Forecasting with ENSO in Australia: The Problems are Not Over Yet!
Tahl KestinMonash University
Clayton, Victoria, Australia
tahl@vortex.shm.monash.edu.au
and
Neville Nicholls
Bureau of Meteorological Research Centre
Melbourne, Australia
n.nicholls@bom.gov.au
In May 1997 the Bureau, noting the developing El Niño, began issuing predictions of a higher likelihood of below average rainfall. Rainfall over the 1997-98 El Niño was below average, yet the Bureau was accused of getting the rainfall forecasts wrong and misleading farmers. There are many obstacles to ensuring that forecasts are used to their full potential. The 1997-98 El Niño demonstrated that we still do not know how to cross them all.
A major difficulty in forecasting rainfall with ENSO is the large uncertainty in the specific effects that an event will have. The 1997-98 El Niño, for example, was one of the strongest on record. But instead of rainfall being below the average (and despite lower rainfall through much of 1997), spring rainfall was the highest recorded for any El Niño and saved most of the crops. This uncertainty is reflected in the forecasts through the use of probabilities. An El Niño increases the probability of below average rainfall, but average or above average rainfall is still possible.
Probability is a very difficult concept for people to understand and use, and the Bureau's probabilistic rainfall forecasts were no exception. There was also a large gap between what the Bureau thought they were saying in the forecasts and how farmers understood it; an increased likelihood of below average rainfall was interpreted as no rainfall (i.e., drought).
The farmers' expectations were shaped further by the media hype surrounding El Niño. Several international meteorological agencies were reported in the Australian media as saying that Australia can expect a devastating drought. Not only did these reports strongly affect the expectations of farmers, but they also conflicted with forecasts issued in Australia and created confusion. So when good spring rains came in 1997, they were not seen as a chance event, but as proof that the Bureau's forecast of a terrible drought was wrong.
In addition to misinterpreting the forecasts, many farmers did not have the right knowledge about ENSO to use the forecasts to make informed planning decisions. We have come across many stories of inappropriate actions taken by farmers: from panic-driven asset selling and following the daily progress of El Niño, to complete disregard of the forecast information "in the hope that it will be wrong".
Lack of knowledge about interpreting and using ENSO forecasts caused many farmers to have inappropriate expectations of El Niño and take inappropriate actions to cope with it. As a result, the forecasts in some cases probably did more harm than good. The failures at each step of the way were mostly due to poor communication: poor communication on behalf of the forecasters and other authorities in preparing the forecasts and providing the necessary information and training for farmers to use them, and poor communication on behalf of the users, in failing to seek all the information they needed to use the forecasts. While our experience during the 1997-98 El Niño is a guide to what we can expect in a La Niña situation, we suspect that the communication problem may be even worse. La Niña is not yet part of the public's awareness, both because it has been a long time since the last big event(1988-89), and because La Niña was rarely mentioned in the context of El Niño.
Successfully forecasting under uncertainty is a complex and difficult problem. It is not a new problem, nor is ti a problem that is confined to ENSO forecasts; we take solace in that even the volcanologists on Montserrat had a similar experience to us. We are far from solving the difficulties of communication between forecasters and users, and until we do we cannot hope for ENSO forecasts to be as useful as they might be. Despite all the difficulties, we believe that an improved focus on communication strategies should become a priority for both ENSO forecasters and forcast users.