Abstract

This study expands on previous studies (Harnack and Landsberg, 1978; Harnack, 1979) in that objective, statistical winter temperature forecast models are tested and verified for three additional winters (1979–81); models have been formulated and tested for the first time for the western one-third of the United States; new models have been formulated and tested using predictors defined for October; and Northern Hemisphere sea level pressure (SLP) has been tested as a predictor. An independent sample was used to test regression models.

The main results include: 1) Among the “November” type predictor models, the sea surface temperature (SST) only model continues to be superior to the others when tested on independent data, including those models using circulation predictors, and it performs significantly better than chance expectation. 2) The SST-only model performed much better in the eastern two-thirds of the United States than in the western third. 3) November SLP did not contribute to skillful winter temperature prediction as assessed by applying a model to 17 independent cases. 4) The October SST-only model showed slight skill relative to random chance but not compared to persistence. The other “October” type prediction models showed no skill. 5) The reliability of predictions increased considerably when only those predictions were verified in which the November SST-only model and persistence produced the same forecast. 62% of these forecasts have been correct (using three categories), which was superior to the performance of persistence alone (45% correct) or of the SST-only model alone (47% correct).

Geographical and yearly forecast performance differences are discussed as well as predictions made for the recent winter of 1980–81.

Abstract

The mean Northern Hemispheric circulation and United States temperature pattern of the winter 1978–79 are discussed and compared with those of the previous two winters. It is found that, although the Arctic and North Atlantic mean circulation for the three winters had some similarities, the North Pacific mean circulation was quite dissimilar in the winter 1978–79 compared with the previous two winters. With regard to the United States temperature anomaly pattern, the winter 1978–79 had below-normal temperatures coast to coast, which implies a westward expansion of the below-normal temperature regime when considering the last three winters in chronological order. A westward displacement of the core of the below-normal regime was also seen.

Possible large-scale air-sea interaction in the North Pacific is also reviewed for each of the three winters. It is proposed that major changes in the sea surface temperature pattern in the North Pacific, which occurred between the winters of 1977–78 and 1978–79, were largely responsible for the observed changes seen in the North Pacific circulation between these two winters. It is observed that the downstream circulation over eastern North America and the North Atlantic in 1978–79 did not behave as expected based on the use of teleconnection charts for the winter season. Possible reasons for this are discussed. Finally, within the context of the foregoing, various winter temperature forecasts are discussed and verified qualitatively. It was noted that most of these shared the error of forecasting above-normal temperatures for the southeastern portion of the United States.

Abstract

Various statistical models were tested for their reliability in predicting mean winter temperatures in the eastern and central United States. This study was designed td extend, and improve upon, a previous study, which also used statistical methods to formulate and test models for forecasting winter temperatures. In the current study, principal component analysis was performed separately on various predictor fields, which were selected mainly on the basis of physical-statistical relationships proposed in the literature. These predictor fields included the mean monthly sea-surface-temperature (SST) distribution for November in the eastern North Pacific Ocean. in the eastern tropical Pacific Ocean and in the North Atlantic Ocean, as well as the November 700 mb heights for a portion of the Northern Hemisphere. The time-varying amplitudes. which are derived by applying principal component analysis, were the predictors used as input for screening multiple linear regression. A Southern Oscillation (SO) index for the fall season was used as an additional predictor. The predictands consisted of area-averaged winter temperatures for nine sub-areas of the region cast of the Rocky Mountains. Four sets were formulated. corresponding to four lengths of winter: December only, December–January, December–February and December–March.

In one experiment all the potential predictors assembled (23) were used in turn with each set of predictands as input to the regression analysis. The dependent sample consisted of the period 1949–71 (23 years). The resulting four sets of prediction equations were then tested on an independent sample consisting of winters in the period 1972–77 (six years). Forecast and observed winter temperatures were put into categories for verification purposes. lime temperature classes were used. The main results of this testing showed that the forecasts of winter temperature categories improved as the length of the winter period used in the model increased. The mean percent correct for the December-only model was 20%, increasing to 56% for the December–March model. The expected mean percent by chance would be 33%.

In a second experiment, various subsets of the full act of predictors assembled were selected in order to formulate various types of predictor models. The predictor models consisted of an SST-only model, a circulation-only model (i.e., 700 mb height components plus the 50 index), and an all-predictors model (circulation and SST predictors) having a reduced number of components. Each of these predictor acts was matched with the predictand set consisting of mean temperatures for the December–February winter period. The main results obtained by testing each of the sets of prediction equations on the independent sample showed that the SST-only model was superior to the others (56% correct) and persistence. The circulation-only model had 33% correct, while the all-predictor reduced component model had 30% correct. More importantly, the SST-only model was able to distinguish correctly between the mild winter of 1975–76 and the cold winters of 1976–77 and 1977–78.

Supporting diagnostic work is also presented.

Abstract

A study has been made to assess the level of predictive skill associated with the application of a simple analog methodology to long-range temperature prediction over the continental United States in the period 1948–78. This approach relies solely on the pattern correlation statistic to select analogs from which monthly and seasonal temperature category forecasts for 68 climatic divisions (CD's) are subsequently, derived. Numerous analog model trials were attempted, employing various combinations of predictor type and domain, forecast period, and forecast lead time. Predictor types include North Pacific sea surface temperature (SST), 700 mb height and 1000–700 mb thickness. The mean percent correct statistic was used to assess the spatial and temporal variations of skill for each analog model trial, as well as for persistence forecasts.

Principal conclusions include:

1) Overall mean percent correct scores for both monthly and seasonal analog models (using three categories) were, for the most part, slightly better than random chance and occasionally better than persistence. Highest overall scores were 45% correct for February forecasts using January 700 mb heights, and 40% correct for winter forecasts using November and fall SST. Counts of significant local skill exceeded chance expectation for many analog models tested.

2) Monthly analog models generally performed best during the period January–June, outscoring persistence and chance in many instances.

3) Seasonal analog models did best for the winter and summer seasons. Winter forecasts were most successful using Pacific SST, while similar results were obtained for summer, using SST or 1000–700 mb thickness. Seasonal analog models also performed well for spring, relative to random chance and persistence, particularly those using 700 mb heights. Thickness models using a forecast lag of one season appeared to be the best overall, with some combinations of domain and lag beating persistence for each season.

The 1976 version of the University of Wisconsin model's ultra long-range forecasts of monthly mean temperature and precipitation were verified for selected United States stations over the period 1976–80. In an overall sense, neither the pentad category forecasts for four months, nor the individual year forecasts for two months, showed significant skill relative to random chance expectation. Slight positive skill was found for the July precipitation forecasts. Considerable variability of skill scores were seen from one month type to another, and from year to year. The lack of demonstrated significant skill overall for the 1976–80 period contrasts with the positive results reported by the modelers for independent sample forecasts made for the period 1961–75.

Revised and complete verification statistics for mainland United States long-range forecasts made for the period 1976–80 by the 1976 version of the University of Wisconsin model are presented. Corrections to earlier published values are given, as well as skill scores obtained using a much more complete set of stations for which forecasts were made.

The overall skill score for the pentad temperature forecasts made for January, April, July, and October is negative (−0.14), while those for pentad precipitation and individual year July precipitation forecasts are positive (0.12 and 0.04, respectively). The individual year January temperature forecast skill score was unchanged at −0.08 overall.

Abstract

The prediction of winter in the United States from Pacific sea surface temperatures was examined using a jackknifed regression scheme and a measure of intraseasonal atmospheric circulation variability. Employing a jackknifed regression methodology when deriving objective prediction equations allowed forecast to be better quantified than in past studies by greatly increasing the effective independent sample size. The procedures were repeated on three datasets: 1) all winters in the period 1950–79 (30 winters), 2) the 15 winters having the highest Variability Index (VI), and 3) the 15 winters having the lowest VI. The Variability Index was constructed to measure the intraseasonal variability of five-day period mean 700 mb heights for a portion of the Northern Hemisphere. Verification results showed that statistically significant skill was achieved in the complete sample (overall mean percent correct of 39 and 59 for three- and two-category forecasts respectively), but improved somewhat for the low VI sample. In that case, corresponding scores were (34 and 64 percent correct. In contrast, the high VI sample scores were lower (34 and 58 percent correct) than for the complete sample, indicating that skill is likely dependent on the degree of interaseasonal circulation variability.

Abstract

Seasonal precipitation is specified for the United States by matching various area-averaged precipitation statistics as predictands with three different predictors in turn: 700 mb heights, North Pacific SST and North Atlantic SST. Predictors are in the form of empirical orthogonal function (EOF) amplitude time series. The predictands used in trials include total precipitation and precipitation frequencies derived using three different critical values: 2.5, 12.7 and 25.4 mm. Screening multiple linear regression is used to relate predictands to predictors for samples ranging from 24 to 35 years in length; initially trials are compared in terms of area-averaged true skill and percent area of local significance. In order to assess specification skill on an independent sample, additional tests are made using a jackknife regression approach.

Results suggest that skillful seasonal precipitation prediction will continue to be very difficult using predictors and methods presently in common use based on the use of specification equations on an independent sample. Generally, area-averaged explained variances are less than 10% and the area of significant local skill is less than 50%. Based on the low level of specification skill, predictive skill for precipitation using specification equations with imprecisely known specifier fields (like 700 mb heights) as input would be effectively zero.

Other conclusions are:

1. 700 mb heights specify seasonal precipitation about equally well in winter, spring and summer, but worse in fall.

2. Among the three predictor types employed, 700 mb heights are best for all seasons but fall, when Pacific SST does best. Specification using Atlantic SST is poor in all instances and inferior to the use of the other predictor fields.

3. Overall among the four precipitation statistics used as predictands, the frequency statistics have a slightly better relationship with 700 mb heights or Pacific SST than do precipitation totals.

Abstract

North Pacific and North Atlantic SST (sea surface temperature) were used separately and in combination to specify seasonal-mean North American 700 mb heights. One of the goals was to quantify these relationships so that the importance of North Atlantic versus North Pacific SST could be assessed. Sea surface temperature predictors were in the form of EOF (empirical orthogonal function) amplitudes while the predictands consisted of seasonal-mean 700 mb heights at each of 25 locations over North America. Linear regression analysis was used in the data period 1949–77 to build three kinds of models: 1) using the first five North Pacific SST EOFs, 2) using the fist five North Atlantic SST EOFs and 3) using five EOFs from each field, but screening to produce the best five predictor models.

The principal findings can be summarized as:

1) Based on area-averaged skill and percent area of significant skill, North Pacific SST is a better specifier of 700 mb height than North Atlantic SST.

2) Pacific SST models have significant overall skill for all seasons except spring, with area-averaged true skill being greatest in winter (¯S = 0.247) and least in spring (¯S = 0.061).

3) Atlantic SST models do not attain field significance in any season, but perform best overall in winter (¯S = 0.095).

4) A portion of the region studied for winter and summer contained grid point locations where testing indicated that Atlantic SST adds significant information to that of Pacific SST in explaining variations of 700 mb height. This amounted to 13 and 15% of the total area, respectively, which was not enough to declare field significance.

Abstract

Interseasonal variability for a portion of the Northern Hemisphere has been quantified by using grid-point standard deviations of five-day mean 700 mb height for individual seasons, standard deviations of empirical orthogonal function amplitude time series for individual seasons, and a specially derived variability index (VI). The various measures or intraseasonal variability of 700 mb height were used in stratification studies (i.e., using difference maps based on high versus low variability) or correlation analyses so that relationships between intraseasonal variability and seasonal mean circulation or sea-surface temperature anomalies could be determined. Approximately 39 years of data (1948–76) were employed. Principal findings include:

1) Intraseasonal variability, characterized by the VI, generally showed little if any persistence from one season to the next. The exception was the autumn-winter period when the autocorrelation was 0.52. The largest contribution to this persistence came from lower latitudes.

2) The VI time series indicates that intraseasonal variability of 700 mb heights was relatively low in the 1950s to early 1960s, then high until the 1970s when near average conditions occurred, Considerable year-to-year fluctuations occurred, especially in the 1970s.

3) Significant differences occurred overall between winter (also spring) mean 700 mb height in years of highest tercile VI and years of lowest tercile VI. For winter, synoptic interpretation of the 700 mb height difference map showed that high (low) variability winters tend to be associated with expanded (contracted) westerlies. For spring, the relationship was strongest in the North Pacific sector such that high (low) variability springs tended to have weak (strong) Aleutian lows and weak (strong) westerlies in the 30–5O°N band.

4) Empirical orthogonal functions (E0Fs), which depict the important modes of seasonal-mean circulation, were correlated and otherwise related (using contingency tables) to VA. The first three EOFs explain 38% of the VI variance for winter, the most for any season, while only 11% of the VI variance is explained for autumn, the least of any season.

5) Significant differences in eastern tropical Pacific sea-surface temperature in fall-winter were related to intraseasonal fluctuations in the expansion/contraction of the 700 mb circumpolar vortex (i.e., “warm” water implied large intraseasonal fluctuation).