Geoscience Reference
In-Depth Information
11.3 Methodology
In this study, monthly rainfall data of 13 districts of Orissa (Fig. 11.1) for 44-
year period (1960-2003) were collected from the Directorate of Economics
and Statistics, Government of Orissa. The district rainfall represents the average
of rainfall of all the administrative blocks of that district. The state revenue
department maintains one raingauge for each of the 324 blocks. In addition,
monthly maximum temperature (
T
max
), monthly minimum temperature (
T
min
),
monthly average relative humidity at 8:30 hours (
RH
fn
) and at 17:30 hours
(
RH
an
) for 16 meteorological stations of the study area for 15-year period
(1987-2001) were obtained from the report 'Climatological Data of Orissa
1987-2001' published by Directorate of Economics and Statistics, Government
of Orissa. The above-mentioned data were also available for two newly
established stations (ANG, TTG) at an industrial region for 8-year period
(1994-2001). For trend analysis, the two new stations having short records
were included for the spatial coverage and importance of the stations at the
industrial region. Trends in three climatic time series, i.e., annual rainfall,
maximum and minimum air temperature and relative humidity, were explored
by applying Mann-Kendall test and Spearman's rho test. Furthermore,
the trends were quantified by applying Sen's slope estimation technique
and homogeneity of the trends was tested by applying F
2
-based homogeneity
test.
The locally weighted scatter-plot smooth (LOWESS) procedure, a robust
nonparametric method for estimating the regression surfaces (Burn and Elnur,
2002; Broers and Grift, 2004), was used to draw the smoothed plots for the
annual rainfalls of the districts of Orissa. The Yule-Kendall skewness (
S
k
), a
resistance measure of the shape of the distribution (Ferro et al., 2005), was
calculated using the 25
th
percentile (
Q
1
), 50
th
percentile
(
Q
2
), and 75
th
percentile
(
Q
3
) as (
Q
1
- 2Q
2
+ Q
3
)(
Q
3
-
Q
1
)
-1
. The probability plot correlation coefficient
(PPCC) (Helsel and Hirsch, 1995) was computed for the test of normality of
the rainfall time series. Furthermore, the change in the linear relationship was
tested using the Chow's
F
-test (Wilby et al., 2004) with rainfall as the response
variable and year as independent variable between the sub-periods.
11.4 Results and Discussion
11.4.1 Trend and Variability in Annual Rainfall Time Series
Figure 11.2 shows the cases of differential annual rainfall pattern in different
physiographic regions of Orissa during 44-year period (1960-2003). The
anomalous events in terms of both drought and flood are increasingly more
conspicuous in recent years. The LOWESS plots and visual inspection of the
annual rainfall indicated a subtle shift in the rainfall pattern of most of the
Search WWH ::
Custom Search