Geoscience Reference
In-Depth Information
Hourly average precipitation at manaus
0.6
0.5
0.4
0.3
0.2
0.1
Figure 13.2 Average diurnal
variation in rainfall for four
years (2000, 2002, 2003, 2004,
2005) at Manaus, Brazil.
0.0
12
34567891011121314151617
Local time (hours)
18 19 20 21 22 23 24
Trends in precipitation
Strictly speaking, the mean values of annual, seasonal, or monthly precipitation is
statistically useful only when the probability distribution they follow is normally
distributed. However, as mentioned above, over short periods such as a day, rainfall
data are always positively skewed. Trends and oscillations in average precipitation
over shorter periods are therefore hard to discern although it may be possible to
identify them in median values if there are enough samples in the sample period to
define these. More usually when seeking to identify trends and oscillations, the
approach is to define methods that involve some form of averaging or summing of
precipitation data over longer periods because this gives values whose probability
distribution is closer to normal.
To illustrate this by example, consider the data given in Table 13.1 which are the
values of the monthly total precipitation, P i , for July at Musoma, Tanzania between
1931 and 1960. These values follow the strongly skewed distribution shown in Fig. 13.3.
The ten year mean of July precipitation of these data, P M , for the periods 1931-1940,
1941-1950 and 1951-1960 are given in Table 13.2 along with the standard deviation,
s M , and coefficient of variation, C M , for each ten year period computed from:
(
PP
N
)
i M
2
(13.2)
s
=
M
and:
100
C
=
s
(13.3)
M
M
P
M
 
Search WWH ::




Custom Search