Is Precipitation Increasing with Temperature?

Data provided by the National Weather Service (NWS) shows that precipitation appears to be increasing with temperature in Houston and Harris County. A reader recently asked whether there was a correlation. The hypothesis: in this climate, if temperature increases, then so will evaporation and rainfall. Eighty to 130 years worth of data at different locations show both variables trending up. But most scientists would consider the coefficient of correlation weak to non-existent.

Behind the Theory

The theory is plausible from several perspectives.

Warm air holds more moisture than cool air. Warm air also rises. As it cools at higher altitudes, precipitation forms. Think “afternoon thunderstorms on hot summer days.”
Precipitation also forms when warm and cool fronts collide.
It often forms when warm moisture-laden air streams in from the Gulf.
Hurricanes form in the hottest parts of the year.

*Distribution of hurricanes by month during the last 100 years.*

But the question concerned correlation, not causation.

Other outside factors could reduce precipitation, such as droughts triggered by changes in Pacific Ocean currents. Those who remember the drought from 2011 to 2014 may also remember how hot it was.

But looking at 80 to 130 years of data highlights long-term climate trends. That “evens out” the influence of individual events.

Qualifiers

NWS plotted all available historical data for precipitation and temperature on line graphs and then superimposed trend lines. The graphs show official data from two sources: Houston-Hobby Airport and the “City of Houston.”

I put City of Houston in quotes because the the official City-of-Houston data is currently collected at Bush Intercontinental Airport. But the location has bounced around. So the “City” isn’t one location, but many:

Cotton Station (July 1881 – September 1909)
Stewart Building at Preston and Fannin (September 1909 – February 1926)
Shell Building at Texas and Fannin (March 1926 – August 1938)
Federal Building at Franklin and Fannin (August 1938 – May 1969)
Intercontinental Airport (June 1969 – Present)

We have less data for Houston-Hobby because Hobby Airport didn’t exist until 1927. That’s when someone turned a 600-acre pasture into a landing field. The City of Houston purchased the field in 1937 and expanded it.

With those qualifiers, see the charts below. Both temperature and rainfall vary from year to year. But rainfall shows extreme variance. Regardless, in all four graphs the trend lines slowly increase.

Houston-Hobby Airport

*Mean temps at Hobby increased from 69 to 73 degrees – a 4 degree increase between 1931 and 2022.*

*During roughly the same time period, precipitation increased approximately 9 inches from about 48 to 57 inches.* *Also notice the extreme range – from less than 30 to more than 80 inches.*

City of Houston Data

The City of Houston data covers a wider time period. Within that, the location varied as noted above. The big jump was from downtown to Bush Intercontinental Airport in 1969. Generally speaking, as you go farther north from the coast, precipitation decreases. But the difference is less than an inch. Atlas 14 shows that a 100-year, 24-hour storm is 17.6 inches at Hobby, 17 inches downtown, and 16.9 inches at Intercontinental.

*City data indicates mean temp has increased roughly 4 degrees in last 120 years.*

*During roughly the same years, precipitation increased about 5-6 inches. Here, the range was even more extreme. From less than 20 inches to 80 – a 4X difference.*

So the change in where the City collects official data actually worked against the hypothesis. And it shows.

Summary of Trend Differences

Summarizing the key differences:

As the temp trend line increased 4 degrees at Hobby, precipitation increased 9 inches.
As temp increased 4 degrees at various City locations, precipitation increased 6 inches.

Low Coefficient of Correlation

Jimmy Fowler of the National Weather Service’s Houston/Galveston office calculated the coefficients of correlation between the two series of data at each location.

For Hobby, the coefficient of correlation is only .03. The City’s is slightly higher at .11.
Jimmy Fowler, Meteorologist, National Weather Service

For those who didn’t study statistics in college, the coefficient of correlation tells you how much one variable changes in response to another.

A perfect positive correlation is 1.0. Example: population growth and food consumption.
A perfect negative correlation is -1. Example: hours worked and free time.

In both cases, one unit of change in the first variable accounts for an equal unit of change in the second. But most correlations fall between the two extremes with different degrees of strength.

The chart below indicates how scientists would characterize correlation coefficients of .03 and .11. Both are considered “very weak” or having “no association.”

correlation strength — *By Wayne W. LaMorte, MD, PhD, MPH, from the Boston University School of Public Health website*.

Fit of Trend Lines to Data

So, if the trends are all up, why is the co-efficient of correlation so low? Part of the answer has to do with those R2 (R squared) values you see at the bottom of the charts. They show the data doesn’t conform to the trend lines very well. Temperature fits moderately well. But precipitation shows extreme variance.

A perfect fit (1.0) would show all the data points on the line. As a rule of thumb, 0.8 (80%) or higher is considered a good fit. But the R2 values in these trend lines range from 0.03 to 0.5.

Conclusion

You can read into this data whatever you want depending on your point of view. Climate change advocates might see proof in the consistent slope of the trend lines that warming temperatures and more precipitation are related. A deeper dig into the data reveals the correlation is weak at best and possibly non-existent. Other factors may be at play and influencing the data.

To demonstrate causation, you need to show a directional relationship with no alternative explanations. But with weather, you have a multitude of alternative explanations.

Remember that weather is global and that we looked only at Houston in this instance.

However, a friend who traded weather-related derivatives before retirement tracked hundreds of temperature sites. He found they all trended warmer over time. But he believed the variance resulted primarily from changes in surrounding ground cover, i.e., replacement of natural ground cover with concrete – also known as the urban heat island effect.

He also tracked variance to changes in measurement locations (as with Houston).

Finally, remember that some of the hottest and coldest places on earth get very little precipitation. The Sahara and the North and South Poles are all considered deserts based of the amount of precipitation they get.

Net: I find the similarities in the graphs interesting enough to keep digging. As my friend suggested, it would be interesting to find the coefficient of correlation between population growth and temperature change. I won’t leap to any generalizations at this point.

Posted by Bob Rehak on 12/30/2022 with thanks to Jeff Lindner, Harris County Meteorologist and Jimmy Fowler of the National Weather Service.

1949 Days since Hurricane Harvey