Return to main weather page
July/August, 2004
Q. Somewhere between 100 and 150 years ago, government agencies in the United States began keeping official weather records. In the early years, new maximum/minimum records were fairly common because the database was so small. As time passed and more years were included in the databases, the frequency of new temperature records correspondingly decreased. Looking ahead to the end of the next 150-year period, new temperature records should, intuitively, be few and far between. It seems to me that a qualified statistician should be able to calculate an expected "new record" frequency as the database expands. Now comes my question: Is today's frequency of new temperature records, either maxima or minima, greater than expected? More specifically, is our weather becoming more extreme in general or hotter in particular?
-Richard Hanauer
Furlong, Pennsylvania
It almost seems common to hear of records being set or nearly broken. "Today set a record high temperature for the area," or "This is the most rainfall for a season since 1912." It can seem with so many years of weather data, that it would be quite unusual to set a new record. Should we interpret this record setting as a sign of climate change? Several groups are looking seriously at the possibility that not only are means in climate changing but also that climate extremes are changing, where extremes are defined in a number of ways. But let's just play the game of looking at the record high for a particular location on a specific day--say one's birthday. The first year that data are collected would set the standard. If each year were independent of the others and there were no underlying change in the environment, Year 2 would have a 1 in 2 chance of setting a new high record. Year 3 would have a 1 in 3 chance of setting a new record, etc. That would mean that after 150 years of collecting data, the chance of the 150th year setting the new record high would be roughly 1 in 150. The time between setting new records would also grow in a near geometric progression. If a new record were set in Year 150, we would expect to wait a little more than 100 years until that record was broken.
So why do record-setting events seem so common? There are at least three explanations. First, there are many records to be broken. Consider breaking the record for a daily high, after we've collected 150 years of data. Given that there are at least 365 days in a year, we might expect to set at least two records for a daily high each year. But we have more than daily highs to consider. Records are broken for daily lows, precipitation amounts, the most snowfall, largest temperature change, etc. Given the number of different ways we can look at weather and the number of new monitoring stations, we can create long lists of records with which to compare.
The second reason we may see a rash of record-setting events is that, unlike many statistical problems, weather does not occur in a random and independent fashion. If we set a record for a daily high yesterday, we have at least a moderate chance of setting a record for a daily high today. The reason is that large weather systems can require several days to pass through an area. This means that weather on successive days tends to be correlated, and implies that record-setting events will likely cluster.
The third reason we may be setting records more frequently than expected is that the climate may be changing in a preferential direction: warmer in most places, wetter in some places, and drier in others. Globally, the year 1998 is the warmest on record. The average global temperature that year broke the previous record set in 1997, which broke the record set in 1995, which superceded the record set in 1990. We know from looking at hundreds of years of data that a warm year is often followed by additional warm years. We also have some understanding of the cause of year-to-year reoccurrence in our data: oceans, volcanic emissions, and solar activity all behave on multi-year time scales and can influence weather. The higher mean temperatures we've been observing in recent years can be accompanied by record-setting daily values because extremes in annual mean temperatures or monthly mean temperatures can boost the chance of setting a record in daily values.
For the above reasons, a rash of record-setting events is, in some senses, more likely than completely randomly spread events. In fact, it's one of the aspects of our atmosphere that makes observing weather so interesting.
Contributing Editor THOMAS SCHLATTER is
a meteorologist at NOAA's Forecast Systems Laboratory in Boulder, Colorado.
