Presentation is loading. Please wait.

Presentation is loading. Please wait.

Raw plume forecast data

Similar presentations


Presentation on theme: "Raw plume forecast data"— Presentation transcript:

1 Raw plume forecast data
Lead time↓ Raw plume forecast data 6 2 4 Obs

2 Correlation skill by target season and lead time

3

4

5 Bias by target season and lead time

6 Standard deviation ratio by target season and lead time

7 Correlation by lead time and target offset time
(all seasons)

8 Correlation by target season and lead time:
Real-time forecasts vs. longer-term hindcasts Real-time forecasts ( ) Hindcasts (~ )

9 9-year sliding correlation
Recent decade has lower predict- ability, not unlike the early 1990s. Attribution to low variability.

10 3-year sliding correlation
Performance was particularly poor more specifically during , similar to Attribution to low variability.

11 ==== ======= poor poor

12 Findings: 1. Gradual upward trend in ENSO prediction performance is outweighed by decadal variations in inherent ENSO predictability, based on signal-to-noise considerations. 2. During , dynamical models outperformed statistical models during the time of year prediction is most difficult. This performance difference is statistically significant. Future Goal: Developing a multi-model ensemble forecast for plume, based on multi-model ensemble means and/or or on multi-model individual ensemble members.


Download ppt "Raw plume forecast data"

Similar presentations


Ads by Google