Has User Confidence in Seasonal Long

advertisement
Has User Confidence in Seasonal Long-Lead Outlooks Changed Since the
1997-98 El Niño? A Midwestern Perspective
David Changnon
Meteorology Program
Department of Geography
Northern Illinois University
DeKalb, IL 60115
Email: changnon@geog.niu.edu
1. Introduction
Prior to the 1997-98 winter the Climate Prediction Center (CPC) identified that a
significant El Niño was developing in the equatorial Pacific. By late summer 1997
winter seasonal outlooks developed by CPC and a number of private forecasters
identified a number of ‘high’ probability temperature and precipitation anomalies across
the United States (Barnston et al 1999). Some weather-sensitive users in the Midwest
(primarily those involved in winter energy decisions) used those outlooks and benefited
because the forecasts generally verified across most parts of the region (Changnon 1999;
Changnon 2000). Those who surveyed climate-sensitive decision makers after the 199798 El Niño found that future seasonal forecasts would be more widely integrated into
both operational and planning decisions (Changnon 2000). Many midwestern decision
makers believed that climatologists could now more accurately than ever predict seasonal
climate anomalies across the U.S. What they didn’t understand is that climatologists still
don’t understand all aspects of our complex climate system and perhaps the successful
seasonal outlooks of 1997-98 created unrealistic expectations! Since the 1997-98 winter
forecast some of the seasonal temperature and precipitation outlooks for the Midwest
have been incorrect (often with the observed anomaly opposite of the forecasted
anomaly). These perceived forecast failures have eroded some of the confidence users
had in this tool.
This paper examines the CPC’s winter (Dec.-Feb.) temperature forecasts for the
upper Midwest since the infamous El Niño of 1997-98 and discusses issues related to the
tool and those who wish to incorporate it (or aspects of it) into decision models for the
2002-03 winter. Winter was chosen because of its impact on critical natural gas
purchasing decisions and the fact that greater forecast skill has been found in this season
(Hartmann 2002). Understanding how well these forecasts verified in the Midwest can
help the climate community, especially those of us involved with interpreting and
disseminating information related to future winter forecasts (e.g. 2002-03). It also raises
some questions about the CPC seasonal outlooks—What roles (developing forecasts,
creating “value-added” products related to forecasts, disseminating information to users,
etc.) should be CPC’s? How should CPC forecast products be used by climatologists and
weather-sensitive decision makers? What other forms of information (related to seasonal
forecasts) would be valuable to the users? Should CPC explain forecast successes and
failures?
2. Review of Recent Winter Temperature Outlooks for the U.S. Midwest
Winter temperature outlooks developed by CPC were examined for four winters
(1998-99 through 2001-02). Again the emphasis was placed on the Midwest and those
involved with natural gas purchasing decisions, especially those facing decisions entering
the 2002-03 winter. Recent research has indicated that seasonal forecast skill in the
Midwest is not as high as it is for other parts of the U.S. (Hartmann et al 2002).
Interestingly, most users are interested in the accuracy of such forecasts, however, most if
not all, do not understand such things as the Heidke Skill Score and other statistical
descriptors of accuracy used by climatologists. Most often those who consider using
seasonal forecast information in a decision want to know, in a general sense, how good
last year’s forecast was, or what was the trend in the winter forecasts over the past five
years. Thus, as this researcher became involved with those making 2002-03 natural gas
purchasing decisions on Northern Illinois University’s (NIU) campus and elsewhere in
northern Illinois, he went back and examined the most recent forecasts. Although many
users perceive that the accuracy of the winter temperature forecast for northern Illinois
has decreased since the 1997-98, the evidence suggests something different (see Table 1).
Table 1. Forecasted and actual winter (December through February) temperature
anomalies for northern Illinois made within two months prior to winter beginning.
Winter
1998-99
1999-00
2000-01
2001-02
CPC Forecast Temperature
Anomaly for Northern IL
Climatology
3% for Above Average
Climatology
5% for Below Average
Actual
Temperature
Anomaly for Northern IL
Near Record Warm
Near Record Warm
Near Record Cold
Near Record Warm
Overall, the winter temperature forecast for northern Illinois was correct once
(1999-00), and incorrect, forecasting opposite of what actually occurred, once (2001-02),
with two years when there was no forecast—insufficient skill (1998-99 and 2000-01).
Because there is a small sample and in two years there was insufficient skill
(climatology) a trend cannot be easily discerned.
The most recent forecast for winter 2002-03 (produced October 17, 2002)
indicates a high probability of warmer than average conditions over the upper Midwest,
with a probability near 15% over northern Illinois. Despite this high probability many
users are more cautious in integrating the warm forecast into winter operational and
planning decisions because the 2001-02 winter temperature forecast identified the
incorrect anomaly in northern Illinois. This year, more than ever, decision makers who
examine and consider the use of seasonal forecasts have asked for more detailed
information relating to the upcoming winter temperature forecast such as: Is there more
confidence (by forecasters) with this year’s winter temperature forecast (e.g. is the risk of
an incorrect forecast smaller)? What are the physical reasons behind the forecast (e.g. El
Niño)? If you were making a decision relating to the purchase of natural gas how would
you use this (seasonal climate outlook) information? Although scientists working at CPC
could answer these questions, the question is which questions are they best qualified to
answer and which should be left to others in the atmospheric sciences community?
3. Discussion and Concluding thoughts on ‘tool’
Several Midwestern climatologists were polled on their use of the CPC forecast
and whether they bring it into discussions with potential users. Despite its successes and
failures, nearly all climatologists said they examined the forecasts monthly and conveyed
the forecast information to the user. However, many climatologists have been involved
in research using regional climate analogs and have identified useful relationships
between certain ENSO phases (El Niño, Neutral, and La Niña) and Midwestern winter
temperatures. Often the climatologist is more confident in these relationships than he or
she is with the CPC forecast. Interestingly, most users are generally interested in how to
use any of these forecasts (which all have some risk attached to them) to enhance their
natural gas purchasing decisions.
Clearly, the CPC should continue to develop and enhance seasonal climate
outlooks. As our understanding of oceanic/atmospheric teleconnections (on various
spatial and temporal scales) continues to improve so will the seasonal outlooks. These
seasonal outlooks represent a great starting point in the efforts to develop and provide
weather-sensitive decision makers with useful climate forecast information. However, as
has been suggested by White (2001), Dutton (2002), Murnane et al. (2002), and Smith
(2002), there is an important need to go beyond the forecast probabilities given in the
CPC outlooks and develop specific climate-related information that can be integrated into
decision models for individual users. Although creating “value-added” or “enhanced”
climate information may involve the use of the CPC outlooks, this activity should be
conducted primarily by private-sector atmospheric scientists in concert with decision
makers in the weather-sensitive organizations (e.g. agribusiness, utilities, insurance,
futures markets, transportation, recreation, etc.). Although some in private sector are
involved in this activity the opportunities for climatologists to work with users are
growing exponentially (White 2001; Dutton 2002). To meet the current demand this
researcher suggests that partnerships between atmospheric scientists in the private sector,
government, and academia should be considered to help develop climate information for
user-oriented decision models.
Overall, a recent assessment of potential seasonal forecast users indicated that
closer interaction between climatologists and weather-sensitive decision makers would be
beneficial. As many decision makers have asked, “Was the successful El Niño-based
winter forecast for 1997-98 a ‘blip on the screen’ or a lucky guess?” The answer is an
emphatic “no.” However, the lack of consistently correct forecasts made for the Midwest
further demonstrates that our understanding of the complex ocean/atmosphere system and
its impact on seasonal climate is far from complete. Greater research efforts (both basic
and applied) need to be focused on forecast improvement, conveying confidence levels
with forecasts (for various regions) to users, how seasonal forecasts are used in decision
models, and understanding impacts (both positive and negative) related to seasonal
climate anomalies. Many people (climatologists and users alike) will be watching closely
to see if the 2002-03 winter forecast, with its high probabilities for warmer conditions,
verifies in the Midwest. Importantly, whether the winter forecast is successful or not,
CPC scientists need to explain why. This enhances credibility in the eyes of those who
use the tool.
References
Barnston, A.G., and Coauthors, 1999: NCEP forecasts of the El Niño of 1997-98 and its
U.S. Impacts. Bull. Amer. Meteor. Soc., 80, 1829-1852.
Changnon, D., 2000: Who used and benefited from the El Niño forecasts? El Niño 19971998: The Climate Event of the Century. Edited by Stanley A. Changnon, Oxford
University Press, New York, 109-135.
Changnon, S.A., 1999: Impacts of 1997-1998 El Niño-generated weather in the United
States. Bull. Amer. Meteor. Soc., 80, 1819-1827.
Dutton, J.A., 2002: Opportunities and priorities in a new era for weather and climate
services. Bull. Amer. Meteor. Soc., 83, 1303-1311.
Hartmann, H.C., Pogano, T.C., Sorooshian, S., and R. Bales, 2002: Confidence builders:
Evaluating seasonal climate forecasts from user perspectives. Bull. Amer. Meteor. Soc.,
83, 683-698.
Murnane, R.J., and Coauthors, 2002: The weather risk management industry’s climate
forecast and data needs. Bull. Amer. Meteor. Soc., 83, 1193-1198.
Smith, M.R., 2002: Five myths of commercial meteorology. Bull. Amer. Meteor. Soc., 83,
993-996.
White, R.M, 2001: The evolving public-private meteorology partnership. Bull. Amer.
Meteor. Soc., 82, 1431-1437.
Download