Quick Dips

It's Raining, It's Pouring

It's raining -- or is it just a shower? The answer to this question affects not only people trying to plan their day, but also meteorologists who need to make this assessment as they gather data for their weather forecasts. Wellington meteorologist John Sansom has been studying ways of distinguishing showers from rain, using new methods of statistical analysis applied to measurements of rainfall.

Sansom, who works at NIWA (the National Institute of Water and Atmospheric Research), explains that rain and its shorter, heavier counterparts, showers, are caused by different mechanisms.

"Rain is associated with a front moving over an area, while showers result from smaller-scale convective instability," he says.

"At any one time and place only one, or neither, of these mechanisms can be operating and, whichever one it is, it will persist for some time compared to the time-scale of variations of rain rate within each period of rainfall."

An expert observer in the area could usually tell whether precipitation was caused by rain or a shower, but Sansom was looking for ways to use automated recording equipment to make this distinction. To do so he used "breakpoint" data, gained from a special type of rain-gauges.

Conventional rain-gauges measure precipitation over fixed, relatively long periods such as whole days. The breakpoint data from the Dine's rain-gauges provides much more detailed information on rates of fall over much shorter time-spans. In particular, the gauges will detect and record changes from one steady state to another, for example from heavy rain to light rain or no rain.

Sansom set out to find statistical patterns in the breakpoint data recordings so that climate researchers and meteorologists could tell from the recordings whether rain or showers were happening at a particular point. He analysed breakpoint data collected over a period of years at Invercargill Airport, and correlated these statistics with hourly reports made by observers at the airport, where the Met Service used to run a 24-hour station.

"I also used some data from Israel," he says. "The results were, surprisingly, quite similar to those from Invercargill."

For his analysis Sansom needed to develop some new statistical tools. Using these, he succeeded in producing systems which could take the breakpoint data and give a probability as to whether recorded rainfall was the result of rain or a shower.

Sometimes the probabilities are similar and the choice between rain or showers difficult, but in these cases the persistence of the conditions could be used to refine the analysis, he says.

"After a period of rain, for example, it's much more likely that you'll get more rain rather than a period of showers."

He has already started work on applying different methods of statistical analysis, replacing the static analysis of sets of data with dynamic methods which can take into account the order in which the weather events have been happening.

"It's not yet at the stage where it can be used by the Met Service, but it could be in the future," Sansom says.

"At NIWA we are now trying to fit these statistical models with data from as many places as possible around the country, to get information on the amount of time each place has showers and rain, and, if it rains, the intensity and duration of the rain.

"As well as being useful for weather forecasting, this information will be important for engineers and planners who need to know about the rainfall patterns in different areas," Sansom says.

"Dr Sansom has cleverly used an array of statistical arguments, probability models and simulation to guide his analysis. This has not only led to a number of international publications in applied meteorology, but also novel and important methodological contributions to the statistical literature," says Dr Peter Thomson, who supervised Sansom's PhD research in Victoria University's Institute of Statistics and Operations Research