.

What is the source of Mann et al’s extra cyclones? Well, it appears that Mann included something called “subtropical cyclones” in their tropical cyclone count. While the names “subtropical” and “tropical” sound similar, they are not the same thing. A good Wikipedia article on subtropical storms can be found here:

http://en.wikipedia.org/wiki/Subtropical_cyclone

Prior to 1969 subtropical cyclones were not recorded. After 1969 they were recorded. There was a change in record-keeping.

To make a study which spans the 1969 record-keeping change one should either exclude the post-1969 subtropical cyclones or reconstruct the pre-1969 records to include subtropical cyclones. Mann et al did neither, so far as I can tell: they simply included the post-1969 subtropical storms in their “tropical cyclone” study.

There is also an argument, which I support, which says that subtropical storms should not be lumped in with tropical cyclones in these studies, as they have different origins and are different “critters”. I’ll leave that alone for now, however, and simply suggest that studies like Mann Emanuel Holland Webster (2007) should seek apples-to-apples comparisons and exclude subtropical cyclones.

August marks the two-year anniversary of Hurricane Katrina’s strike on New Orleans. I recently visited the city to see how things have progressed. Below are photos (all from July 2007) and a few comments.

For those unfamiliar with the city, New Orleans is part American and part something else, perhaps Latin American. It’s a city with a relaxed lifestyle, a place where efficency and hustle are not highly valued and where most people shrug their shoulders at corruption. Education levels are low and poverty is endemic. It is not a prime candidate for a robust recovery.

Katrina flooded about 80% of the city. After my tour of the flooded areas I estimate that  15% of the buildings have been restored and are reoccupied. Some reconstruction continues but the amount is modest and at best may add another 10% to the restored tally. Added together, I estimate that 50% to 60% of New Orleans structures are permanently abandoned and the 150,000 to 200,000 refugees will not return. The city is forever changed.

This was the home of a pharmacist. Visible on the door is the familiar “X” written by search parties, which gives information on their search of the building during the flood. On the roof (though hard to see in this photo) is the search hole cut by the rescuers. The owner evacuated prior to the storm and, two years later, has not returned. The house is abandoned, as are thousands across the city.

The photo above is the house interior (living room or parlor). Flood damage is apparent, including strewn furniture and mold on the walls. Portions of the ceiling have collapsed onto the floor, adding to the debris.

This is the kitchen. At some point this house was probably looted, as happened all across the city and evidenced by the liquor bottles sitting atop debris.

Another abandoned house. Large weeds are common across the city, with any weed control done voluntarily by the few remaining neighbors. Some of these weed patches contain marijuana plants, the result of seeds from household pot stashes dispersed by the flood.

The slab on the left is all that is left of a house. The owner requested that the government demolish the badly damaged house, which it did for free. I’m told that the government will also remove the slab for free and even buy the property.
Next to the slab is a house under repair. Damaged by the flood, it was bought by a couple who are making repairs and who hope to ultimately sell it and move back to their native Philippines. Damaged houses which have been gutted (debris removed, stripped to the studs and mold-treated) sell for about $25 per square foot of living area. FYI the houses in the background, while appearing normal, are unoccupied.

Above is a FEMA trailer, present in the thousands across the city. Living two years in a small trailer is quite difficult, especially for those with younger children, but it provides shelter free of charge. Recently a scandal developed when formaldehyde vapors were detected in several trailers, fumes known to attract trial lawyers.

An owner with a nice sense of humor. This house is in the area damaged by water, wind and crude oil that leaked from a nearby tank. This part of the New Orleans area is not far from where the nursing home residents drowned.

This truck rig is one of thousands which roam the city collecting debris. I’m told that FEMA pays US$27 for a cubic yard of debris collected, but the people doing the actual collection make only about $3 per cubic yard. The $24 difference finds its way into many pockets.

An abandoned McDonalds two years after the storm. Sights like this are very common and eerie.

An abandoned shopping center left to the weeds and the rats. This, too, is a common sight. Closed and abandoned stores like this don’t pay taxes into the local government, which creates a long term nightmarish collapse of services. Many services in New Orleans are gone, including the hospitals and especially the one which provided no-charge medical services for the region. Patients without health insurance go to suburban hospitals, which are required to treat them free of charge. These suburban hospitals are now running large unsustainable losses, creating yet another looming crisis in the region.

Above is a large golf course in the geographical center of the city, abandoned and left for whatever animals take residence. It (City Park) was an attractive course and park, with many hundreds of large live oak trees. The rumor is that a golf pro is considering buying it. The problem is that there are few locals around to use it, so the economics of any purchase are problematic.

This looks like a normal city stop sign except that it is made of cardboard and sits on an obviously short stick. Much of the flooded region was without traffic signs or lights for over a year after the storm and some are still waiting. This cardboard sign may have been installed by a government body or (more likely) by a concerned and frustrated citizen.

The photo above looks unremarkable except for an odd rise extending down the boulevard median. The rise is actually the old surface level, maybe 20 years ago, and the surrounding land has sunk by about a foot. This is indicative of the soil subsidence problem which causes New Orleans to slowly sink further and further below sea level and which makes construction of anything quite difficult.

 
This is one of the few reoccupied houses in the eastern suburbs, the area hardest hit. An American flag, though battered and ripped, still ripples in the breeze.

I don’t know where this story ends or what New Orleans will be ten or twenty years from now. Katrina may prove to be a crippling blow or the catalyst of rebirth. Money and other resources continue to flow into the city in massive amounts but it will be the citizens, not the US government, who decide the future.

Over the years the ability to detect and measure Atlantic tropical cyclones has improved markedly. In the Nineteenth Century detection was limited to storms which hit land, or passed near an unfortunate ship, and where the storm report later found its way into a historical record. Quantitative measurement was almost non-existent.

As time passed, the number of ships and land observers grew, followed by other advances:

*   land-based weather stations

* a centralized weather bureau

* initial use of aircraft for reconnaissance

*crude first-generation weather radar

*modified aircraft and airborne capabilities

*first-generation earth-circling satellites

*geosynchronous satellites

* satellites with IR and higher-resolution capabilities

* deepwater buoy networks

*oil rigs in the Gulf

* QUIKSCAT satellite wind detection

* radar-scanning satellites

* shore-based Doppler radar

and so forth.

This expansion in the coverage and capabilities of detection and monitoring methods improved the sampling of tropical cyclones. This improved sampling made it increasingly likely that the peak intensity would be captured. Tropical cyclones often vary greatly and quickly in intensity, and storms which are minimally sampled may well not have their peak intensity determined.

Consider the case of weak tropical cyclones, storms where winds are barely of minimal gale force. These weak systems, especially ones in their early hours of formation, typically have wind small wind fields. An observer has to be “in the right place at the right time’ to observe storm-strength winds. If the sampling is sparse then the chances of proper classification of this storm are small.

On the other hand, as detection and sampling tools have improved, the chances of detecting even a short-term surge to storm strength have become nearly 100%.

The Atlantic basin occasionally generates weak, short-lived tropical storms which, prior to modern sampling methods, may well have been undersampled, with their peak intensity missed, and thus excluded from the HURDAT record. (This is somewhat different from the problem of entirely failing to detect an at-sea system due to a lack of ship coverage, a point to be explored later.)

What does HURDAT show? I checked HURDAT for tropical storms which existed at tropical storm strength for 24 hours or less. These were, to a large extent, weak and small systems which would be difficult to detect even with modern shipping density. Detection generally required more-sophisticated tools, like recon flights, advanced satellite methods or Doppler radar. Here is the time series of these storms since 1900:

 

As can be seen, these weak systems are barely present in the early records even during the active 1930s to 1950s. As the first crude satellites and shore-based radars came into use the detection/sampling improved somewhat, as did the number of storms in the database. The number of detected storms increased further in the 1980s, concurrent with improvements in satellite capabilities and buoy networks. This pattern continued as the Atlantic shifted into an active mode around 1995.

Is this increase in the count of very weak storms important? Yes. In the last thirty years there have been thirty-one such storms, one per year on average. That is a significant part of the reported increase in storm count, which is important if an attempt is made to compare modern storm count with those of earlier decades.

Could this increase be due to higher SST and global warming? Here’s the plot of SST (10-25N, 20-90W, three-year smoothed) since 1950:

What it shows is that the tropical Atlantic temperatures were no higher in the early and mid 1990s than they were forty years earlier, yet the reported weak system storm count was much higher in the later period. And note that the SST graph does not extend back into the warmer 1930s and 1940s.

Could other, natural causes account for the increase? Possibly, but what those causes might be are hard to fathom.

What does this all mean? Well, it is one more illustration of the hazards inherent in comparing modern storm counts with those of earlier decades. Apples and oranges.

UPDATE (23 NOV07):

Here’s the updated time series as well as the geographical distribution of these storms -

Here is the geographical distribution for the 55 years before the start of aircraft recon and other modern detection techniques. This early era used ship and shore detection only -

And here is the modern map (detection via aircraft, satellite, doppler radar, GPS dropsonde, oil platforms, buoy network, airborne microwave radar, etc)

Suspend, for a moment, any disbelief about the global temperature records. Pretend, for a moment, that both the reported surface anomalies and the reported satellite-derived anomalies are accurate. 

In this perfect world the surface record measures its namesake while the satellite record measures a large chunk of the atmosphere generally termed the lower troposphere. AGW models show the troposphere warming as fast as, or faster than, the surface.

As an exercise, let’s see how the two measures have performed relative to each other over the last ten years. For the surface record take the average of the GISS and NCDC anomalies. For the satellite-derived record take the average of the UAH and RSS lower troposphere anomalies. Then, calculate the difference between these averages for each month of the last ten years. (Note: I did a 1-1-1 smoothing of each of the anomalies before averaging.)

Here is the plot:

In a broad and fuzzy sense, higher values (greater differences between surface and troposphere) directionally indicate increases in the lapse rate. Lower values represent a directional decrease in the lapse rate.

In an even broader and fuzzier sense, lower values indicate an accumulation of heat in the troposphere (relative to the surface) while higher values represent relative dissipation of heat. 

AGW hypotheses, as I understand them, expect to see flat-to-falling values.

What does the plot show? Interestingly, there seems to be some pattern to the madness, generally along the lines of an ENSO relationship. Note 1998, the famous El Nino year, when the difference dropped sharply lower. Note 1997 and especially 2000, when cool La Ninas emerged and the difference increased. The 2002-2006 period was one of modest El Nino-like conditions.

Physically, what does this mean? Well, to a noticeable extent the troposphere’s heat content depends on heat input by tropical thunderstorms - when El Nino is active, considerable heat accumulates in the troposphere and the lapse rate decreases a bit. Surface temperatures rise, but troposphere temperatures rise faster. Vice-versa for cool La Ninas.

How about the recent pattern? It’s intriguing, but the data is sparse and conjecture is risky. The recent pattern is as if, troposphere-wise, we’re sliding towards a La Nina-like situation but it’s not due to La Nina (we’re currently in ENSO-neutral conditions). Perhaps the Indo-Pacific Warm Pool, the major region of tropical thunderstorms, is cooling/shrinking somewhat. Perhaps there are other atmospheric oscillations in progress in this odd weather year. Who knows?

Is this ten-year plot consistent with the AGW hypothesis? Hard to say - I need to extend the chart back in time, perhaps to 1980, and see what the longer-term plot indicates. I’ll do that (but it’s a time-consumer). Stay tuned and we’ll see.

a rel=”attachment wp-att-110″ href=”http://davidsmith.auditblogs.com/2007/07/04/charts/0730074jpg/” mce_href=”http://davidsmith.auditblogs.com/2007/07/04/charts/0730074jpg/” title=”0730074.JPG”>

x

The TAO Project Office ( link ) offers data on the volume of warm water in the equatorial Pacific. This includes both surface and near-surface water (generally within 100 meters of the surface) and broadly represents the heat content of an important part of the upper ocean.

The water in question is warmer than 20C and is located between 5S and 5N and 120E to 80W (basically from The Philippines to Peru). This is an important region because it puts a large and varying amount of heat (both sensible and in the form of water vapor) into the atmosphere. The region is home to the well-known El Nino and La Nina.

How has this warm water volume varied in recent decades? Well, here’s the TAO-derived chart:

( Link to Plot )

What it shows is ENSO-related interannual variability but no decadal trend. The volume of warm water has not grown over the last 25 years, despite increases in CO2 concentrations and reported global temperature.

Now, this is not a closed region, as water moves in and out, especially in the form of cool upwelling water. Also, windspeed varies, affecting the amount of heat removed by evaporation. Still, it is interesting to note that in this well-measured region (better measured than other tropical regions) the volume of warm water, and presumably its heat content, have not grown in over two decades.

Perhaps the best use of this data is as an imperfect predictor of ENSO (El Nino / La Nina) activity. The basic idea is that upper-ocean heat accumulates and then, after a while, is released to the atmosphere. Here is the TAO chart showing warm water volume and subsequent SST (which equates to ENSO and heat release to the atmosphere):

( Link to plot )

It looks like a reasonably good correlation between heat accumulation and its subsequent release, with the heat accumulation predictably preceeding SST. Using this as a leading indicator of future activity, we should remain in ENSO-neutral or La Nina conditions in the coming months of 2007.

Unfortunately, the plots do not cover the period before 1976, the time of the last PDO shift. The period prior to 1976 was dominated by La Nina conditions, in which winds helped upwell cool deep water into the region, affecting heat content and, to some extent, global temperatures. 

So, what does all this mean? Well, warm water volume is foremost a reasonable indicator of future ENSO activity, not perfect by any stretch but still worth a look. Second, it is interesting that the volume has not grown despite rising CO2 and reported temperatures.

 Third, in the conjectural realm, I remain in the “big-bang” camp which suspects that there was a rather sudden global warming circa 1976, associated with Pacific Ocean behavior, and that after this big-bang the global temperature remained mostly constant for about two decades. The trendless Pacific warm water volume is consistent with this idea.

Then, a second shift occurred in the 1990s, possibly associated with Atlantic thermohaline activity, which elevated global temperatures in a way that was partially masked by volcanic and ENSO activity. Since that 1990s shift global temperature has again been flat.

Does this mean that CO2 did not play a role? Nope, CO2 could well have played a major role - my suggestion is that, in the last thirty years, there have been other, natural facors which likely also contributed to warming. And, in the future, these factors could, for natural reasons, shift back to their cooler states. Sometimes Mother Nature oscillates.