Introduction to statistics used in seasonal forecasting

1
Introduction to statistics used in seasonal
forecasting

• Scott Power
• Bureau of Meteorology
• Ackowledgements
• D. Jones, P. Reid, R. Fawcett, NCC
• W. Drosdowsky, N. Nicholls, L. Chambers, O.
  Alves, BMRC

2
Statistics and Probabilities Averages and
means Percentiles and deciles Terciles
Seasonal predictions as probabilities
Distribution curves and chocolate wheels
3
Commonly used climate statistics
Mean or arithmetic average Adding up all values
and dividing by the number of values Eg, the
average monthly rainfall (MM) recorded on an
island from 1960 to 1990 was 121, 165, 98, 65,
32, 31, 34, 68, 95, 108, 132, 135 Average annual
rainfall between 1960 and 1990 was 1084 mm The
average monthly rainfall during this period was
1084/12 90.3 mm
4
The Median Here, we rank the rainfall amounts
from smallest to greatest 31, 32, 34, 65, 68,
95, 98, 108, 121, 132, 135, 165 The median
monthly rainfall is about 96.5 mm (a little
higher than the arithmetic mean) One half of the
months recorded rain above 96.5 mm and the other
half below 96.5 mm. This is also known as the
50th percentile.
5
Percentiles The 80th percentile, for example, is
that figure below which rainfall in 80 of the
months will lie. 31, 32, 34, 65, 68, 95, 98,
108, 121, 132, 135, 165 80 of 12 is 9.6 So
our 80th percentile is about 126 mm
6
Deciles The rainfall is said to be in decile 1
if it less than or equal to the 10th
percentile In decile 2 if less than or equal to
the 20th percentile but above the 10th
percentile In decile 3 if less than or equal to
the 30th percentile but above the 20th
percentile
7
Terciles Three possible outcomes for a
particular series of events For example, with
rainfall, it could be (a) Wetter than normal
range (7th to 9th decile) (b) Normal range (4th
to 7th decile) (c) Drier than normal range (1st
to 4th decile)