close

Вход

Забыли?

вход по аккаунту

?

469

код для вставкиСкачать
INTERNATIONAL JOURNAL OF CLIMATOLOGY
Int. J. Climatol. 19: 69–80 (1999)
THE CLIMATOLOGICAL REGIONS OF TANZANIA BASED ON THE
RAINFALL CHARACTERISTICS
C.P.K. BASALIRWAa,*, J.O. ODIYOb, R.J. MNGODOb and E.J. MPETAb
Makerere Uni6ersity, Department of Geography, PO Box 7062, Kampala, Uganda
b
Water Resources Engineering Programme, Uni6ersity of Dar-es-Salaam, PO Box 35131, Dar-es-Salaam, Tanzania
a
Recei6ed 1 May 1995
Re6ised 22 April 1998
Accepted 4 June 1998
ABSTRACT
In this paper, principal component analysis (PCA) was used to delineate the raingauge network of Tanzania into
homogeneous groups. The monthly rainfall records for the years 1961 – 1990 inclusive at 150 raingauge stations used
in the study were extracted from the records at the Directorate of Meteorology, Tanzania. The spatial patterns of the
rotated PCA dominant modes delineated Tanzania’s raingauge network into 15 homogeneous groups. Statistical tests,
climatological information, topographic features and other data supported the physical reality of the 15 delineated
groups. The delimited rainfall regions may be useful for Tanzania in agricultural planning, the assessment of water
resources potential, delineation of drought or flood risk zones and as a basis of ensuring collection of climatologically
representative rainfall data by the inclusion of a station(s) from each homogeneous rainfall region. Copyright © 1999
Royal Meteorological Society.
KEY WORDS: Tanzania;
East Africa; principal component analysis (PCA); rainfall regions; Intertropical Convergence Zone
(ITCZ); climatological regions
1. INTRODUCTION
One of the objectives of climate classifications is the identification of the spatial limits of different climate
types. Although the climate elements such as temperature, pressure, humidity, etc., vary, especially with
altitude and distance from the sea of a specific region or location, in the tropics, the element with the
highest variations in space and time is rainfall. Thus, in East Africa, climate classification schemes have
been based mainly on analyses of rainfall data (Griffiths, 1958, 1972; Johnson, 1962; Kagenda, 1975;
Ogallo, 1980, 1989).
Knowledge of the spatial extent of regions which have similar rainfall characteristics is advantageous in
the planning and management of not only rainfall dependent agricultural activities, but water resources
as well. In an equatorial region, rainfall would be associated with synoptic scale circulations; for example,
convergent low level winds in the Intertropical Convergence Zone (ITCZ) surface locations. However, in
East Africa, superimposed on the synoptic scale circulation patterns are meso-scale systems induced by
regional factors such as large water bodies and topographic features. Also, considering its latitudinal
position, East Africa does not experience much rainfall. Areas of 35%, 20%, 41% and 4% of East Africa
receive mean annual rainfall in 4 years out of 5 of less than 500, 500–750, 750–1250 and greater than
1250 mm, respectively (Griffiths, 1972). Thus, the rainfall patterns of the East African region are complex,
with rainfall amounts changing markedly over short distances such that there is no simple scheme based
on synoptic factors, such as the ITCZ, that can be used in the determination of the spatial extent of the
homogeneous rainfall regions.
Early studies of East African rainfall climatology used graphical mapping of areas with similar onset
and withdraw of the rains, similar seasonal/annual rainfall amounts etc. (Glover et al., 1954; Griffiths,
1958, 1972). Graphical rainfall mapping techniques are, however, tedious and time consuming.
* Correspondence to: Makerere University, Department of Geography, PO Box 7062, Kampala, Uganda.
CCC 0899–8418/99/010069 – 12$17.50
Copyright © 1999 Royal Meteorological Society
70
C.P.K. BASALIRWA ET AL.
Recent rainfall studies have, therefore, mostly relied on delineations of homogeneous rainfall regions
(zones) derived from empirical orthogonal functions (EOF) solutions, (Gregory, 1965, 1975; Dyer, 1977;
Morin et al., 1979; Ogallo, 1980, 1989; Nyenzi, 1992; Basalirwa, 1991). The principal component analysis
(PCA) studies on East African rainfall series, for example, those of Ogallo (1989), delineated ten
homogeneous regions over Tanzania from rotated PCA seasonal rainfall series while those of Nyenzi
(1992) showed the existence of four dominant seasonal rainfall regions over Tanzania with different
rainfall characteristics. However, both Ogallo (1989) and Nyenzi (1992) used sparse networks of rainfall
stations over Tanzania, and hence identified large scale regional climatic regions influenced by synoptic
systems. These regions could not adequately account for the meso-scale induced climatic differences.
Thus, this study attempts to derive homogeneous rainfall regions of Tanzania (an East African state)
using PCA based on a large network of 150 raingauge stations. The details of the methods are briefly
discussed after the next section.
2. AREA OF STUDY
Tanzania lies within 1 – 12°S and longitudes 29 –40°E; between the great East African lakes, namely: lakes
Victoria in the north, Tanganyika to the west and Nyasa to the south. To the east, lies the Indian ocean.
The country includes Africa’s highest point (Mount Kilimanjaro, 5950 m above sea level) and lowest part
(the floor of lake Tanganyika, 358 m below sea level). However, most of Tanzania, except the eastern
coastline lies above 200 m above mean sea level. Figure 1 depicts the relief of the country.
The total rainfall amounts for stations in Tanzania vary from year-to-year as well as having large
seasonal variations. The mean annual rainfall totals range from below 500 mm in the drier central areas
to just over 1000 mm in the wet areas, although the coastal region including the Islands of Zanzibar and
Pemba and parts of south-western Tanzania may receive over 1500 mm, (Griffiths, 1972). Hence,
climatological rainfall regions based on analyses of annual rainfall totals may be misleading for planning
purposes. Over most of Tanzania, the rains begin between mid-October and early December and continue
until May to early June, (EAMD, 1963; Alusa and Mushi, 1973; Mhita, 1990). Tanzania may be said to
have areas with two rainy maxima (bimodal) concentrated in the northern parts of the country and areas
with one long rainy (unimodal) seasonal rainfall distribution patterns found in the central and southern
regions.
3. DATA AND METHODS
In this study, monthly rainfall totals, within the years 1961–1990, inclusive at 150 stations over Tanzania,
whose approximate geographical locations are shown in Figure 2, were extracted from the records of the
Directorate of Meteorology, Tanzania. The few missing records in the data set were first estimated using
correlation and isopleth methods. The data set was then subjected to standard data quality control
techniques to remove any ambiguities that arise from observer error, changes in site, equipment, etc. and
declared near error free before PCA analyses. Details of estimation of missing records and data quality
control techniques can be obtained from Siegel (1956), WMO (1983, 1986), and Basalirwa (1991).
The PCA S-mode analyses first derived the inter-stations correlation matrices from the smoothed
monthly data series which have the advantage of weighting all the input stations equally which avoids
biasing the wetter stations with higher variability. The PCA provide criteria which are used to group the
locations with similar temporal rainfall characteristics.
The selection of stations in the initial PCA solutions was achieved by using a 4 × 4° square grid mesh
over Tanzania. Stations in each grid square were selected in a way that ensured an even representation of
the whole country taking into account the computing limitations of 44 stations that could be accommodated in the PCA analysis at a time. The initial PCA solutions identified locations that clustered on any
one factor or group of factors to form nucleus groups.
Copyright © 1999 Royal Meteorological Society
Int. J. Climatol. 19: 69 – 80 (1999)
CLIMATOLOGICAL REGIONS OF TANZANIA AND RAINFALL CHARACTERISTICS
71
Subsequent PCA solutions took a nucleus group of stations at a time to include all neighbouring
locations. If one dominant mode emerged then the group was considered as homogeneous. However,
where more than one dominant mode emerged, stations in the set clustering on a similar factor(s) were
identified and re-grouped. The locations were also re-grouped using vector space plots which helped to
determine their spatial relationship relative to each other. The data sets of each of the re-grouped
locations were then subjected to PCA separately until only one dominant PCA mode was extracted for
each group of locations. This is because, in principle, stations within any individual homogeneous region
should have the same factors(s) dominating. Hence, for any homogeneous group of stations, PCA
application to their data set independently should have only one factor extracted.
Boundary locations were included in the data sets of adjacent groups and PCA applied. Each boundary
location was then assigned to the group where its communality was highest. This is because the
Figure 1. The relief of Tanzania
Copyright © 1999 Royal Meteorological Society
Int. J. Climatol. 19: 69 – 80 (1999)
72
C.P.K. BASALIRWA ET AL.
Figure 2. The location of stations used in the study
communality of a variable is an indicator of the degree of its association with the other variables in the
set (see Child, 1970; Harman, 1976 for a discussion of communality of variables). In this way, all 150
stations were allocated to unique groups.
PCA solutions are advantageous over the graphical methods because of their flexibility and ability to
separate the complex variables based on the unique temporal characteristics of the individual locations.
The statistical significance of the results can also be determined (Child, 1970).
The number of significant principal components of the PCA solutions were determined using the Kaiser
criterion (Kaiser, 1959) and the scree test (Catell, 1966; Harman, 1976). The PCA solutions derive
orthogonal (independent) groups of variables. However, rainfall is influenced by synoptic and regional
factors. Thus, some degree of similarities (non-orthogonality) between rainfall patterns of neighbouring
locations in the different groups should be expected. Such similarities are not accounted for in the PCA
solutions. Therefore, some of the derived, PCA solutions may be physically unrealistic. This problem was
minimised through the rotation of the principal components which has been noted to reduce some
Copyright © 1999 Royal Meteorological Society
Int. J. Climatol. 19: 69 – 80 (1999)
CLIMATOLOGICAL REGIONS OF TANZANIA AND RAINFALL CHARACTERISTICS
73
ambiguities associated with the initial PCA solutions (Child, 1970; Harman, 1976; Richman, 1981, 1986).
Hence, the varimax method, where the significant principal components remain orthogonal during
rotation to an alternative position which better explains the data used.
The number of the PCA dominant modes underlying the spatial variance of the variables indicates the
number of climate types that may be expected. Statistical tests at 95% level of significance were used to
determine the significance of any variable’s loading on a PCA dominant mode. The spatial patterns of the
derived homogeneous raingauge groups delineated the homogeneous climatological rainfall regions.
Details of the PCA method are not given here but these can be obtained from Child (1970), Nie et al.
(1970) and Harman (1976).
Table I. PCA factor matrix for initial statistics at 44 locations
Variable
Communality * Factor
Eigenvalue
Variance (%)
Cumulative percentage
R002
R007
R011
R014
R005
R016
R034
R047
R040
R063
R078
R066
R100
R094
R101
R110
R107
R089
R062
R121
R147
R141
R118
R115
R130
R139
R125
R126
R122
R105
R103
R081
R072
R061
R090
R088
R108
R020
R019
R025
R070
R013
R030
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
1.00000
22.23604
6 29038
1.84728
1.12746
0.95891
0.78527
0.67708
0.56044
0.53078
0.49996
0.45443
0.43548
0.40889
0.40024
0.39172
0.36065
0.31340
0.30644
0.29374
0.28732
0.27025
0.26697
0.25131
0.25070
0.23226
0.22958
0.21009
0.19609
0.18348
0.17779
0.16698
0.15882
0.14805
0.13525
0.13208
0.12837
0.12339
0.11912
0.10298
0.09771
0.09473
0.08208
0.07600
51.7
14.6
4.3
2.6
2.2
1.8
1.6
1.3
1.2
1.2
1.1
1.0
1.0
0.9
0.9
0.8
0.7
0.7
0.7
0.7
0.6
0.6
0.6
0.6
0.5
0.5
0.5
0.5
0.4
0.4
0.4
0.4
0.3
0.3
0.3
0.3
0.3
0.3
0.2
0.2
0.2
0.2
0.2
51.7
66.3
70.6
73.3
75.5
77.3
78.9
80.2
81.4
82.6
83.6
84.7
85.6
86.5
87.5
88.3
89.0
89.7
90.4
91.1
91.7
92.3
92.9
53.5
94.0
94.6
95.1
95.5
95.9
96.4
96.7
97.1
97.5
97.8
98.1
98.4
98.7
98.9
99.2
99.4
99.6
99.8
100.0
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Copyright © 1999 Royal Meteorological Society
Int. J. Climatol. 19: 69 – 80 (1999)
74
C.P.K. BASALIRWA ET AL.
Figure 3. The scree test from PCA
Finally other verification methods which included the use of relief features and known rainfall climatic
patterns such as onset, withdraw and duration of the rains over Tanzania were used to determine the
reality of the delineated PCA rainfall patterns. In the next section, results and discussion from the study
are presented.
4. RESULTS AND DISCUSSION
Initial PCA results on 44 evenly selected different sets of stations (the maximum number that could be
handled by the computer in this analysis) chosen from all over Tanzania gave PCA dominant modes
indicating an expectation of at least 15 different climatic types.
Table I gives sample results from a preliminary PCA run. The Kaiser criterion of retaining factors with
eigenvalues greater or equal to 1 gave four significant eigenvectors. The variance explained by the first
four dominant modes was 51.7, 14.6, 4.3 and 2.6% for factors 1, 2, 3 and 4, respectively, giving a total
explained variance of 73.3%. Figure 3, the scree test derived from the results given in Table I, also shows
that only four dominant PCA modes may be retained in the rotation. In all further PCA, therefore, the
Kaiser criterion was used to determine the number of significant factors retained in the PCA varimax
rotations.
Figure 4 shows the PCA delineated homogeneous regions (zones) of Tanzania arbitrarily labelled A, B,
C, . . , M, N, P after subsequent use of PCA on all data sets, the final results from the varimax rotations
and the composite spatial mapping of the PCA significant modes groupings. The spatial boundaries of
these groups were derived using topographic features, the dominant wind flow patterns (Asnani and
Kinuthia, 1979), among other factors. Altogether 15 homogenous rainfall regions of Tanzania were
identified.
Table II gives three examples (zones B with 7 stations, E with 11 stations and K with 15 stations) of
the results obtained when the data at all the stations in an homogeneous group are independently
subjected to PCA. It can be observed that in each of the three cases, only one PCA dominant mode could
be extracted. The dominant PCA mode explained 70.8, 73.6 and 76.61% of the variance at each of the
three groups B, E and K, respectively. Similar results were obtained for each of the other delineated
rainfall regions. The uniqueness of each of the delineated zones was also reflected in the clustering of
vector-space plots and other statistical tests.
Copyright © 1999 Royal Meteorological Society
Int. J. Climatol. 19: 69 – 80 (1999)
CLIMATOLOGICAL REGIONS OF TANZANIA AND RAINFALL CHARACTERISTICS
75
By comparing Figures 1 and 4, it can be noted that zone B, for example, includes only those stations
surrounding lake Victoria, while zone E describes lee areas of the northern highlands, zone H the northern
coastal areas excluding Pemba, while regions K and L the central plateau and southern highlands,
respectively. Examples of the seasonal rainfall distribution patterns for stations in regions B, E, K and M
are given in Figures 5 – 8, respectively, for a selected highest communality station (the station that has the
highest spatial coherence with all the other stations in that group) in each zone.
In zone B (Figure 5), it can be noted that the seasonal rainfall distribution is bimodal with rains in
March–May with a peak in April of about 220 mm. The second rainfall period occurs from October–December with a peak in November of more than 150 mm. Zone E (Figure 6), has a similar seasonal rainfall
distribution pattern to that of B. However, it can be observed that for zone E, the rainfall amounts at E
are only ca. 50% of those at B for most of the year. This may be due to the regional influence of lake
Victoria in zone B which generates a lot of moisture from the land/lake breeze effects while at E, in spite
Figure 4. The homogenous climatological zones of Tanzania derived from PCA
Copyright © 1999 Royal Meteorological Society
Int. J. Climatol. 19: 69 – 80 (1999)
76
C.P.K. BASALIRWA ET AL.
Table II. PCA factor matrices for delineated zones B, E and K
Factor
Communality
Variance (%)
Zone B
R002
R003
R004
R005
R006
R014
R015
0.72036
0.88687
0.87575
0.89201
0.80534
0.84917
0.84568
0.51892
0.78654
0.76693
0.79569
0.64857
0.72108
0.71518
70.8
Zone E
R039
R041
R042
R043
R044
R045
R046
R047
R048
R050
R051
0.80861
0.88972
0.92298
0.79756
0.90223
0.69275
0.79644
0.89557
0.91560
0.84973
0.93303
0.65386
0.79161
0.85189
0.63610
0.81403
0.47991
0.63432
0.80205
0.83832
0.72205
0.87054
73.6
Zone K
R057
R058
R059
R060
R061
R071
R072
R073
R074
R081
R082
R083
R087
R090
R106
0.88743
0.89637
0.84189
0.88798
0.88987
0.87220
0.92638
0.92314
0.87971
0.88913
0.89540
0.83509
0.79205
0.80975
0.88644
0.78753
0.80349
0.70878
0.78850
0.79187
0.76073
0.85819
0.85219
0.77388
0.79056
0.80174
0.69737
0.62735
0.65569
0.78578
76.6
of orographic lifting, the rainfall amounts are less due to limited moisture in the prevailing easterly winds.
It is, however, clear that both regions B and E respond to the synoptic influence of the movements of the
ITCZ behind the overhead sun as evidenced by the double rainfall maxima when the ITCZ is in the area.
In zone K (Figure 7), the seasonal rainfall distribution indicates that the rains begin in late October and
continue until early May with a seasonal rainfall maximum of less than 150 mm occurring in December/
January. Only negligible amounts of rain occur between June and October. This may be due to the low
elevation of the zone, K lying 1000 – 1500 m above mean sea level. Hence rainfall in this zone is influenced
greatly by the movements of the ITCZ.
In zone M (Figure 8), the seasonal rainfall distribution reveals that there is one rainfall season
occurring between October and May similar to that of zone K. However, the peak occurring during April,
with a mean monthly rainfall total amounts in excess of 600 mm, almost five times the peak amount in
zone K. This may be a result of the fact that zone M is a ridge of about 3000 m above mean sea level
to the north of lake Tanganyika which plays a major role in rainfall enhancement from lake/land breeze
effects and from orographic lifting. The April rainfall peak in M, like the one of December/January is a
response of the seasonal rainfall to the ITCZ movements.
Copyright © 1999 Royal Meteorological Society
Int. J. Climatol. 19: 69 – 80 (1999)
77
CLIMATOLOGICAL REGIONS OF TANZANIA AND RAINFALL CHARACTERISTICS
Table III. Onset, duration and cessation of rains for sample of homogeneous rainfall zones of Tanzania
Long rains
Study case
no.
Delineated zone (bold)
and names of stations
Short rains
Pentad of
onset
Pentad of
cessation
Duration
Pentad of
onset
Pentad of
cessation
Duration
Ngara
58
28
43
61
71
10
Bukoba
Musoma
8
11
29
29
21
18
59
61
72
71
13
10
Mwanza
59
40
40
40
Loliondo
Mbulu
Arusha
62
63
11
28
27
29
37
37
28
61
1
13
Tanga
15
31
16
58
67
10
Dar-es-Salaam
Kilindoni
12
12
29
28
17
16
60
62
72
5
12
14
Mahenge
Nachingwea
Mtwara
68
64
66
26
24
26
33
33
33
26
31
6
Morogoro
63
28
38
Singida
Dodoma
Iringa
65
66
67
24
23
24
32
30
30
Songea
66
22
29
Mpanda
Chala
Mbeya
62
65
66
25
24
24
30
32
31
Kibondo
Mwadui
Kigoma
Tabora
61
66
61
63
27
28
26
24
39
35
38
34
A
7
B
2
3
C
11
D
16
26
38
G
79
H
101
110
I
117
143
148
J
84
K
57
81
106
L
140
N
80
103
112
P
18
24
52
69
Similarly the other regions can be identified with either characteristic relief features, synoptic factors,
regional factors or two or more of these as influences in their seasonal rainfall distribution patterns.
However, in all the PCA delineated regions, the seasonal rainfall distribution patterns have a characteristic response to the synoptic factors, especially the ITCZ movements. Some details about the other rainfall
regions are outlined in Odiyo (1994), Ogallo (1989) and Nyenzi (1992).
Table III, from the onset, duration and cessation of the rains in East Africa, adopted from Alusa and
Mushi (1973) for Tanzania’s stations indicates that the sample PCA delineated homogeneous rainfall
regions indicated have different pentads of onset, duration and withdraw of the rains. The table shows,
for example, that in region B there are two rainy seasons, the long rains occurring in pentads 8–29, while
the short rains occur during pentads 59 – 72 (taking January 1–5 as the first pentad). Table III further
Copyright © 1999 Royal Meteorological Society
Int. J. Climatol. 19: 69 – 80 (1999)
78
C.P.K. BASALIRWA ET AL.
Figure 5. The seasonal rainfall distribution at station 6 in zone B
shows that in zone P there is only one rainy season with the rains beginning about pentad 61 and lasting
until about pentad 27 of the following year. These differences in onset, duration and withdraw of the rains
in the various PCA delineated homogeneous rainfall regions indicate climatological differences in the
rainfall patterns of the delineated zones.
Thus, the 15 PCA delineated homogeneous rainfall climatological regions are consistent with the
statistical analyses, relief patterns, the rainfall climatological and graphical data of Tanzania.
Figure 6. The seasonal rainfall distribution at station 51 in zone E
Copyright © 1999 Royal Meteorological Society
Int. J. Climatol. 19: 69 – 80 (1999)
CLIMATOLOGICAL REGIONS OF TANZANIA AND RAINFALL CHARACTERISTICS
79
Figure 7. The seasonal rainfall distribution at station 72 in zone K
5. SUMMARY
It has been shown from the study that PCA solutions extracted 4 dominant PCA modes from the rainfall
records accounting for about 75% of the data variance. The PCA results delineated Tanzania into 15
homogeneous rainfall regions. The definition of the spatial extents of the PCA delineated rainfall zones
were based on a large network of 150 widely distributed rainfall stations.
The identification of regions with similar spatial and temporal rainfall characteristics is of significance
in the planning and management of rainfall dependent activities; for example, in determining planting
Figure 8. The seasonal rainfall distribution at station 129 in zone M
Copyright © 1999 Royal Meteorological Society
Int. J. Climatol. 19: 69 – 80 (1999)
80
C.P.K. BASALIRWA ET AL.
dates of seasonal crops for the different areas, the delineation of risk zones for drought forecasting or in
warning of a risk of floods. Homogeneous rainfall regions can also be used for the identification of
suitable areas for rain water harvesting and water storage. Finally, in the procurement of climatological
records for Tanzania, the adequacy of such climatological rainfall data samples necessitate the inclusion
of rainfall records of at least one station in each region.
ACKNOWLEDGEMENTS
The authors wish to acknowledge with thanks the financial support of the Irish Government through the
Water Resources Engineering Project coordinator, Dr R.K. Kachroo, the University of Dar-es-Salaam
Hydrology Section for the research facilities and all members of staff in the Hydrology Section, University
of Dar-es-Salaam who helped, in one way or another, to make this work possible.
REFERENCES
Alusa, A.L. and Mushi, M.T. 1973. ‘A study on the onset, duration and cessation of the rains in East Africa’, Reprints, International
Meteorological Meeting, Nairobi, 1974, pp. 133–140.
Asnani, G.C. and Kinuthia, J.K. 1979. Diurnal Variations of Precipitation in East Africa, Kenya Meteorological Depertment
Research Report, 8(79), 58 pp.
Basalirwa, C.P.K. 1991. Raingauge Network Designs for Uganda, Ph.D. Thesis, University of Nairobi, 1991, 235 pp.
Catell, R.B. 1966. ‘The scree test for the number of factors’, Multi6ar. Beha6. Res., 1, 245 – 276.
Child, D. 1970. Factor Analysis, Holt, Rinehert and Winston, 107 pp.
Dyer, T.G.J. 1977. ‘The assignment of rainfall stations into homogeneous groups. An application of principal component analysis’,
Q. J. R. Meteorol. Soc., 103, 1005–1013.
EAMD, 1963. Climatic Seasons of East Africa, East African Meteorological Department Report No. 8, 4 pp.
Glover, J., Robinson, P. and Henderson, J.P. 1954. ‘Provisional maps of reliability of rainfall in East Africa’, Q. J. R. Meteorol.
Soc., 80, 602 – 609.
Gregory, S. 1965. Rainfall o6er Sierra Leone, Department of Geography, University of Liverpool, Research Paper No. 2, 58 pp.
Gregory, S. 1975. ‘On the delimitation of regional patterns of recent climatic fluctuations’, Weather, 30, 276 – 287.
Griffiths, J.F. 1958. ‘Climatic zones of East Africa’, E. Afr. Agric. J., 23(3), 179 – 185.
Griffiths, J.F. 1972. ‘Eastern Africa’, in Climates of Africa, World Survey of Climatology, Vol. 10, Elsevier, 604 pp.
Harman, H.H. 1976. Modern Factor Analysis, University of Chicago Press, 487 pp.
Johnson, D.H. 1962. ‘Rainfall in East Africa’, Q. J. R. Meteorol. Soc., 88, 1 – 19.
Kagenda, A. 1975. ‘A factor analytic approach to delimitation of rainfall regions of Uganda’ E. A. Geogr. Re6., 13, 9 – 36.
Kaiser, H.F. 1959. ‘Computer program for varimax rotation in factor analysis’, Educ. Psych. Meas., 19, 413 – 420.
Morin, G., Fortin, J.P, Sochansk, W., Lardeau, J.P. and Charbonneau, R. 1979. ‘Use of principal component analysis to identify
homogeneous precipitation stations for optimal interpolation’, Water Res., 15, 1841 – 1850.
Mhita, M.S. 1990. ‘The analysis of rainfall data for agriculture in Tanzania’, Tanzan. Met. Res. Pub. (T.M.R.P.), 2(90), 61 pp.
Nie, N.H., Bent, D.H. and Hull, C.H. 1970. Statistical Package for Social Science (SPSS), Compiled by Glasgow University,
MacGraw-Hill, U.K., 343 pp.
Nyenzi, B.S. 1992. ‘An analysis of the interannual variability of rainfall over East Africa’, J. Afr. Met. Soc. (SMA), 1, 57 – 79.
Odiyo, J.O. 1994. Application of Principal Component Analysis in Delineation of Tanzania into Homoyeneous Rainfall Regions, M.Sc.
Thesis, University of Dar-es-Salaam, 155 pp.
Ogallo, L.J. 1980. ‘Regional classification of East African rainfall stations into homogeneous groups using the method of principal
component analysis’, Stat. Clim. De6el. Atmos. Sci., 13, 255 – 266.
Ogallo, L.J. 1989. ‘The spatial and temporal patterns of East African seasonal rainfall derived from principal component analysis’,
Int. J. Climatol., 9, 145–167.
Richman, M.B. 1981. ‘Obliquely rotated principal components. An improved map typing technique?’, J. Appl. Meteorol., 20,
1145 – 1159.
Richman, M.B. 1986. ‘Rotation of principal components’, J. Climatol., 6, 293 – 335.
Siegel, S. 1956. Non-Parametric Statistics for Beha6ioral Sciences, MacGraw-Hill, Kogakusha, Tokyo, 312 pp.
WMO. 1983. Guide to Hydrological Practices, WMO 168, Geneva, Switzerland.
WMO. 1986. Guidelines to the Quality Control of Surface Climatological Data, WCP-85, WMO/AD-No. 111, 56 pp.
Copyright © 1999 Royal Meteorological Society
Int. J. Climatol. 19: 69 – 80 (1999)
Документ
Категория
Без категории
Просмотров
4
Размер файла
290 Кб
Теги
469
1/--страниц
Пожаловаться на содержимое документа