January-March 2009 T&P anomalies vs. trends & La Niña: How did the CPC forecast pan out?

1. January-March 2009 T&P anomalies vs. trends & La Niña: How did the CPC forecast pan out? Klaus Wolter (NOAA-ESRL) - 14 may09 •What was the CPC forecast for late winter 2009? • What happened in January-March 2009? •Were the anomalies consistent with the tools employed (La Niña + trend)? • What role did the snowcover play in ‘tweaking’ the temperature outcome?

2. CPC forecasts for January-March 2009: Temperature Unambiguous weak-to-moderate La Niña conditions (right) + OCN were main tools used in CPC forecast (top right: gridded version)

3. CPC forecasts for January-March 2009: Precipitation Unambiguous weak-to-moderate La Niña conditions (right) + OCN were main tools used in CPC forecast (top right: gridded version)

4. Verification for January-March 2009: Temperature Non-EC forecasts: 16.6 (2008: -5.9) All: 10.1 (2008: -4.7) Coverage: 61% (81%)

5. Verification for January-March 2009: Temperature Non-EC forecasts: 16.6 (2008: -5.9) All: 10.1 (2008: -4.7) Coverage: 61% (81%) Whoops! Two-category errors on both sides!

6. Verification for January-March 2009: Precipitation Non-EC forecasts: 41.2 (2008: 31.5) All: 13.1 (2008: 20.3) Coverage: 32% (64%) - big drop in coverage!

7. Verification for January-March 2009: Precipitation Non-EC forecasts: 41.2 (2008: 31.5) All: 13.1 (2008: 20.3) Coverage: 32% (64%) - big drop in coverage! Whoops! Two-category errors on dry side!

8. U.S. temperature anomalies based on OCN (10yr) Out of 344 climate divisions, 97 have an average standardized temperature anomaly for ‘99-08 JFM that exceeds +0.43 sigma (~upper tercile), none are below -0.43 sigma. Most of these ‘warming’ divisions reside in the Western U.S. (TX, AZ, NM, CO+).

9. U.S. temperature anomalies based on OCN (10yr) In January-March 2009 (below), 59 out of 97 flagged climate divisions end up in the upper tercile which translates into a non-EC Heidke Skill Score of +41 for OCN-based temperature forecasts. Total national HSS: +12

10. U.S. temperature anomalies based on GHG-based warming only Out of 344 climate divisions, 80 have an average standardized temperature anomaly from GHG-based forcing that exceeds 0.43 standard deviations, none are below -0.43 sigma. Most of these ‘warming’ divisions reside in the western third of the U.S., with a handful in the northeast quadrant.

11. U.S. temperature anomalies based on GHG-based warming only In January-March 2009 (below), 27 out of 80 flagged climate divisions end up in the upper tercile which translates into a Heidke Skill Score of +01 for GHG-based temperature forecasts. Total national HSS: +00

12. U.S. temperature anomalies in La Niña winters Ten strongest La Niña cases (left): out of 344 climate divisions, 184 (or 53%) show composite anomalies in excess of 0.43 standard deviations (33% tails) vs. 18 below -.43. Warmth overlaps with OCN-flagged area in southwestern U.S. Moderate La Niña cases (MEI ranks 6-15; right): out of 344 climate divisions, 107 show composite anomalies above 0.43 standard deviations (33% tails) vs. 21 below -.43. Pattern correlation with strong La Niña cases: +0.96 - similar anomalies, just weaker.

13. U.S. temperature anomalies in La Niña winters In January-March 2009 (top right), 63 out of 184 strong La Niña-flagged climate divisions (top left) and 37 out of 107 such divisions for weak-to-moderate La Niñas (left) end up in the upper tercile, while 12 out of 18 strong (11 out of 21 weak) La Niña cold divisions end up in the lower tercile. This translates into a Heidke Skill Score of +06 for both La Niña-based temperature forecasts. Total HSS: +3 (2)

14. 2007-08 impact of anomalous snowpack Snowdepth anomaly map for 6z, 22dec07 - Source: National Operational Hydrologic Remote Sensing Center (NOHRSC) JFM’08 was cold in the Intermountain West and Midwest with a warmish pocket in between (MT+). Except for New England, temperature anomalies match snowdepth anomalies quite well.

15. 2008-09 impact of anomalous snowpack Snowdepth anomaly map for 6z, 22dec08 - same scale as previous year! Source: National Operational Hydrologic Remote Sensing Center (NOHRSC) Cold winter from northeastern MT eastward across ND into WI and further east matches snowpack influence, but not La Niña nor OCN composites! Conversely, lack of snow in southern MT and WY helps OCN to ‘win’ against La Niña impacts. On the other hand, southern UT and southwestern CO lost their snowcover and cold air to end up warm in JFM’09 despite high snowpack early on. Conclusions from two winters: southern (<40N?) and coastal snowpack anomalies need ‘help’ to persist&exert influence!

16. U.S. precipitation anomalies based on OCN (15yr) Out of 344 climate divisions, 17 have an average standardized precipitation anomaly for JFM ‘94-08 that exceeds +0.43 sigma (~upper tercile), 28 are below -0.43 sigma. Most coherent wet (dry) trend is found in northeast (southeast) U.S.

17. U.S. precipitation anomalies based on OCN (15yr) In January-March 2009 (below), only 2 of the 17 flagged wet-trending climate divisions ended up in the upper tercile, but 20 of the 28 flagged dry-trending divisions verified in lower tercile. Total OCN-based skill score: +23. Total national HSS: +3

18. U.S. precipitation anomalies in La Niña winters Ten strongest La Niña cases (left): out of 344 climate divisions, 119 show composite anomalies in excess of 0.43 standard deviations (33% tails) vs. 67 below -.43. Dryness overlaps with OCN-flagged area in southeastern U.S. Moderate La Niña cases (MEI ranks 6-15; right): out of 344 climate divisions, 65 show composite anomalies above 0.43 standard deviations (33% tails) vs. 51 below -.43. Pattern correlation with strong La Niña cases: +0.80 - similar anomalies, just weaker.

19. U.S. precipitation anomalies in La Niña winters In January-March 2009 (top right), 21 out of 119 strong La Niña-flagged climate divisions (top left) and 12 out of 65 such divisions for weak-to-moderate La Niñas (left) end up in the upper tercile, while 62 out of 67 strong (46 out of 51 weak) La Niña dry divisions end up in the lower tercile. This translates into HSS: +17 (+25) for top 10 (ranked 6-15) La Niña-based precipitation forecasts. Total HSS: +9 (8)

20. U.S. temperature and precipitation anomalies in models driven by global SST (AMIP) in JFM’09 Modeled temperature anomalies (left) covered 190 climate divisions with anomalies in excess of +0.43 sigma (13 below -0.43). Observed anomalies matches the positive forecasts in 70 divisions (low skill), but 12 out of 13 negative divisions, yielding an overall non-EC HSS of +11 (national score: +6). For precipitation (right), only 20 climate divisions reached the negative 0.43 sigma threshold (none above +0.43), and matched the observations 18 times (HSS: +85; national score: +5).

21. Summary • CPC’s winter forecast scored better than last year - a combination of less ambition (smaller area coverage) and good use of OCN and La Niña information. • Compared to the previous winter, warming trends re-asserted themselves in much of the country, leaving OCN as the best temperature forecast tool this time around. GHG-based temperature projections for the U.S. were less well matched than either La Niña- or OCN-based forecasts. • La Niña was not as strong as in the previous winter, but provided for good guidance for cold temperatures and low precipitation anomalies. •In a rare almost-repeat of the previous year, fairly extreme snowdepth anomalies (both positive and negative) could have been used to tweak temperature forecasts in the right direction and avoid 2-category forecast misses which may be the most damaging to reputation. •Hindcast model forecasts for U.S. temperature and precipitation based on global SST did a very competitive job in predicting both fields.