Five changes have been made to the procedure to estimate the probabilities for above, near, and below normal U.S. seasonal mean temperatures and precipitation associated with La Nina. The four changes that impact the precipitation probabilities have been implemented, while one of the changes (resampling) remains to be completed for the temperature probabilities. The changes are the following:

*1. Cases*

Case years from 1950 on for a particular season now almost exactly conform
to *CPC's
official list* of moderate to strong La Ninas. All but two (OND
1954 and JFM 1955) of the seasons on the official list classified as moderate
to strong La Ninas, met or nearly met the *quantitative
case selection criterion* previously used here so nothing was compromised
by adding several new cases. In fact the addition of these new cases usually
reinforced our confidence in the La Nina signals.

*2. U.S. Divisions*

Climate divisions have been combined so that the lower 48 states are now covered by 102 approximately equal-area super divisions instead of 358 unequal-area divisions. This had the dual effect of reducing the raggedness of the patterns and of enhancing signal strengths.

*3. Category Definitions*

Categories of above, near, and below normal now refer to categories that are equally probable over the 45-year period 1953-97 rather than for the full record dating back to 1895. This makes the probabilities displayed in the maps and bar graphs more meaningful for use with current forecasts of La Nina.

*4. Resampling*

To further reduce pattern raggedness and sampling error and enhance signal confidence the probabilities have been developed for a particular season through a bootstrap technique that consists of building up sample seasons through resampling (with replacement) a pool of all the months from the case year seasons.

*5. Correction for Global-Change Signal*

Unlike for precipitation, long-term trends in U.S. temperatures associated
with global change are large with respect to El Nino/La Nina signals. Therefore
the simple compositing techniques used here to describe the effects of
La Nina cannot be used as conditional probabilities for U.S. seasonal mean
temperatures this year if a La Nina does occur because long-term trends
have not been taken into account in them. CPC forecasters encountered the
same difficulty last year in attempting to apply El Nino-based composites
to winter/spring temperature predictions. A first attempt has been made
to quantitatively take the global change signal into account, guided by
insight into this signal provided by other work done at CPC. An estimate
of the signal season-by-season and division-by-division is removed from
the data and U.S. La Nina probability distributions estimated from the
residuals. The global change signal is then extrapolated forward to this
year and the La Nina probability distributions added back into it. The
results amount to deterministic trend forecasts with the uncertainties
associated with La Nina effects modifying them. We are in the process of
modifying the resampling strategy to also take into account uncertainties
of the trend estimates. The resulting distributions will replace those
currently displayed here as soon as they are available.

Maps encompassing the United States are provided for overlapping three-month
periods. These maps show the conditional probabilities, given a **moderate
to strong **La Nina episode and the continuation of long-term trends
this year, that mean temperatures will rank among the warmest or coldest
third of the 45-year climate division record. This information is depicted
only for those climate divisions where the probabilities for the tercile
classes of above, near, and below normal departed sufficiently from a uniform
distribution that the odds of the departure being an accident were less
than about 10%.

Thus, this set of charts not only provides insight into what kind of conditions La Nina will favor in a specific area for a specific time of year, but also how reliably those conditions are expected to be observed with a moderate to strong La Nina episode.

The probability for the tercile class that the distribution is skewed towards is color coded; for example when the highest probability is for the warm class at a location this probability (in number of cases out of the total number of moderate to strong episodes for the period) is denoted by a red shade, but when the highest probability is for the cold class it is indicated by a blue shade. For each colored division on a chart the probability for the opposite tercile class (for example the coldest third for divisions that are colored a red shade) is denoted by a number (again out of the total). Thus for every three-month period at every climate division the complete distribution among temperature terciles for moderate to strong La Nina episodes can be determined from the diagrams.

For specific examples refer to the [October
thru December] temperature chart constructed from trend extrapolations
and 8 moderate to strong La Ninas: For eastern New Mexico the probability
estimated for the warmest third is 6 out of 8, the middle third 2 out of
8, and zero out 8 for the coldest third, while for northern Minnesota the
probabilities for the warmest and coldest thirds are exactly the opposite.
Thus, suitably smoothed, the information on the diagrams can be used to
formulate *a priori* probabilities of the different temperature classes
conditional on a high-confidence La Nina forecast.

The years representing moderate to strong La Ninas change from period to period. This is because the part of the year for which the central Equatorial Pacific sea surface temperatures (SSTs) are well below normal differs from episode to episode. The cases that were included are those for which the average SST in a prescribed area was close to or greater than one degree Celsius below normal in at least one of three months spanning a particular period and close to or greater than eight-tenths of a degree Celsius below normal in the remaining months. The key area used for case selection is bounded by the Dateline and 150 west longitude and 5 north and 5 south latitudes. This area was used because it approximates the region in the equatorial Pacific where tropical convection and rainfall (the major source of atmospheric energy in the tropics) are the most sensitive to relatively small changes in SST. Thus, the SST anomaly in this area should be a good index of how strong a La Nina's impact on the global atmosphere will be.

For precipitation cases were selected for the period 1930/31 to 1996/97
with the addition of MJJ and JJA 1910 to ensure minimum sample sizes of
four for all periods, while for temperature cases from 1939/40 were selected.
Cases from 1950 on conform exactly to *CPC's
official list* of moderate to strong La Ninas with the exception
of OND 1954 and JFM 1955 which did not satisfy the criterion but otherwise
behaved like moderate La Ninas.** **The diagrams shown here reflect,
complement, and extend the information recently presented by Livezey *et
al.* (1997: *J. Climate,*** 10**, 1787-1820; hereafter L97),
who used similar selection criteria.