Introduction


Part 1: Why Are Some Climate Variations Predictable At All?
+ Part 1: Sect 2
+ Part 1: Sect 3
+ Part 1: Sect 4
+ Part 1: Sect 5
+ Part 1: Sect 6
+ Part 1: Sect 7
+ Part 1: Sect 8
+ Part 1: Sect 9
+ Part 1: Sect 10
+ Exercise 1


Part 2: Using Models As Tools to Estimate the Predictability of Seasonal Climate
+ Part 2: Sect 2
+ Part 2: Sect 3
+ Part 2: Sect 4
+ Part 2: Sect 5
+ Exercise 2


Part 3: Seasonal Climate Forecasts: Basic Methods for Large-Scales and Downscaling
+ Part 3: Sect 2
+ Part 3: Sect 3
+ Part 3: Sect 4
+ Part 3: Sect 5
+ Part 3: Sect 6
+ Exercise 3


Part 4: Creating Information that can Better Support Decisions: Downscaling
+ Part 4: Sect 2
+ Part 4: Sect 3
+ Part 4: Sect 4
+ Part 4: Sect 5
+ Part 4: Sect 6
+ Part 4: Sect 7
+ Part 4: Sect 8
+ Part 4: Sect 9
+ Exercise 4


Conclusion
PART 4 : SECTION 2

This lack of predictability on sub-seasonal timescales is very important to understand in order consider one of the most often asked questions of atmospheric science: to forecast a few weeks in advance the onset of the rainy season. An ability to anticipate this would have enormous value for agriculture, allowing more reliable planting strategies. However, such a forecast requires at least the approximate timing of weather events. We discussed in the first lecture that there are theoretical reasons why weather events are not predictable beyond about 10-14 days. Therefore, this limits the potential for rainfall onset forecasts. There are two caveats to this statement:

(i) The rainfall onset phenomenon is often not a single synoptic event, but a cluster of events that may be linked to the large scale evolution of the atmosphere. Predictability of, for example, 5-day averages of rainfall with a lead-time of a few weeks in the tropics has not been extensively investigated, yet some empirical analyses with relatively new datasets are suggesting there may be sufficiently coherent large-scale atmospheric features that herald subsequent synoptic developments over a timescale of weeks, and this may be a fruitful avenue for research. However, such a major advance in thinking about atmospheric predictability will likely take time, even if it turns out to be possible.

(ii) In some situations, the control of the SST on the time mean circulation during the period of normal rainfall onset may be so strong as to modify the statistics of the onset date. This will only have information of substance in a few situations of extremely strong SST forcing. Examples may include parts of Indonesia and Central America, and the October-December rainy season in East Africa. However, more typically, the SST forcing will be very weak relative to the other factors that control the rainfall onset date, which in most situations may be internal to chaotic atmosphere dynamics and, even with a perfect atmospheric model, not predictable beyond the limit of deterministic weather predictability (10-14 days). In this typical situation, interannual variations in the rainfall onset date will show no relation with the prevailing SST and will have no potential predictability from the ocean-atmosphere coupling on seasonal timescales.

Previous Next