How does rising atmospheric CO2 affect marine organisms?

Click to locate material archived on our website by topic


Detecting a Human Influence on Weather
Reference
Cerveny, R.S. and Balling, R.C., Jr.  1998.  Weekly cycles of air pollutants, precipitation and tropical cyclones in the coastal NW Atlantic region.  Nature 394: 561-563.

What was done
Three independent meteorological data sets were analyzed for evidence of a seven-day cycle, which is indicative of human influence, since no natural phenomena recur on such a time scale.  In the first analysis, concentrations of ozone and carbon monoxide were examined for weekly cycles at an Atlantic island off the east coast of Canada over the period 1991 to 1995.  In the second analysis, northwest Atlantic ocean precipitation data obtained from various satellites from 1979 through 1995 were examined; and in the final analysis, two variables associated with tropical cyclones - maximum wind speed and minimum central air pressure - were studied for the periods 1946-1996 and 1970 to 1996.

What was learned
All data sets exhibited statistically significant weekly cycles.  Ozone and carbon monoxide concentrations were highest on Wednesdays through Saturdays, and lowest on Sundays through Tuesdays, coastal precipitation on Saturdays was 22 percent higher than that received on Mondays, and tropical cyclone wind speeds on Saturdays were from 3.6 to 5.0 meters per second slower than those observed on Fridays, while their central pressures were significantly higher (less severe) on the weekend than they were during the preceding five days.

What it means
The identification of a weekly cycle in any meteorological variable is extremely important, because no natural meteorological phenomenon has a consistent seven-day periodicity.  Weekly weather cycles, therefore, are a clear manifestation of human activity.  In this case, the authors suggest it is the increased particulate pollution that results from vehicular traffic and industrial activities, which are greater on weekdays than on weekends, that is responsible for their observations.  They are thus able to truly report a "discernible human influence" on regional weather, a topic on which there is as yet no such consensus with respect to global climate.

What else the data reveal
The authors analyzed maximum tropical cyclone wind speeds in two different time domains - 1946 through 1996 (their "long" database) and 1970 through 1996 (their "short" record) - because of a known overestimation bias of 4.9 meters per second in the data collected prior to 1970.  Hence, letting Vf (for "first") represent the mean maximum wind speed over the initial 24-year period 1946-1969, and letting Vs (for "second" or "short") represent the mean maximum wind speed over the subsequent 27-year period 1970-1996, we can write Vf = Vs + 4.9 meters per second.  We can also write the equation that defines the mean maximum wind speed over the entire 51-year period as Vl (for "long") = (24Vf + 27Vs) / 51.  Substituting the expression for Vf defined by the first of these equations into the second equation then allows us to solve for the difference Vl - Vs, obtaining a value of 2.3 meters per second.  But the observed value of Vl - Vs, as read from the authors' Figure 2c, is approximately 4.5 meters per second, or 2.2 meters per second more than what is calculated on the assumption of no real change in wind speed between the 1946-1969 and 1970-1996 time periods.  Consequently, true wind speed had to have dropped between the midpoints of these two intervals.

We can calculate the magnitude of this real change in wind speed by again defining Vl = (24Vf + 27Vs) / 51, subtracting Vs from each side of this equation, setting the resultant left-hand side equal to the observed data-derived value of 4.5 meters per second, and solving for Vf - Vs to obtain 9.6 meters per second.  Then, subtracting the Vf overestimation bias of 4.9 meters per second from this result yields the true decrease in mean wind speed between the two periods 1946-1969 and 1970-1996: 4.7 meters per second, which represents a decline of about 16% between the midpoints of these two periods, or a drop of approximately 0.64% per year over the midpoint-to-midpoint time interval.

Is this secular decline in maximum tropical cyclone wind speed the result of some anthropogenic activity?  Everything else examined by the authors would tend to suggest that it is.  Is it the result of "extensive regional pollution advection into the Atlantic"?  Studies of historical trends of Atlantic seaboard pollution levels should be capable of answering this question with little ambiguity.  If they provide a negative answer, some other gradually changing property of the atmosphere must be responsible for the historical decline in tropical cyclone wind speeds.

Reviewed 15 September 1998