Experimental CCA Forecasts of Canadian Temperature and Precipitation --- Jul-Aug-Sep 2003

Amir Shabbar

Climate Research Branch, Meteorological Service of Canada, Downsview, Ontario, Canada

For the Canadian seasonal forecasts, the multivariate statistical technique of Canonical Correlation Analysis (CCA) has been used since 1995. These forecasts have been presented in the Experimental Long-Lead Forecast Bulletin over the past several years. For Canada, predictive relationships between evolving large-scale patterns of quasi-global sea surface temperature, Northern Hemisphere 500 mb circulation, and the subsequent Canadian surface temperature and precipitation have been developed. Here, we present the forecasts for Jul-Aug-Sep 2003 using the predictor fields through May 2003. This is a 4-month lead forecast. More details about the Canadian CCA-based seasonal climate prediction can be found in Shabbar (1996a, 1996b) and Shabbar and Barnston (1996).

Figure 1 shows the CCA-based temperature forecast for the Jul-Aug-Sep 2003 period expressed as a standardized anomaly. Table 1 shows the value of the standard deviation in ^oC at selected stations. The mean skill over all 51 stations is given in the caption beneath each forecast map. The field significance is also shown, reflecting the probability of randomly obtaining overall map skill equal to or higher than that which actually occurred. Field significance is evaluated using a Monte Carlo procedure in which the forecast versus observation correspondences are shuffled randomly 1000 times. The field of cross-validated historical skill (correlation) for the Jul-Aug-Sep forecast time period at this lead is shown in Fig. 2. The forecast has a modest expected skill - a mean national score of 0.18. Some random process can realize the field significance of 0.168, reflecting the fact that the overall map skill is rather low. Overall, the skill of the temperature forecast is highest in winter followed by spring and early summer but drops considerably in fall.

Relatively high local skill is found from the Great lakes extending northward into Nunavut Territory. Above normal temperatures are expected from the prairie provinces extending eastward to Labrador and Baffin Island. Below normal temperatures are forecast for British Columbia and the Yukon.

Figure 3 shows the CCA-based precipitation forecast for the Jul-Aug-Sep 2003 period, expressed as a standardized anomaly. Table 1 shows the value of the standard deviation (in millimeters) at a selected few stations. Cross-validated historical skill (correlation) for this lead and time period is shown in Fig. 4. The forecast has a rather modest expected skill: a mean national score of 0.13 and a field significance of 0.471 is considerably poorer than the traditional 0.05 level. Local skills are low throughout most of Canada except along the west coast of British Columbia and the Mackenzie Valley. Below normal precipitation is forecast over southern Alberta and southern Ontario. The Maritime Provinces, west coast of British Columbia, the northern portions of the prairies and the Mackenzie valley are expected to have above normal precipitation. Skill scores, however, indicate reliability in the precipitation forecast only over coastal British Columbia, the south coast of Nova Scotia and over the Mackenzie Valley.

Both atmospheric and oceanic indices are now showing beginning stages of the cold phase of ENSO. Statistical and dynamical models are indicating a slow progression toward the mature stage by the end of 2003 and early 2004. Traditionally, the linkage between ENSO and the Canadian climate during the summer season is rather weak. Therefore, climatic influences other than ENSO constitute the main part of the Jul-Aug-Sep forecast. It is expected that the summer warming trend over central Canada contributes towards warmer than normal temperatures. Additionally, drier than normal forecast over southern Alberta and southern Ontario is consistent with recent drying trend in those regions.

References:

Shabbar, A., 1996a: Seasonal prediction of Canadian surface temperature and precipitation by canonical correlation analysis. Proceedings of the 20th Annual Climate Diagnostic Workshop, Seattle, Washington, Oct. 23-27, 1995, 421-424.

Shabbar, A., 1996b: Seasonal forecast of Canadian surface temperature by canonical correlation analysis. 13th Conference on Probability and Statistics in Atmospheric Sciences. American Meteorological Society, San Francisco, California, Feb. 21-23, 339-342.

Shabbar, A. and A. G. Barnston, 1996: Skill of seasonal climate forecasts in Canada using canonical correlation analysis. Mon. Wea. Rev., 124, 2370-2385.

Figure captions:

Table 1. Standard deviation of temperature (Temp) and precipitation (Prcp) for the 3 month period July through September at selected Canadian stations.

Station Temp (oC) Prcp(mm)

Whitehorse 1.3 19.2

Fort Smith 1.5 23.9

Innujjuak 1.6 23.1

Eureka 1.5 9.9

Vancouver 1.0 30.2

Edmonton 1.7 34.0

Regina 1.7 30.9

Winnipeg 1.6 40.8

Churchill 1.4 24.1

Moosonee 1.5 33.2

Toronto 1.4 35.1

Quebec City 1.3 45.2

Halifax 1.0 56.6

St. John's 1.3 48.5

Fig. 1. CCA-based temperature forecast for the 3 month mean period of Jul-Aug-Sep 2003. Forecasts are represented as standardized anomalies.

Fig. 2. Geographical distribution of cross-validated historical skill for the forecast shown in Fig. 1, calculated as temporal correlation coefficient between forecasts and observations. Areas having forecast skill of 0.30 or higher are considered to have utility. The mean score over 51 stations is 0.18. Field significance is 0.168.

Fig. 3. CCA-based precipitation forecast for the 3-month mean period of Jul-Aug-Sep 2003. Forecasts are represented as standardized anomalies.

Fig. 4. Geographical distribution of cross-validated historical skill for the forecast shown in Fig. 3, calculated as temporal correlation coefficient between forecasts and observations. Areas having forecast skill of 0.30 or higher are considered to have utility. The mean score over 69 stations is 0.13. Field significance is 0.471.