documentary reconstruction of monsoon rainfall variability over
TRANSCRIPT
Documentary reconstruction of monsoon rainfall variability over western India, 1781-1860
George C.D. Adamson and David J. Nash
G.C.D. Adamson
School of Environment and Technology, University of Brighton,Lewes Road, Brighton BN2 4GJ, UKe-mail: [email protected]
D.J. Nash
School of Environment and Technology, University of Brighton,Lewes Road, Brighton BN2 4GJ, UK
and, School of Geography, Archaeology and Environmental Studies,University of the Witwatersrand, Private Bag 3, Wits 2050, South Africae-mail: [email protected]
Abstract
Investigations into the climatic forcings that affect the long-term variability of the Indian summer
monsoon are constrained by a lack of reliable rainfall data prior to the late 19 th century. Extensive
qualitative and quantitative meteorological information for the pre-instrumental period exists within
historical documents, although these materials have been largely unexplored. This paper presents
the first reconstruction of monsoon variability using documentary sources, focussing on western
India for the period 1781-1860. Three separate reconstructions are generated, for (i) Mumbai, (ii)
Pune and (iii) the area of Gujarat bordering the Gulf of Khambat. A composite chronology is then
produced from the three reconstructions, termed the Western India Monsoon Rainfall
reconstruction (WIMR). The WIMR exhibits four periods of generally deficient monsoon rainfall
(1780-85, 1799-1806, 1830-1838 and 1845-1857) and three of above-normal rainfall (1788-1794,
1813-1828 and 1839-1844). The WIMR shows good correspondence with a dendroclimatic drought
reconstruction for Kerala, although agreement with the western Indian portion of the tree-ring
derived Monsoon Asia Drought Atlas is less strong. The reconstruction is used to examine the
long-term relationship between the El Nino-Southern Oscillation (ENSO) and monsoon rainfall over
western India. This exhibits peaks and troughs in correlation over time, suggesting a regular long-
term fluctuation. This may be an internal oscillation in the ENSO-monsoon system or may be
related to volcanic aerosol forcings. Further reconstructions of monsoon rainfall are necessary to
validate this. The study highlights uncertainties in existing published rainfall records for 1817-1846
for western India.
Keywords: Summer monsoon; documentary reconstruction; ENSO; western India
1
1
23456789
101112131415161718192021
222324
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
1 Introduction
Unravelling the climatic forcings that drive variability in the Indian monsoon is of vital importance
given the fundamental role of monsoon rains in the economy of the subcontinent. Current thinking
holds that monsoon rainfall variability is affected by internal intraseasonal and interannual
fluctuations (Gadgil and Srinivasan 1990; Krishnan et al. 2000; Annamalai and Slingo 2001; Gadgil
2003), as well as teleconnections with tropical/subtropical atmospheric-ocean circulation patterns
such as the Indian Ocean Dipole (IOD) (Saji et al. 1999; Ashok et al. 2001; Ashok and Saji 2007)
and El Niño-Southern Oscillation (ENSO) (Cole et al. 2000; Krishnamurthy and Goswami 2000;
Fasullo and Webster 2002; Goswami and Xavier 2005; Krishna Kumar et al. 2006; Lim and Kim
2007). Understanding the exact nature of these relationships on interdecadal timescales is
important in order to enable robust long-term forecasting. However, this is constrained by an
absence of reliable meteorological information from the subcontinent before the late 19 th century.
For example, debate is currently ongoing as to the nature of the relationship between ENSO and
monsoon rainfall. Whilst some studies suggest a weakening in the ENSO-monsoon relationship
due to recent warming (Krishna Kumar et al. 1999; Kinter et al. 2002), others point towards longer-
term fluctuations, with periods of unusually strong coupling from c.1885-c.1910 and c.1965-c.1980
and weaker correlation during other periods (Torrence and Webster 1998; Maraun and Kurths
2005; Robinson et al. 2008). It is not possible to reconcile these arguments when dealing with a
dataset for India as a whole that goes back no further than 1871.
The first systematic rainfall observations in India started at Madras in 1813. However, by the
time meteorological records began to be collected at the Colaba Observatory in Bombay in 1843,
the number of rain gauges in India was only 11 and a national gauging network did not appear until
1871 (Sontakke et al. 2008). Regional rainfall reconstructions prior to 1871 (Sontakke et al. 2008)
rely on statistical inferences using this small network of gauges. Climatic reconstruction using
natural proxies is also problematic within peninsular India. Reconstructions using ice-cores are
feasible only within the Himalayan regions, and dendroclimatological investigations are challenging
due to a lack of clearly defined growth rings in the majority of indigenous species. Climatic
reconstruction using teak is being developed, but this research is still only in its infancy (see
Bhattacharyya and Yadav 1999; Ram 2011).
The reconstruction of historical rainfall levels in India is possible using documentary sources
(see Nash and Endfield 2002, 2008; Nash and Grab 2010; Nicholson et al. 2012). Europeans
resident during the late 18th and early 19th centuries, including representatives of the British East
India Company (EIC) and western missionary societies, recorded a wealth of climatic information
on a non-systematic basis. This ranged from detailed weather diaries to ad hoc climatic
observations in personal letters and commercial/government records, reflecting the fascination with
tropical climates exhibited by colonists (Grove 1997; Harrison 1999; Endfield and Nash 2002;
Adamson 2012). Much of this information survives within archives in the UK, India and USA, but,
2
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
with few exceptions (e.g. Pant et al. 1993; Walsh et al. 1999; Adamson and Nash 2012), has been
unexplored as a climatic resource.
This study extends the climatic record for the subcontinent by synthesising meteorological
information contained within historical documents to produce a semi-quantitative reconstruction of
monsoon rainfall variability in western India (Figure 1) for the period 1781-1860. Western India has
been selected as the study area for two reasons. First, abundant historical English-language
sources are available, including newspapers, materials relating to the EIC and, from 1823,
extensive documentation written by missionaries. Second, rainfall levels in the meteorological
subdivisions of ‘Konkan and Goa’ and ‘Gujarat’ are highly correlated at interannual timescales with
Niño-3.4 sea surface temperature (SST) anomalies (correlation coefficient = ~0.45; Parthasarathy
et al. 1993), which are understood to be a driver of rainfall variability over India (Krishnamurthy and
Goswami 2000; Fasullo and Webster 2002; Krishna Kumar et al. 2006; Lim and Kim 2007).
In this paper, seasonal monsoon rainfall indices are derived using historical documentary
materials and calibrated against available instrumental rainfall data. Indices are generated for three
reconstruction areas: (i) Mumbai, (ii) Pune, and (iii) an area of southern Gujarat bordering the Gulf
of Khambat (hereafter referred to as the ‘Gulf of Khambat’). The three reconstructions are
combined to produce a composite chronology for the entire study area, termed the Western India
Monsoon Rainfall reconstruction (WIMR). The WIMR chronology is compared to two
dendroclimatic studies and a regional rainfall reconstruction based on statistical inferences from a
small network of rain gauges. We conclude with a consideration of the implications of our results
for the understanding of ENSO-monsoon dynamics over time.
2 Climatology of western India
The rainfall regime of western India is dominated by the southwest monsoon. Winter circulation in
the region is characterised by lower-tropospheric northeasterlies, producing a net continent-ocean
flow. Thermal insolation generates a lower-tropospheric heat low over the Thar Desert to the north
of the study area during March-May (Gadgil 2003), which, together with the migration of the
Tropical Convergence Zone, drives the monsoon (Gadgil and Srinivasan 1990; Srinivasan et al.
1993; Annamalai and Slingo 2001; Lawrence and Webster 2001). Monsoon onset occurs in
Mumbai on 10 June (1781-2011 mean; Adamson and Nash 2012) as the zone of maximum
convection migrates northwards from its winter position at 0°-5°N to its mean summer latitude of
20°N (Gadgil 2003). Rains may occur before this date, particularly in coastal regions, due to the
action of onset vortices formed over the Arabian Sea (Mooley and Shukla 1987).
The monsoon season generally lasts until late-September or mid-October. Rainfall in western
India is sourced predominantly from subtropical cyclones, which develop over the northeastern
Arabian Sea and are driven by a low-level westerly jet, the cross-equatorial Somali jet, which forms
in late May (Mohanty et al. 2005). Deflection of air over the Western Ghats causes conditional
3
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
instability (Grossman and Durran 1984), resulting in an area of rainfall maximum located just off
the west coast (Krishnamurti et al. 1983). Mumbai lies within this maximum and receives around
2000mm rainfall per year, predominantly falling during the monsoon months (Bhowmik et al. 2008).
The Deccan region is situated in the rain shadow zone to the east of the Western Ghats and
receives significantly lower rainfall than adjoining coastal areas (Gunnell 1997). Precipitation is less
regular in the Deccan than in coastal regions, although this phenomenon is more marked to the
south of the study area (Gunnell 1997). Rainfall in northern parts of the study area originates from
depressions formed in the Bay of Bengal, which are transported across the subcontinent by the
upper easterlies (Gadgil 2003). These form twice a month on average and propagate in a westerly
direction, with a lifespan of approximately 5 days (Yoon and Chen 2005). Depressions are weak by
the time they reach the study area, resulting in low average rainfall in this region (~550mm per
year; GHCN 2010).
During the monsoon season, western India experiences ‘active’ periods of rainfall separated by
‘break’ periods of several days duration when little or no rainfall occurs. The timing and duration of
break periods is subject to the position of the zone of maximum convection. This zone is normally
located at a mean latitude of 25°N during the monsoon season, but periodically migrates
northwards to 30°N, bringing heavy rainfall to the southern Himalaya and reduced precipitation
over the remainder of the country (Krishnamurthy and Goswami 2000). Break periods are also
generally associated with the production of new zones of maximum convection at 5°N, which then
commence a northward propagation and lead to a return to active conditions (Gadgil 2003).
Easterly anomalies (Krishnamurthy and Shukla 2000), and a weakening in lower tropospheric
vorticity (Goswami et al. 2003), result in 75-85% departures from the long-term average rainfall
during break periods in the study region (Gadgil 2003). The region of the Western Ghats is the first
area to experience increased rainfall at the recommencement of active periods (Krishnan et al.
2000; Krishnamurthy and Shukla 2007).
3 Data sources
The locations of the main archives consulted for this study, together with the coding used to cite
individual sources in the text, are provided in Table 1. The archive consulted most extensively was
the India Office Records collection housed in the British Library, St Pancras, London, UK. This
archive contains the records of the EIC, which comprises of copies of correspondence and minutes
sent either as annual reports or synthesised into volumes relating to specific events. Three
volumes on drought conditions were particularly useful, covering droughts in 1812-13, 1823-25 and
1832-33. A travel diary by a Dr Anton Hove, entitled ‘Tours for Scientific and Economical Research
of Guzerat-Kattiawar and the Conkuns, in 1787-88’ (BL IOR/V/22/212 No. 16) was also consulted
in depth. The British Library at St Pancras additionally contains an extensive collection of Private
Papers, which includes letters, diaries, scrapbooks and personal business records from individuals
or families who were resident in India during the colonial period. All Private Papers containing
4
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
materials relating to western India during the study period were consulted, comprising thirty
volumes.
The British Library Newspaper Collection, Colindale, London, was also consulted. This archive
contains microfilmed back-issues of English-language newspapers from Mumbai. The earliest
available newspaper was the Bombay Gazette, which has issues stored from August 1792,
although the holdings of this publication are incomplete. The primary newspapers used for this
study therefore were the Bombay Courier and Bombay Times. The Courier was published from
1793 until 1846. Meteorological information within the Courier was sparse during the last two years
of its publication, so from 1844 to 1860 the Times was used as the main newspaper source. For
years where meteorological information was lacking within these publications, or where issues
were missing, issues of the Bombay Gazette and later the Bombay Monthly Times and Bombay
Standard were consulted. The publication frequency of the various non-monthly newspapers
increased almost simultaneously, rising from weekly in 1793 to twice weekly in 1835, thrice weekly
from 1840 and daily from 1850. All newspapers followed generally the same format, with stories
often repeated in several publications. The quality of meteorological information and the
terminology adopted was therefore very similar between publications. From 1833, all newspapers
included identical fortnightly State of the Weather and the Crops reports (see Table 2).
A number of other collections of historical documents were explored in addition to materials held
in the British Library. These included the archives of three western missionary associations: the
Church Missionary Society, American Board of Commissioners for Foreign Missions and Scottish
Missionary Societies (Table 1). Documentation within these archives consists of reports and letters
from missionaries in the field to their respective central offices, together with some private
correspondence. The archives of the Government of Maharashtra, located at Elphinstone College,
Mumbai, India, were consulted for the records of the colonial Government of the Bombay
Presidency. These comprise predominantly of letters, minutes, official proclamations and circulars,
and petitions from the local population. Miscellaneous materials within the National Archives of
Scotland, and the archives of the Royal Society, London, and Aberdeen Medico-Chirurgical
Society were also analysed.
Some previously published materials, mainly scholarly articles published during the late 18th and
early-mid 19th centuries, were used as primary sources. These included two weather diaries that
appeared in the Transactions of the Royal Society (Banks 1790; Sykes 1835) and a weather diary
and famine report published in the first edition of the Transactions of the Literary Society of
Bombay (Carnac 1819; Nicholls 1819). A synthesis of information on famines in colonial western
India, the Report of Past Famines in the Bombay Presidency, was the only secondary source
utilised (Etheridge, 1868). This was originally collated from oral and written records, and was
published eight years after the end of the study period.
4 Methodology5
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
4.1 Data collection and sorting
Information on climate or climate-dependent phenomena within each historical source was
recorded verbatim. This included direct references to climatic conditions, as well as reports of
droughts and floods. Reports of harvests yields were also recorded as a broad indicator of
seasonal rainfall levels, particularly those included within the State of the Weather and the Crops
sections of Mumbai newspapers from 1833 onwards (see Table 2). For each observation, the
author, place of publication, location referred to, date written, date (range) referred to, and recipient
(if applicable) were recorded in a central database. Information included within individual sources
ranged from general comments on the conditions during a season to daily weather accounts within
systematic or semi-systematic weather diaries. Instrumental temperature and pressure readings
were also sometimes included. Certain documents were used extensively for climatic
reconstruction due to the quality and quantity of information recorded (Table 2).
Recorded information was sorted into monthly blocks by location. Climatic reconstruction was
undertaken for three areas (Figure 1). The ‘Mumbai’ region comprises approximately the
contemporary area of Greater Mumbai, including the 19th century Bombay Island together with
Salsette and Colaba. The region of ‘Pune’ comprises the administrative district of Pune, including
the present day city. The ‘Gulf of Khambat’ region incorporates the administrative divisions of
Surat, Bharuch, Vadodara, Anand, Kheda, Ahmadabad and Bhavnagar that surround the Gulf.
Where no other data were available, information from the peninsular of Kathiawar was also used
for the Gulf of Khambat reconstruction. The number of datapoints (i.e. individual quotations) used
for monsoon reconstruction for each of the three areas, together with the number of sources from
which these quotations were derived, is summarised in Figure 2.
Quotations were used to produce summaries of rainfall conditions during the monsoon for each
of the reconstruction areas. Four ‘monthly’ summaries were created for the monsoon period:
May/June; July; August; and September/October (hereafter referred to as the ‘rainfall months’).
May and June were combined to allow for fluctuations in the date of monsoon onset, which
generally occurred during early-mid June but occasionally in late May (cf. Adamson and Nash
2012). Rainfall associated with cyclonic activity before the date of onset was not included.
September and October were combined to take into account variations in the end-monsoon date.
An example of a monthly summary table for the Mumbai reconstruction area during the monsoon
season of 1853 is provided in Table 3.
4.2 Generation of calibration tables
Calibration of the reconstruction was undertaken where reliable instrumental data overlapped with
the documentary record. For Mumbai, the Global Historical Climatology Network (GHCN) publishes
homogenous monthly instrumental data collected at the Colaba Observatory from 1847. Data
spanning the monsoon months only are also available from 1817; however, our archival
6
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
explorations have revealed that these data may be unreliable. The data derive from a network of
rain gauges across Mumbai, published in newspapers; published monthly data are not always
derived from the same gauge and in some years are averages of several gauges. Furthermore, no
evidence exists as to the ways in which these data were collected or the instruments used. The
data are therefore likely to contain inhomogeneities. As a result, a 13-year calibration period was
selected for the Mumbai reconstruction, encompassing 1847-1859. For Pune, the GHCN publishes
homogenous instrumental rainfall data collected at the Poona Observatory from 1856. Five years
of rainfall data are also available from 1826-1830. These were collected as part of Colonel Henry
Sykes’ reports on the Statistic of the Deccan (BL MSS Eur K388) and can be assumed to be
reliable with a high degree of confidence. A 10-year calibration period was therefore selected for
the Pune reconstruction, comprising the years 1826-1830 and 1856-1860. No instrumental data
are available for the Gulf of Khambat during the study period so calibration was not possible for
this region.
Calibration was achieved by first assigning a 5-point monthly rainfall index value – ranging from
-2 (‘scanty rainfall’) to +2 (‘heavy rainfall’) – to each month within the two calibration areas. Indices
were assigned based on instrumental data, using categories adopted by the Indian Meteorological
Department (IMD) to describe rainfall variability at short intervals (Table 4). The boundaries of each
category were delimited by percentage deviations from a Long Period Average (LPA) of monthly
rainfall. The LPA was calculated using data from 1847-1950 for Mumbai and from 1826-1830 and
1856-1950 for Pune. This is different to the LPA currently used by the IMD – an average of the last
30 years – and was selected in order to (i) incorporate the calibration periods and (ii) avoid any
recent changes in the rainfall regime which may have occurred due to anthropogenic warming
(Robertson et al. 2001) or multidecadal fluctuations in rainfall intensity (Parthasarathy et al. 1987;
1994; Guhathakurta and Rajeevan 2008). These LPAs are used for all further analyses in this
paper.
Documentary information for rainfall months falling within each index class was analysed to
determine key descriptors and conditions associated with each category. For example, in Mumbai,
five rainfall months during the calibration period were classified as +1 (excess rain). Content
analysis of documentary material revealed that this particular rainfall class was associated
regularly with (i) the descriptor ‘seasonable rain’, (ii) slight crop damage due to flooding, and (iii)
periods of heavy rainfall interspersed with drier intervals. This information was used to generate
calibration tables detailing the predominant rainfall properties and descriptors associated with each
index category. The calibration table for Mumbai is presented in Table 5. The calibration table for
Pune was the same as for Mumbai, with the exception that ‘moderate/more or less’ was used
regularly by observers to refer to average rainfall, ‘satisfactory’ to above average rainfall, and
‘sufficient’ to excess rainfall. As no calibration was possible for the Gulf of Khambat, it was
assumed that descriptors for this reconstruction area had the same meaning as for Mumbai, since
7
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
the climatic regime in southern Gujarat is more similar to that of Konkan than the western Deccan
Plateau (Gadgil and Joshi 1983; Gunnell 1997).
4.3 Classification of rainfall across the reconstruction period
Where meteorological descriptions within the documentary materials were of sufficient quantity and
quality, the calibration tables were used to classify rainfall months across the reconstruction period.
Full monthly reconstruction was possible from 1833 onwards. Prior to this date, some years had
either very few quotations, or quotations that were not sufficiently detailed to allow for classification
of rainfall intensity at the monthly scale. Coverage was particularly sparse prior to 1811.
The derived monthly classifications were used to generate seasonal classifications (May-
October), in order to represent rainfall variability during an entire monsoon. The IMD uses a 5-point
index scale to classify rainfall at the seasonal scale, referred to as the 5-Parameter Statistical
Ensemble Forecasting System (5-PSEFS; Table 6). Like the weekly/monthly rainfall descriptors
shown in Table 4, the 5-PSEFS ranks rainfall according to deviation from a LPA, although the
percentage brackets are narrower to reflect the lower variability of rainfall at the seasonal scale. In
order to maintain continuity with IMD terminology, a methodology was devised to convert monthly
indices to seasonal. Monthly rainfall data within the instrumental period (post-1847 for Mumbai,
post-1856 for Pune and post-1871 for the Gulf of Khambat) were first “degraded” into the 5 monthly
classes shown in Table 4. This was achieved by assigning descriptive rainfall classes for each
month according to the deviation of rainfall from the LPA for that month. Seasonal (MJJASO)
rainfall within the instrumental record was also degraded into the classifications used in the 5-
PSEFS. (Where this approach is used in this paper, the resulting monthly and seasonal rainfall
data are referred to as “degraded instrumental data”.)
Several methods were attempted to determine the optimal relationship between the monthly
and seasonal classes. The method that produced the highest correlations involved summing
monthly values and assigning a seasonal classification accordingly. Where the summed monthly
values were -1, 0 or +1, these classifications were assigned to the season overall; years with
summed values of ≥ +2 or ≤ -2 were assigned a category of +2 or -2 respectively. This approach
produced Pearson’s correlation coefficients of 0.79 for Mumbai and 0.85 for Pune (significant at
99%).
For some years, meteorological descriptions were insufficient in number and/or detail to allow
for reconstruction using the monthly calibration table. Quotations, however, often existed that
described conditions across the entire season. For example, only one quotation is available for the
monsoon at Pune in 1813. This is from Mountstuart Elphinstone, a reliable observer (Adamson
2012), and states that the rains were “the heaviest ever known here” (BL MSS Eur F88/370, 21
September 1813). Likewise, no contemporary reports are available for the Gulf of Khambat relating
to 1790. However, the Report of Past Famines in the Bombay Presidency mentions famine in five
8
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
locations in southern Gujarat in 1790/91 due to lack of rain (Etheridge 1868). A second set of
calibration tables was therefore created in order to provide rainfall classifications for these years.
These were generated in the same way as the monthly calibration tables, but were compared
against degraded instrumental data for the whole monsoon season using the percentage
deviations listed in Table 6. The seasonal calibration table for Mumbai is provided in Table 7.
No documentary data were available for the period 1780-1786. Instrumental rainfall
measurements exist for this period within a weather diary presented to the Royal Society in 1790
(RSCHS MA.213). Seasonal monsoon indices were derived for 1781-1786 by degrading the data
using the percentage boundaries in Table 6. Rainfall measurements were taken only from the
onset of the monsoon (which varies in each year; Adamson and Nash 2012) until the withdrawal,
generally in mid-October. Furthermore, little information is provided on the design of the instrument
or on the methods and times of data collection. It is possible therefore that the instrumental data
may be unreliable. 1780 was excluded as the rain gauge was not set until 4 July 1780.
4.4 Derivation of composite Western India Monsoon Rainfall reconstruction
A composite reconstruction, the WIMR, was produced for the entire study area in order to
represent monsoon variability across western India. To achieve this, the seasonal monsoon indices
derived for each of the three reconstruction areas were assigned a representative percentage
rainfall value. The indices +1, 0 and -1 were assigned values equivalent to the mid-point of their
respective 5-PSEFS percentage ranges (Table 6), i.e. 107%, 100% and 93% respectively. Years
classified as +2 and -2 were assigned a value equivalent to the average percentage deviations
corresponding to these classes within instrumental records for the LPA. These were found to be
132% for the class +2, and 69% for the class -2. For the Gulf of Khambat, the LPA was taken as
1871-1950 (due to data availability) and the instrumental data used was an average of data for the
principal meteorological stations within the region (Surat, Vadodara, Bharuch, Kheda, Ahmedabad,
Bhavnagar and Rajkot).
For each year, an average of the three reconstruction areas was generated, weighted according
to mean MJJASO rainfall totals during the LPA (1797mm for Mumbai, 641mm for Pune and
811mm for the Gulf of Khambat; GHCN 2010). Categories for Mumbai were assigned a weighting
of 0.55, Pune 0.20, and the Gulf of Khambat 0.25. Where reconstruction had not been possible for
all three stations, the composite index was based on the available data only. The final result was
then re-assigned into one of the five seasonal rainfall classes based on the percentage deviations
specified for the 5-PSEFS.
4.5 Methodological process for selected years
The following section is designed to illustrate the methodological process. Years from each rainfall
category are presented to show how individual rainfall months were classified and combined to
form the three regional reconstructions and WIMR. The five selected years span the full 9
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
reconstruction period and illustrate how variations in data quality and quantity were addressed.
Monthly and seasonal rainfall classifications for each of these years are presented in Table 8,
together with the associated WIMR category.
4.5.1 Deficient monsoon year: 1838
An example of a deficient (-2) monsoon year is 1838, when significantly below average rainfall
occurred in all three reconstruction areas. The months of May/June contained reports of a “want of
rain” at Surat and Bharuch and rainfall of an “unsatisfactory nature” in Kheda within the State of the
Weather and the Crops (BLNC MC 1112, 19 July 1838). Rainfall at the Gulf of Khambat was
therefore classified as scanty. A “favorable appearance” was reported at Pune and “very favorable
accounts” from Thane, resulting in a classification of normal in Mumbai and excess in Pune (BL
MC 1112, 19 July 1838). Mumbai and the Gulf of Khambat recorded scanty rainfall in July, with
deficient rain reported in Pune. The Bombay Courier reported in mid-July “we regret extremely to
learn that not a drop of rain has fallen at Rajkote” (BLNC MC 1112, 10 July 1838). In Surat it was
reported that “the rice was suffering very seriously” (BLNC MC 1112, 4 August 1838).
Documents indicate that rainfall increased slightly during August in Mumbai and the Gulf of
Khambat (normal in Mumbai, deficient in the Gulf of Khambat), although in Pune it was reported
that “rain was urgently required every where in the Principal and Sub-Divisions of this Zillah, and
gloomy apprehensions were entertained of the serious consequences of a drought” (BLNC MC
1112, 11 September 1838). Scanty rainfall was again reported in Mumbai and the Gulf of Khambat
in September and October, although “most abundant” rains were recorded in Pune in early October
(BL MC 1122, 5 October 1838), resulting in a classification of normal rain. On 3 October the
Bombay Gazette reported "there is every probability of an almost total failure of the crops of every
description, in consequence of the long continued drought" (BL MC 1122, 3 October 1838).
4.5.2 Below normal monsoon year: 1818
The year 1818 was classified as a below normal (-1), based on average rainfall at Mumbai and
deficient rainfall at Pune. Rainfall at Mumbai was variable throughout the season, with scanty
rainfall in May/June, deficient in July, heavy in August, and excess in September/October. On 4 th
July, the Bombay Courier stated that the rainfall in June “does not amount to one half of what fell in
June last year” (BLNC MC 1112, 4 July 1818). In early August, the Courier stated that the weather
had been “much too fine for this season of the year” (BLNC MC 1112, 1 August 1818). In early
September, however, the Courier reported “abundant” rains in Thane and flooding in Panvel (BL
MC 1112, 5 September 1818), resulting in a classification of excess rainfall for September/
October.
In the Pune district, no information is available for May/June, although in mid-July it was
reported in Karandi that no rain had yet fallen (BL MSS Eur F88/363, 11 July 1818), and in Shirur
no rain had been recorded by 1 August and “the cattle were dying for want of forage” (BLNC MC 10
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
1112, 1 August 1818). This suggests deficient/scanty rainfall for May/June and July. Few reports
are available for August, although the Courier reported a fall of rain in “Deckan” which contributed
to a diminution in reported cases of cholera. No further rainfall was reported for
September/October, and a famine was recorded in late 1818 in Solapur, immediately to the east of
the Pune district (Etheridge 1868). This indicates overall deficient rainfall for the season at Pune.
No reports are available from the Gulf of Khambat.
4.5.3 Normal monsoon year: 1853
The 1853 monsoon season was classified as normal (0), based on heavy rainfall during the first
half of the monsoon season and deficient during the second, with a similar pattern repeated across
all reconstruction areas. An observer in Bombay noted on 20 June that “all the flat parts of the
island are under water” (BL MSS Eur Photo 431, 20 June 1853), and the Bombay Times reported
on the same day “we seldom remember a heavier fall of rain in June than we have had during the
past ten days” (BL SM 73, 20 June 1853). The State of the Weather and the Crops likewise
reported “heavy” and “excessive” rain in Gujarat, with reports of flooding and crop damage in
Ahmedabad, Surat and Kheda (BL SM 73, 19 July 1853). Rainfall in Pune towards the end of June
was categorised as “every where abundant, and in some places, excessive” (BL SM 73, 26 July
1853). Rainfall in May/June was therefore classified as heavy in all areas. Accounts from July also
indicate heavy rainfall, with “plentiful”, “heavy”, and “unceasing” rain reported in the Gulf of
Khambat (State of the Weather and the Crops reports; BL SM 73, 2 August 1853). Rainfall from 10
to 13 July in Pune was reported as being the heaviest remembered by “that most sapient of sages
the ‘oldest inhabitant’” (BL SM 73, 18 July 1853). “Seasonable” and “plentiful” rain was reported in
Thane and Colaba (BL SM 73, 19 August 1853), resulting in a classification of excess rain in
Mumbai, with rainfall classified as heavy over Pune and the Gulf of Khambat.
“Little rain” was reported in Pune in August (BL SM 73, 3 September 1853), with “partial” rain in
Thane and reports of subsequent crop damage (BL SM 73, 23 September 1853). Likewise
“deficient” rainfall was reported at Ahmedabad, with “fair” conditions over Kheda and reports of
crop damage due to drought (BL SM 73, 23 September 1853). Rainfall was therefore classified as
scanty in all regions in August. Similar accounts of very low or no rainfall were reported in the State
of the Weather and the Crops reports during September/October, with accounts of crop damage
across all three reconstruction areas (BL SM 73, 3 November 1853 and 9 November 1853). A
classification of scanty rainfall was therefore assigned for all regions during September/October.
4.5.4 Above normal monsoon year: 1793
Although few records are available, 1793 was classified as an above normal (+1) monsoon year.
Reconstruction for Mumbai was based predominantly on one quotation from the Bombay Courier,
which stated: "The late favourable Monsoon among its other benefits, must be attended with
peculiar advantage to the Speculative Cultivators of Salsette” (BLNC MC 1112, 19 October 1793).
11
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
Two further articles within the Courier mention “favourable” weather (BLNC MC 1112, 27 July 1793
and 3 August 1793), and a letter dated 4 August mentions the recent “Rainy and Stormy” weather
(NLS MSS.19207, 4 August 1793). The reconstruction for Pune is based on only one quote, also
included within the Bombay Courier of 19 October. This article states “by intelligence from Poonah
of the 12th instant we learn that the cutting of the Grain has commenced there—that the Crops
have been very favourable” (BLNC MC 1112, 19 October 1793). This indicates elevated rainfall,
but not enough to damage crops (i.e. category +1).
4.5.5 Excess monsoon year: 1827
Overall, excess rainfall occurred in Mumbai and the Gulf of Khambat in 1827 and above average
rainfall in Pune, generating a WIMR category of +2. Only one report is available for Gujarat for
1827, from 13 October in the Bombay Courier. This states that “the monsoon has been one of the
heaviest, and at the same time one of the most agreeable, rainy seasons, that has occurred for
several years” (BLNC MC 1112, 13 October 1827). Rainfall for the season over the Gulf of
Khambat was therefore classified as excess. Reports for Mumbai indicate that rain in June fell with
“unusual severity” (BLNC MC 1112, 23 June 1827) (i.e. heavy rainfall), although later reports are
inconclusive with no indication of either below- or above-average rainfall. Reports for Pune also
mention heavy rainfall in early June. An observer in Pune recorded “Violent” rainfall in her diary of
10, 11 and 12 June 1827 (BL MSS Eur D888, 10/11/12 June 1827), and a report in the Bombay
Courier of 13 June reports “a considerable quantity of rain… in the Deccan” (BLNC MC 1112, 13
June 1827). Rainfall in July was, however, described as “only showers” (BL MSS Eur D888, 7
August 1827), and therefore classified as scanty. August likewise recorded only “scanty showers”
(BLNC MC 1112, 5 September 1827), although with more regular reports of rainfall (BL MSS Eur
D888) and was classified as deficient. Daily rain was recorded in early September (BL MSS Eur
D888, 16 September 1827), with “violent” rain on 26 September (BL MSS Eur D888, 16 September
1827) and “tremendous rain” on 7 October (BL MSS Eur D888, 16 September 1827). A
classification of heavy was therefore assigned for rainfall during September/October.
5 Results
5.1 Monsoon rainfall reconstructions for Mumbai, Pune and the Gulf of Khambat
Reconstructed seasonal monsoon indices for the Mumbai region are presented in Figure 3a.
Reconstruction was possible from 1781 onwards, with monthly rainfall indices generated for all but
seven years (1790, 1793, 1796, 1806, 1809, 1829, 1831); reconstruction for these years was
based on seasonal information. No reconstruction could be undertaken for a further six years
(1791, 1792, 1797, 1805, 1807, 1811) due to an absence of documentary meteorological
evidence. Three periods of below-average rainfall are evident from the reconstruction: 1781-1787,
1799-1806 and 1833-1840. Above-average rainfall occurred from 1788-1798 with a sustained
12
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
period of above-average rainfall apparent from c.1813 to c.1830. A further period of above average
rainfall is evident from 1846 to 1854.
Figure 3a also displays instrumental rainfall data for Mumbai, which show a strong agreement
with the reconstructed monsoon rainfall indices (Kendall Tau correlation coefficient of 0.59,
significant at 99%). Degrading the instrumental data into the five rainfall categories allows for direct
comparison; this indicates a strong relationship ( coefficient of 0.64, significant at 99%). This is
comparable with coefficients calculated for other documentary series in Europe (cf. Brázdil et al.
2005). Only seven years between 1817 and 1846 exhibit differences of more than one class
between the reconstructed and degraded data – 1818, 1821, 1825, 1831, 1837, 1839, 1842. No
years indicate differences of more than two classes. This supports the adopted methodology (but
note the comments in section 4.2 regarding the homogeneity of the GHCN data).
For Pune, documentary evidence was available for reconstruction from 1792 onwards, with data
gaps from 1796-1802, 1806-1810 and 1820-1822. Reconstructions for 1792, 1793, 1805, 1811,
1813, 1823, 1831 and 1832 were reliant on seasonal descriptors. The reconstructed monsoon
index for the Pune region is presented in Figure 3b. Periods of predominantly below-normal rainfall
are evident from c.1824-1833 and c.1845-1858. A sustained period of above-normal rainfall is
evident from 1834-1844, with only one year of below-normal rainfall during this period (1838). It is
difficult to infer trends before 1823 due to gaps within the data.
The reconstructed monsoon rainfall indices for the Gulf of Khambat are presented in Figure 3c.
It was not possible to reconstruct rainfall for a number of years (1782-1785, 1789, 1791-1793,
1795-1801, 1806, 1808-1811, 1814-1815, 1817-1818, 1820, 1823 and 1826) due to a lack of
documentary evidence. Seasonal reconstructions only were undertaken for 13 years (1781, 1786,
1790, 1794, 1803, 1805, 1813, 1816, 1819, 1822, 1827, 1830, 1831 and 1832). Sparse data
before 1827 makes it difficult to infer trends. However, after this date, three periods of above-
normal (1834-1837, 1842-1844 and 1854-1859) and three of below-normal (1830-1833, 1838-1841
and 1846-1850) rainfall are evident.
The rainfall reconstructions for Mumbai, Pune and the Gulf of Khambat generally show strong
agreement, with the majority of years exhibiting differences of two classifications or less between
the three regions. Between Mumbai and Pune, 40 out of 51 years (88%) show differences of two
classifications or less. Between Mumbai and the Gulf of Khambat the figure is 45 of 50 (90%) and
between Pune and the Gulf of Khambat 39 of 44 (89%). This is comparable to figures within the
instrumental record: from 1871 (the first year for which instrumental rainfall data for the Gulf of
Khambat are available) to 1950, degrading instrumental data into the five seasonal rainfall classes
(Table 6) shows differences of greater than two categories on 18 occasions between Mumbai and
Pune (23%), 15 occasions between Mumbai and the Gulf of Khambat (19%) and 13 occasions
between Pune and the Gulf of Khambat (16%). Deviations of four categories (i.e. -2 to +2) occur on
six occasions during the reconstruction period: 1804 (between Mumbai and Gulf of Khambat),
13
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
1826 (Mumbai and Pune), 1832 (Mumbai and Pune/Gulf of Khambat), 1835 (Mumbai and Gulf of
Khambat), 1840 (Pune and Gulf of Khambat) and 1851 (Mumbai and Pune). This phenomenon
also occurs on 16 occasions between 1871 and 1950.
Differences between Mumbai and Pune are likely to be related to the rain shadow effects of the
Western Ghats (Gunnel 1997), and it is significant to note that on all four occasions where four
classes separated rainfall in these regions the location experiencing the heavier rainfall was
Mumbai. Differences between the Gulf of Khambat and Mumbai/Pune are likely to be related to the
influence of northwesterly-moving depressions from the Bay of Bengal, which affect rainfall over
Gujarat but have little impact over western peninsular India (Gadgil 2003).
5.2 Western Indian Monsoon Rainfall reconstruction
The WIMR reconstruction, which combines the seasonal monsoon rainfall indices for Mumbai,
Pune and the Gulf of Khambat, is displayed in Figure 3d. The WIMR uses data from all three
reconstruction areas during 45 years of the study period (including every year from 1827-1860),
from two areas during 16 years, and from one area only in a further 18 years. No data were
available for 1791 and 1797. The record exhibits four periods of generally deficient monsoon
rainfall: 1780-85, 1799-1806, 1830-1838 and 1845-1857. Above-normal rainfall is evident for 1788-
1794, 1813-1828 and 1839-1844. Using power spectrum analysis, a periodicity is observed of
approximately 6-7 years, although this is not statistically significant.
In general, the number of quotes available for reconstruction increases throughout the
reconstruction period. Records are particularly sparse prior for the period 1781-1799 (with the
exception of 1787-1788 and 1794) and 1804-1811 (Figure 2). Regional reconstructions during
these periods were undertaken using mainly seasonal rather than monthly descriptors, and are
reliant occasionally on only a handful of quotes. Likewise, years during these periods often do not
contain data for all three regions, with some years displaying no data. The WIMR therefore will
necessarily be of lower reliability during these periods, although this is difficult to quantify. Years
where reconstruction was reliant on seasonal descriptors only are indicated on Figure 3.
During the compilation of the rainfall indices for Mumbai, Pune and the Gulf of Khambat, one
period was identified when primary and secondary sources present potentially conflicting climatic
signals. This coincides with the famine of 1790-1792 documented within the secondary source, the
Report on Past Famines in the Bombay Presidency (see Table 2). In this report, Etheridge (1868)
suggests that famine occurred in Pune, Satara and various parts of Gujarat during 1790-1791, and
in Pune and Satara in 1792-1793. Famine is also reported in Kheda and Pune in 1791-1792. One
contemporary record, taken from the reports of the Public Department of the colonial government,
confirms that the famine in Bombay in 1790 followed a severe drought. The record notes that
Bombay received “the smallest quantity of rain ever remembered” in 1790 leading to an “almost
total failure of the Crops of Grain” (BL IOR/F/4/428/10490, 24 December 1790). This year was
14
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
therefore given an index value of -2. The famine reported in Pune and Satara during 1792
(Etheridge 1868) may not be entirely attributable to drought, as the only available contemporary
report for that year (BLNC MC 1122, 15 August 1792) states that elevated rainfall fell in August. As
a result, an index value of -1 was assigned. The information for 1791 is more ambiguous. There is
one account of drought conditions (Etheridge 1868: 95), but a documentary report for Vadodara
describes “a luxuriant crop of Cotton and Grain” (NLS MS.13695, 11 January 1804). Etheridge
(1868) notes that the famine of 1791 may have been due to a “bringing together of troops” (p. 95).
It is also possible that the report of drought conditions in 1791 was the result of a dating error.
Early (i.e. prior to English occupation) reports of famines were generally reliant on oral tradition,
creating potential issues with dating, particularly as some dates are stated using the Samvat
calendar which is itself geographically variable. To avoid potential error, no index value has been
assigned for 1791.
6 Comparison of monsoon rainfall reconstructions with other proxies
The following section compares the WIMR chronology presented in section 5 with other proxy-
based climatic reconstructions pertaining to western India. As noted in the introduction,
dendroclimatic reconstructions are sparse for peninsular India. However, Borgaonkar et al. (2010)
have derived a 523-year tree-ring record from teak using 64 cores from three sites in Kerala,
c.1000km from Mumbai. This record was used to generate a ring-width index from 1481 onwards,
which is statistically robust from 1748. The chief determinant of tree growth was found to be
moisture availability during the growing season, represented by the June-September (JJAS)
Palmer Drought Severity Index (PDSI) (Dai et al. 2004). Significant correlation was also found
between the reconstructed ring width chronology and All-India Monsoon Rainfall (AIMR)
(correlation coefficient = 0.32, rising to 0.52 on decadal timescales; Borgaonkar et al. 2010).
Borgaonkar et al. (2010) identify that whilst there is a statistically significant (>95%) relationship
between low growth years and low moisture availability, this relationship breaks down in years with
above normal rainfall. This renders the ring-width index a poor proxy for overall seasonal rainfall,
but a useful indicator of past drought years (Borgaonkar et al. 2010). A simple comparison can
therefore be made between teak low-growth years and WIMR (Table 9). Of the 16 low-growth
years presented by Borgaonkar et al. (2010), eight also exhibit reduced rainfall in the WIMR
chronology (including 1832 which exhibits deficient rainfall in Pune and Gujarat, but excess rainfall
in Mumbai). No reconstructed rainfall data are available for 1791 (section 5.1); however, as ring
width is dependent upon moisture in the previous and concurrent monsoon (Borgaonkar et al.
2010), this low-growth year may be related to the deficient monsoon of 1790. The deficient rainfall
years of 1801 and 1823 may have also contributed to low growth years in 1802 and 1823. Seven
years exhibiting low growth, however, display average or above-average rainfall in the WIMR, and
an additional nine years with deficient rainfall do not exhibit low teak growth. This suggests either
15
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
uncertainty in the teak reconstruction or WIMR, or localised drought in either southern or western
India.
Agreement of the WIMR with the gridded Monsoon Asia Drought Atlas (MADA) of PDSI for the
East and South Asian monsoon region (Cook et al. 2010) is weaker. Ten years within the WIMR
display deficient rainfall within two or more regions: 1790, 1803, 1812, 1823, 1824, 1832, 1838,
1845, 1848 and 1855. Of these, only 1845 and 1855 are associated with drought within the MADA,
and then only for Gujarat but not the Deccan. The monsoon season of 1824 was deficient in
Gujarat but the MADA identifies positive PDSI values over western peninsular India. The years
1790 and 1832 display highly positive PDSI values across western India, indicating elevated
rainfall. Neutral PDSI values are displayed for 1803, 1812, 1823, 1838 and 1848.
These discrepancies between the WIMR and MADA classifications may be related to problems
within the methodology adopted in the documentary reconstruction. However, several previous
studies have highlighted the particular skill of documentary records in capturing extreme
meteorological conditions (cf. Bradley 1999; Brázdil et al. 2005; Jones et al. 2009), and it is likely
that reports of drought are reliable. For example, a report from the town of Porbandar on the
Kathiawar peninsular in September 1812, describes: “no rain has fallen here and of course now the
Season is too far advanced for us to expect any, the face of the Country is miserable, resembling a
sandy beach, more than anything else” (BL MSS Eur D666, 30 September 1812). A report to the
Literary Society of Bombay in April 1815 by the British Resident at Baroda (Vadodara) describes in
great detail a severe famine in Gujarat in 1812 and 1813, caused by “a failure of the rain” (Carnac,
1819). Six other quotations in 1812 indicate deficient rainfall in Gujarat, and famine in this region is
mentioned within the Report of Past Famines in the Bombay Presidency (Etheridge, 1868). It is
therefore highly unlikely that the PDSI of ~ +2 assigned to southern Gujarat within the MADA for
this year is correct. The disagreement may instead be reflective of the spatial coverage of tree-ring
records within the MADA, which is derived from a network of 327 tree-ring records of which only
four are from peninsular India (Cook et al. 2010). It is hoped that, as new tree-ring chronologies are
developed for the Indian subcontinent, more rigorous verification of the WIMR series against
dendroclimatic datasets will be possible.
Sontakke et al. (1993, 2008) and Sontakke and Singh (1996) used rain gauge data to produce
an extended JJAS rainfall record for India and for eight homogenous climate regions within India.
The study area crosses two of these homogenous regions, with Mumbai and Pune located within
the West Peninsular India (WPI) region and the Gulf of Khambat within the Northwest India (NWI)
region. No direct comparison can therefore be made between these studies and WIMR.
Correlations can, however, be drawn between the results presented in Sontakke et al. (2008) and
a composite of reconstructed rainfall at Mumbai and Pune (Figure 4). To enable this comparison,
an average of the instrumental rainfall data for Mumbai and Pune post-1860 was degraded into the
5-point classification system using the percentage boundaries presented in Table 6.
16
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
For the instrumental period (1861-2000), the Kendall Tau coefficient between rainfall recorded
at the Colaba Observatory, Mumbai, and WPI rainfall (degraded into the 5-point classification
system) is 0.33. The correlation between Pune rainfall and WPI for the same period is 0.44, and
the correlation between WPI rainfall and the average of rainfall at Mumbai and Pune is 0.39. For
1817-1860 – the portion of the reconstruction period for which comparison with Sontakke et al.
(2008) is possible – these correlations decrease to = 0.19, 0.20 and 0.20 respectively, which are
not significant at the 95% level.
This apparent breakdown in correlation may represent an error in the methodology adopted for
this study. However, the poor correlations established for the reconstruction period could instead
arise due to uncertainties within early data used in Sontakke et al. (2008). During the early years of
the WPI reconstruction, Sontakke et al. (2008) rely on 1817-1846 published instrumental data for
Mumbai, available through the GHCN. As mentioned in section 4.2, there are uncertainties in these
data (which may be inhomogenous). We do not challenge the statistical methodology adopted by
Sontakke et al. (2008). However, due to the unreliable rainfall data from 1817-1846 we suggest
that the regional and national monsoon rainfall reconstructions presented by Sontakke et al. (2008)
for years prior to 1847 should be treated cautiously when analysing long-term trends.
7 Long-term relationship between ENSO and Western Indian Monsoon Rainfall
The extended rainfall record provided by the WIMR permits an examination of the long-term
dynamics of monsoon variability. The lack of proxy-derived reconstructions of the IOD prohibits
analysis of the relationship between this climate mode and monsoon rainfall on longer timescales.
Annual ENSO indices based on SLP and SST have been constructed for as far back as 1500
(Cole et al. 2000; Braganza et al. 2009; Li et al. 2011). However, these indices generally fail to
replicate correlations between AIMR and ENSO observed in modern instrumental series such as
Niño 3.4 (Parthasarathy et al. 1993). A number of documentary-derived ENSO chronologies are
also available, the most commonly cited being that produced by Quinn and Neal (1992) and
updated by Ortlieb (2000).
Gergis and Fowler (2009) developed a multi-proxy reconstruction of ENSO events dating back
to 1525. This utilised instrumental data, tree-ring reconstructions, coral records, ice-cores and
documentary records from several locations on the rim of the Pacific and Indian Oceans to
generate a quantified magnitude score for each El Niño and La Niña event, which was transposed
into a 5-point strength index using percentiles. The discrete ranked nature of the Gergis and
Fowler (2009) reconstruction, hereafter referred to as GFENSO, provides a suitable comparison
with the WIMR chronology. However, as Gergis and Fowler (2009) include a drought chronology
for India (Whetton and Rutherfurd 1994) as an El Niño proxy in their reconstruction, a slight
alteration to the GFENSO indices was necessary to avoid circularity. For years classified as El
Niño in GFENSO and by Ortlieb (2000), and as Indian drought years by Whetton and Rutherfurd
(1994), the GFENSO chronology was left unaltered. For years categorised as El Niño in GFENSO 17
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
but not in Ortlieb (2000), the proxy relating to the Indian drought was removed from GFENSO and
the magnitude classification for that year adjusted using the methodology presented by the
authors. This discounted El Niño events identified by Gergis and Fowler (2009) in 1782, 1838 and
1860.
The modified GFENSO indices were combined into a single chronology, with extreme El Niño
years given a ranking of -5 and extreme La Niña years +5. Years exhibiting neither El Niño nor La
Niña conditions were ranked as zero. The validity of using this index was determined by correlating
the chronology with annual mean Niño-3.4 SST data for 1871-2002. This generates a Kendall Tau
coefficient of 0.46, significant at 99%. Correlation between AIMR and GFENSO produces a
coefficient of 0.26 (significant at 99%). This is lower than the coefficient of 0.40 calculated
between AIMR and annual mean Niño-3.4 SST. However, sliding correlation analysis between
GFENSO and degraded rainfall data during the instrumental period clearly isolates two periods of
strong coupling from c.1885-c.1910 and c.1965-c.1980 (Figure 5), which have been identified in
previous analyses of the long term relationship between Niño 3.4 SSTs and Indian monsoon
rainfall (Torrence and Webster 1998; Maraun and Kurths 2005). This suggests that GFENSO is
suitable for examining long-term changes in correlation between ENSO and monsoon rainfall, if not
the exact magnitude of the correlation.
The long-term relationship of ENSO with localised rainfall in western India was determined by
correlating GFENSO with the seasonal monsoon indices derived in this study for the three
separate reconstruction areas and for WIMR. As shown in Table 10, all areas produced generally
stationary correlations between the reconstruction and instrumental period. The correlation
between ENSO and WIMR also appears to retain stationarity. To examine this further, 21- and 31-
year sliding Pearson correlations between the modified GFENSO and WIMR were generated
(Figure 5), with monsoon rainfall indices prior to 1860 derived from the documentary reconstruction
and from 1860 using instrumental data.
As noted above, Figure 5 replicates the high correlations between levels of monsoon rainfall
and ENSO observed in previous studies (Torrence and Webster 1998; Maraun and Kurths 2005),
with peaks in correlation at c.1905 and c.1976. A further peak in correlation is noted between
c.1835 and c.1845. This is a potentially significant finding, particularly given current debates over
whether the apparent breakdown in the ENSO-monsoon relationship since 1976 is a product of
anthropogenic climate change or represents the downward arm of a longer-term fluctuation (cf.
Krishna Kumar et al. 1999; Gershunov et al. 2001; Kinter et al. 2002; Maraun and Kurths 2005;
Goswami and Xavier 2005; Xavier et al. 2007; Robinson et al. 2008). If the apparent periodicity of
approximately 75 years were a persistent feature, our results would support the latter
interpretation. Modelled variability in the ENSO-monsoon relationship over 1,000 years exhibits a
similar fluctuation (Kitoh 2007), suggesting an internal oscillation of the system.
18
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
Maraun and Kurths (2005) proposed that the periods of coherence between ENSO and the
Indian monsoon centred on c.1905 and c.1976 may be related to volcanic activity. Both periods
commence with major volcanic events in Indonesia (Krakatau in 1883 and Mount Agung in 1963),
and occur during periods of high volcanic activity, associated with high mean stratospheric optical
aerosol depths (Bertrand et al. 1999). Maraun and Kurths (2005) suggested that volcanic activity
may increase the phase coupling between ENSO and the monsoon by increasing the variance of
ENSO and/or through global cooling due to the release of sulphate aerosols, rendering the
monsoon more sensitive to ENSO fluctuations.
The third period of strong ENSO-monsoon relationship indicated through our analysis (1815-
1856; Figure 5) also begins with a major Indonesian eruption (Tambora: 1815), and spans a period
of enhanced stratospheric aerosol depth associated with higher volcanic activity. In contrast, the
identified periods of poor ENSO-monsoon coupling (1866-1886 and 1936-1956) are both
associated with lower volcanic activity (Bertrand et al. 1999). Analysis of ENSO variance through
time exhibits similar peaks at c.1970, c.1900 and c.1820 (Fowler et al. 2012). Our results therefore
add further weight to the argument presented by Maraun and Kurths (2005). Whether the apparent
fluctuation in the ENSO-monsoon relationship is an internal oscillation or related to volcanic-
induced cooling has significant implication for forecasting, and we support the call by Maraun and
Kurths (2005) for model simulations of this effect.
8 Conclusions
This study has used English-language documentary records to derive a semi-quantitative
reconstruction of monsoon rainfall variability over western India for the period 1781-1860. This
represents an extension of 66 years on the existing homogenous instrumental record for the
region, which begins at the Colaba Observatory, Bombay, in 1847. Monsoon rainfall
reconstructions have been presented for Mumbai, Pune and the area of southern Gujarat
surrounding the Gulf of Khambat, and combined to produce a composite WIMR chronology for
western India as a whole. The WIMR exhibits four periods of generally below-normal (1780-85,
1799-1806, 1830-1838 and 1845-1857) and three of above-normal (1788-1794, 1813-1828 and
1839-1844) monsoon rainfall. Ten widespread droughts are evident, in 1790, 1803, 1812, 1824,
1833, 1838, 1845, 1847-48, 1850 and 1855.
The WIMR exhibits agreement with a tree-ring derived drought chronology for Kerala.
Agreement with the Monsoon Asia Drought Atlas is weaker, but this may be related to the paucity
of tree-ring data used to reconstruct conditions over India within the MADA. Our results suggest
that regional rainfall series derived by statistical inference from early instrumental rainfall data may
exhibit a degree of unreliability prior to 1847. This is due to the use of instrumental data for Mumbai
from 1817-1846 (currently available through the GHCN), which, our archival explorations reveal,
may be inhomogenous.
19
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
Comparison between our rainfall reconstruction and ENSO indices suggests periods of
generally higher and lower correlation between ENSO and monsoon rainfall over western India,
with a periodicity of approximately 75 years. This may be related to either volcanic activity or an
internal oscillation in the ENSO-monsoon system. Future reconstructions of pre-instrumental
monsoon variability in India should be analysed to determine whether such periodicity is also
observed, as this may have implications for current debates regarding recent reductions in the
correlation between ENSO and monsoon rainfall.
Taken as a whole, this study demonstrates the considerable potential for historical
climatological studies of moisture variability in the tropics and subtropics, particularly for regions
with strongly seasonal rainfall regimes and rich, well-preserved archives of historical documents. It
is hoped that the results will enable further analysis of the long-term variability of monsoon rainfall
and associated climatic forcings for the Indian subcontinent.
Acknowledgements
GCDA was in receipt of a University of Brighton doctoral research scholarship whilst the archive-
based research for this paper was undertaken. The Dudley Stamp Memorial Fund (administered by
the Royal Geographical Society), the Royal Historical Society and the Indian National Trust for Arts
and Cultural Heritage (INTACH) UK Trust Grants provided funding to GCDA to support visits to the
Houghton Library, Harvard, USA and the Archives of the Government of Maharashtra, Mumbai,
India. Our thanks go to the two anonymous reviewers whose comments greatly improved this
paper.
20
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
References
Adamson GCD (2012) ‘The languor of the hot weather’: Everyday perspectives on weather and climate in western India, 1819-1828. Journal of Historical Geography 38:143-154
Adamson GCD, Nash DJ (2012) Long-term variability in the date of monsoon onset over western India. Climate Dynamics doi: 10.1007/s00382-012-1494-x
Annamalai H, Slingo JM (2001) Active/break cycles: diagnosis of the intraseasonal variability of the Asian Summer Monsoon. Climate Dynamics 18:85-102
Ashok K, Saji NH (2007) Impacts of ENSO and Indian Ocean Dipole events on the sub-regional Indian summer monsoon rainfall. Natural Hazards 42:273-285
Ashok K, Guan Z, Yamagata T (2001) Impact of the Indian Ocean dipole on the relationship between the Indian monsoon rainfall and ENSO. Geophysical Research Letters 28:4499–4502
Banks J (1790) Diary of the Rain at Bombay from 1780 to 1787, and of the Heat for the Year 1788. Philosophical Transactions of the Royal Society 80:590
Bertrand C, van Ypersele J-P, Berger A (1999) Volcanic and solar impacts on climate since 1700. Climate Dynamics 15:355-367
Bhattacharyya A, Yadav RR (1999) Climatic reconstructions using tree-ring data from tropical and temperate regions of India – a review. IAWA Journal 20(3):311-316
Bhowmik SKR, Roy SS, Kundu PK (2008) Analysis of large-scale conditions associated with convection over the Indian monsoon region. International Journal of Climatology 28:797-821
Borgaonkar HP, Sikder AB, Ram S, Pant GB (2010) El Niño and related monsoon drought signals in 523-year-long ring width records of teak (Tectona grandis L.F.) trees from south India. Palaeogeography, Palaeoclimatology, Palaeoecology 285:74-84
Bradley RS (1999) Paleoclimatology: Reconstructing Climates of the Quaternary – Second Edition. Harcourt Academic Press, Boston.
Braganza K, Gergis JL, Power SB, Risbey JS, Fowler AM (2009) A multiproxy index of the El Nino-Southern Oscillation, A.D. 1525-1982. Journal of Geophysical Research (Atmospheres) 114 D05106 doi:10.1029/2008JD010896
Brázdil R, Pfister C, Wanner H, von Storch H, Luterbacher J (2005) Historical climatology in Europe – the state of the art. Climatic Change 70 363-430
Carnac JR (1819) Some Account of the Famine in Guzerat in the years 1812 and 1813, in a Letter to William Erskine, Esq. Transactions of the Literary Society of Bombay 1:296-303
Cole JE, Dunbar RB, McClanahan TR, Muthiga NA (2000) Tropical Pacific forcing of decadal SST variability in the Western Indian Ocean over the past two centuries. Science 287:617–619
Cook E, Anchukaitis KJ, Buckley BM, D’Arrigo R, Jacoby GC, Wright WE (2010) Asian Monsoon failure and megadrought during the last millenium. Science 328:486-489
Dai A, Trenberth KE, Qian T (2004) A global data set of Palmer Drought Severity Index for 1870-2002; relationship with soil moisture and effects of surface warming. Journal of Hydrometeorology 5:117-1130
Endfield GH, Nash DJ (2002) Missionaries and morals: climatic discourse in nineteenth century central southern Africa. Annals of the Association of American Geographers 92:727-742
Etheridge AT (1868) Report on Past Famines in the Bombay Presidency. Government of the Bombay Presidency, Bombay.
Fasullo J, Webster PJ (2002) Hydrological Signatures Relating the Asian Summer Monsoon and ENSO. Journal of Climate 15:3082-3095
21
719
720721
722723
724725
726727
728729
730731
732733
734735
736737
738739740
741742
743744745
746747
748749
750751
752753
754755756
757758
759760
761762
Fowler AM, Boswijk G, Lorrey AM, Gergis J, Pirie M, McCloskey SPJ, Palmer JG and Wunder J (2012) Multi-centennial tree-ring record of ENSO-related activity in New Zealand. Nature Climate Change 2 172-176
Gadgil S (2003) The Indian monsoon and its variability. Annual Review of Earth and Planetary Science 31:429-467
Gadgil S, Joshi NV (1983) Climatic clusters of the Indian region. Journal of Climatology 3:47-63
Gadgil S, Srinivasan J (1990) Low Frequency Variation of Tropical Convergence Zones. Meteorology and Atmospheric Physics 44:119-132
Gergis JL, Fowler AM (2009) A history of ENSO events since A.D. 1525: implications for future climate change. Climatic Change 92:343-387
Gershunov A, Schneider N, Barnett T (2001) Low-frequency modulation of the ENSO-Indian monsoon rainfall relationship: signal or noise. Journal of Climate 14:2486-2492
GHCN (2010) Global Historical Climatology Network. http://www.ncdc.noaa.gov/ghcnm/. Accessed January 2010
Goswami BN, Ajayamohan RS, Xavier PK, Sengupta D (2003) Clustering of low pressure systems during the Indian summer monsoon by intraseasonal oscillations. Geophysical Research Letters 304:1431 doi:10.1029/2002GL016734
Goswami BN, Xavier PK (2005) ENSO control on the south Asian monsoon through the length of the rainy season. Geophysical Research Letters 32 L18717 doi:L10.1029/2005GL023216
Grossman RL, Durran DR (1984) Interaction of low-level flow with the western Ghat Mountains and offshore convection in the summer monsoon. Monthly Weather Review 112:652-672
Grove RH (1997) The East India Company, the Australians and the El Niño: colonial scientists and ideas about global climatic change and teleconnections between 1779 and 1930. In: Ecology, Climate and Empire. White Horse Press, Cambridge, pp.124-146
Guhathakurta P, Rajeevan M (2008) Trends in the rainfall pattern over India. International Journal of Climatology 28:1453-1469
Gunnell Y (1997) Relief and climate in South Asia: The influence of the Western Ghats on the current climate pattern of peninsular India. International Journal of Climatology 17:1169-1182
Harrison M (1999) Climates and Constitutions: Health, Race, Environment and British Imperialism in India 1600-1850. Oxford University Press, Oxford.
Indian Meteorological Department (1943) Climatological Atlas for Airmen 3. IMD, New Delhi.
Jones PD, Briffa KR, Osborn TJ, Lough JM, van Ommen TD, Vinther BM, Luterbacher J, Wahl ER, Zwiers FW, Mann ME, Schmidt GA, Ammann CM, Buckley BM, Cobb KM, Esper J, Goosse H, Graham N, Jansen E, Kiefer T, Kull C, Küttel M, Mosley-Thompson E, Overpeck JT, Riedwyl N, Schulz M, Tudhope AW, Villalba R, Wanner H, Wolff E, Xoplaki E (2009) High-resolution palaeoclimatology of the last millennium: a review of current status and future prospects. The Holocene 19:3-49
Kinter JL, Miyakoda K, Yang S (2002) Recent changes in the connection from the Asian monsoon to ENSO. Journal of Climate 15:1203-1214
Kitoh A (2007) Variability of the Indian monsoon-ENSO relationship in a 1000-year MRI-CGCM2.2 simulation. Natural Hazards 42:261-272
Krishna Kumar K, Rajagopalan B, Cane, MA (1999) On the weakening relationship between the Indian monsoon and ENSO. Science 284:2156-2159
Krishna Kumar K, Rajagopalan B, Hoerling M, Bates G, Cane M (2006) Unraveling the mystery of Indian monsoon failure during El Niño. Science 314:115-119
Krishnamurthy V, Goswami BN (2000) Indian Monsoon-ENSO Relationship on Interdecadal Timescale. Journal of Climate 13:579-595
22
763764765
766767
768
769770
771772
773774
775776
777778779
780781
782783
784785786
787788
789790
791792
793
794795796797798799
800801
802803
804805
806807
808809
Krishnamurthy V, Shukla J (2000) Intraseasonal and Interannual Variability of Rainfall over India. Journal of Climate 13:4366-4377
Krishnamurthy V, Shukla J (2007) Intraseasonal and Seasonally Persisting Patterns of Indian Monsoon Rainfall. Journal of Climate 20:3-20
Krishnamurti TN, Cocke S, Pasch R, Low-Nam S (1983) Precipitation estimates from raingauge and satellite observations, summer MONEX. FSU Report 83-7, Department of Meteorology, Florida State University, Tallahassee.
Krishnan R, Zhang C, Sugi M. (2000) Dynamics of breaks in the Indian summer monsoon. Journal of the Atmospheric Sciences 57:1354-1372
Lawrence DM, Webster PJ (2001) Interannual variations of the intraseasonal oscillation in the South Asian summer monsoon region. Journal of Climate 14:2910-2922
Li J, Xie S-P, Cook ER, Huang G, D'Arrigo R, Liu F, Ma J, Zheng X-T (2011) Interdecadal modulation of El Niño amplitude during the past millennium. Nature Climate Change 1:114-118
Lim Y-K, Kim K-Y (2007) ENSO Impact on the space-time evolution of the regional Asian summer monsoons. Journal of Climate 20:2397-2415
Maraun D, Kurths J (2005) Epochs of phase coherence between El Niño/Southern Oscillation and Indian Monsoon. Geophysical Research Letters 32:L15709
Mohanty UC, Raju PVS, Bhatla R (2005) A study on climatological features of the Asian summer monsoon: dynamic, energetics and variability. Pure and Applied Geophysics 16:1511-1541
Mooley DA, Shukla J (1987) Variability and forecasting of the summer monsoon rainfall over India. In: Chang C-P, Krishnamurti TN (eds.) Monsoon Meteorology. Clarendon Press, Oxford, pp.26-59
Nash DJ, Endfield GH (2002) A 19th century climate chronology for the Kalahari region of central southern Africa derived from missionary correspondence. International Journal of Climatology 22:821-841
Nash DJ, Endfield GH (2008) ‘Splendid rains have fallen’: links between El Niño and rainfall variability in the Kalihari, 1840-1900. Climatic Change 86:257-290
Nash DJ, Grab SW (2010) “A sky of brass and burning winds”: documentary evidence of rainfall variability in the Kingdom of Lesotho, Southern Africa, 1824-1900. Climatic Change 101:617-653
Nicholls J (1819) Remarks on the temperature of the Island of Bombay during the years 1803 and 1804. Transactions of the Literary Society of Bombay 1:6-11
Nicholson SE, Klotter D, Dezfuli, AK (2012) Spatial reconstruction of semi-quantitative precipitation fields over Africa during the nineteenth century from documentary evidence and gauge data. Quaternary Research 78:13-23
Ortlieb L (2000) The documented historical record of El Niño events in Peru: An update of the Quinn and Neal record (Sixteenth through Nineteenth Centuries. In: Diaz HE, Markgraf V (eds.) El Niño and the Southern Oscillation: multiscale variability and global and regional impacts. Cambridge University Press, Cambridge, pp.207-296
Pant GB, Rupa Kumar K, Sontakke NA, Borgaonkar HP (1993) Climate variability over India on Century and Longer Time Scales. In: Keshavamurty RN, Joshi PC (eds.) Advances in Tropical Meteorology. Tata McGraw-Hill Publishing Company, New Delhi, pp.71-84
Parthasarathy B, Sontakke NA, Monot AA, Kothwale DR (1987) Droughts/floods in the summer monsoon season over different meteorological subdivisions of India for the period 1871-1984. Journal of Climatology 7:57-70
Parthasarathy B, Rupa Kumar K, Munit AA (1993) Homogenous Indian monsoon rainfall: variability and prediction. Proceedings of the Indian Academy of Sciences (Earth and Planetary Science) 102:121-155
23
810811
812813
814815816
817818
819820
821822
823824
825826
827828
829830831
832833834
835836
837838839
840841
842843844
845846847848
849850851
852853854
855856857
Parthasarathy B, Munot AA, Kothawale DR (1994) All India monthly and seasonal rainfall series 1871-1993. Theoretical and Applied Climatology 45:217-224
Quinn WH, Neal VT (1992) The historical records of El Niño events. In: Bradley RS, Jones PD (eds.) Climate Since A.D. 1500. Routledge, London, pp.623-648
Ram S (2011) On the recent strengthening of the relationship between Palmer Drought Severity Index and teak (Tectona grandis L.) tree-ring width chronology from Maharashtra, India: A case study. Quaternary International 248:92-97
Robertson A, Overpeck J, Rind D, Mosely-Thompson E, Zielinski G, Lean J, Koch D, Penner J, Tegen I, Healy R (2001) Hypothesized climate forcing time series for the last 500 years. Journal of Geophysical Research 106:14783-14803
Robinson LF, de la Peña VH, Kushnir Y (2008) Detecting shifts in correlation and variability with application to ENSO-monsoon rainfall relationships. Theoretical and Applied Climatology 94:215-224
Saji NH, Goswami BN, Vinayachandran PH, Yamagata T (1999) A dipole mode in the tropical Indian Ocean. Nature 401:360–363
Sontakke NA, Pant GB, Singh N (1993) Construction of All-India Summer Monsoon rainfall series for the period 1844-1991. Journal of Climate 6:1807-1811
Sontakke NA, Singh N (1996) Longest instrumental regional and all-India summer monsoon rainfall series using optimum observations: reconstruction and update. The Holocene 6:315-331
Sontakke NA, Singh N, Singh HN (2008) Instrumental period rainfall series of the Indian region (AD 1813-2005): revised reconstruction, update and analysis. The Holocene 18:1055-1066
Srinivasan J, Gadgil S, Webster PJ (1993) Meridional propagation of large-scale monsoon convective zones. Meteorology and Atmospheric Physics 52:15-35
Sykes WH (1835) On the atmospheric tides and meteorology of Dukhun (Deccan), East Indies. Philosophical Transactions of the Royal Society 125:161-220
Torrence C, Webster PJ (1998) The annual cycle of persistence in the El Niño/South Oscillation. Quarterly Journal of the Royal Meteorological Society 124:1985-2004
Walsh RPD, Glaser R, Militzer, S. (1999) The climate of Madras during the eighteenth century. International Journal of Climatology 19:1025-104
Whetton P, Rutherfurd I (1994) Historical ENSO teleconnections in the eastern Hemisphere. Climatic Change 28:221–253
Xavier PK, Marzin C, Goswami BN (2007) An objective definition of the Indian summer monsoon season and a new perspective on the ENSO–monsoon relationship. Quarterly Journal of the Royal Meteorological Society 133:749-764
Yoon J-H, Chen T-C (2005) Water vapour budget of the Indian monsoon depression. Tellus 57A:770-782
24
858859
860861
862863864
865866867
868869870
871872
873874
875876
877878
879880
881882
883884
885886
887888
889890891
892893
List of Figures
Figure 1: Map of India displaying the study area, principal locations noted in the text, plus
average onset dates for the southwest monsoon (1900-1941) across the subcontinent
(after IMD 1943). Mean monthly rainfall distribution patterns are also shown for
Mumbai, Pune, and the Gulf of Khambat (the latter calculated using data from Surat,
Vadodara, Bharuch, Kheda, Ahmedabad, Bhavnagar and Rajkot) (GHCN 2010).
Figure 2: Number of datapoints (i.e. quotations) used during the reconstruction of monsoon
rainfall variability in: (a) Mumbai; (b) Pune; (c) Gulf of Khambat. (d) Number of sources
available across all three reconstruction areas.
Figure 3: Reconstructed seasonal monsoon indices for the three regional reconstruction areas:
(a) Mumbai, including instrumental rainfall data taken from GHCN (2010); (b) Pune; (c)
Gulf of Khambat. Dark grey (pale grey) bars indicate years for which monthly (seasonal
only) reconstruction was possible. (d) The Western India Monsoon Rainfall
reconstruction. Indices for 1781-1786 for Mumbai are derived from instrumental data
published in the Philosophical Transactions of the Royal Society (RSCHS MA.213).
Figure 4: Composite monsoon rainfall reconstruction for Mumbai and Pune.
Figure 5: 21- and 31-year sliding correlations for the relationship between ENSO intensity (using
a modified version of the Gergis and Fowler [2009] historical El Niño and La Niña
series) and the Western India Monsoon Rainfall reconstruction.
25
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
List of Tables
Table 1: Details and codes for archival collections consulted in the UK, USA and India.
Table 2: Key sources used for documentary reconstruction of monsoon intensity.
Table 3: Summary of descriptive rainfall conditions from documentary materials, together with
illustrative quotes for the monsoon season of 1853, Mumbai reconstruction area.
Table 4: Indian Meteorological Department terminology to describe localised rainfall at a weekly
or monthly timescale, used to generate monthly indices of monsoon strength. The final
category (in italics) is an addition for the purposes of this study.
Table 5: Calibration table of monthly indices of monsoon strength for Mumbai, describing
prevailing characteristics of each monsoon classification noted within documentary
archive material during the calibration period.
Table 6: Rainfall categories and percentage of long-period average as used in the IMD 5-
Parameter Statistical Ensemble Forecasting system.
Table 7: Calibration table of seasonal descriptors of monsoon strength for Mumbai, describing
prevailing characteristics of each monsoon classification noted within documentary
archive material during the calibration period.
Table 8: Regional monthly, seasonal and Western India Monsoon Rainfall (WIMR) rainfall
categories for 1838, 1818, 1853, 1793 and 1827. See section 4.5 for descriptions of
conditions. N.D. indicates no data.
Table 9: Kerala teak series low-growth years (Borgaonkar et al. 2010) with reconstructed
monsoon indices for the three reconstruction areas and Western India Monsoon
Rainfall (WIMR). Years in agreement are shaded. N.D. indicates no data.
Table 10: Correlations between the modified GFENSO index (Gergis and Fowler 2009; see text)
and reconstructed seasonal monsoon indices for the three study areas and the
combined Western India Monsoon Rainfall (WIMR) record. Correlations post-1860 are
calculated using degraded instrumental data.
26
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
Table 1: Details and codes for archival collections consulted in the UK, USA and India.
Name of collection Archive Location Archive codeIndia Office Records British Library, St Pancras,
London, UKBL followed by catalogue number
India Office Private Papers British Library, St Pancras, London, UK
BL followed by catalogue number
British Library Newspaper Collections
British Library, Colindale, London, UK
BLNC followed by catalogue number
Church Missionary Society University of Birmingham Library, UK
UBL followed by catalogue number
American Board of Commissioners for Foreign Missions
Houghton Library, Harvard, USA HLH followed by catalogue number
Scottish Missionary Societies National Library of Scotland, Edinburgh, UK
NLS followed by catalogue number
National Library of Scotland Archives
National Library of Scotland, Edinburgh, UK
NLS followed by catalogue number
Archives of the Government of Maharashtra
Department of Archives, Government of Maharashtra, Mumbai, India
DAGM followed by catalogue number
Royal Society Archives Library of the History of Science, Royal Society, London, UK
RSA followed by catalogue number
Aberdeen Medico-Chirurgical Society
University of Aberdeen Library, UK (available online)
UAL followed by catalogue number
27
939
Table 2: Key sources used for documentary reconstruction of monsoon intensity.
Name CodeDiary of the Rain at Bombay from 1780 to 1787 RSCHS MA.213 / Banks (1790)
8-year weather diary of rainfall, which forms the earliest available data for the study period. The diary comprises of an instrumental record of rainfall for the monsoon period, commencing at the onset (as prescribed by the author of the diary).
Tours for scientific and economical research, made in Guzerat-Kattiawar, and the Conkuns, in 1787-88 by Dr. Hove
BL IOR/V/22/212 No. 16
The book consists of a printed transcription of diaries kept by Dr. Anton Pantaleon Hove whilst on tour in northwest India. It records descriptive weather conditions almost daily, with sporadic temperature readings in Fahrenheit.
Case books of James McGrigor, 1799-1803 ABCMS 4/1/4/15-23
Contain weekly summaries of climatic conditions at Bombay, as well as information on sick and wounded soldiers. The case books are held at the library of the Aberdeen Medico-Chirurgical Society.
Weather diary of Jasper Nicholls, 1802-1805 BL MSS Eur F/172/2-9
Nicholls’ diaries contain several ad hoc weather observations and monthly ‘meteorological abstracts’, including temperature observations, general comments on the weather and days of ‘heavy’ and ‘light’ rains.
Diaries of Mountstuart Elphinstone and Lucretia West, 1811-1828
BL MSS Eur F88/278-374; BL MSS D888/1
Although not systematic weather diaries, the journals contain frequent references to climate, including occasional temperature readings in Fahrenheit. Together they comprise a near-daily record of meteorological conditions in Bombay during the 1820s.
Bombay Courier, Bombay Gazette, Bombay Times, Bombay Standard
BL MC 1112; BL MC 1122; BL SM 44; BL SM 72
Climatic information printed within the newspapers ranges from brief comments on the progress of the monsoon to detailed descriptions of weather conditions including temperature and pressure observations. Extreme weather events that had an impact upon human livelihoods, such as droughts, floods or storms were generally recorded, particularly within Bombay itself. Occasional weather diaries were also published, as well as sporadic rain gauge information.
State of the Weather and the Crops (Within newspapers)
From 1833, bi-weekly reports of the ‘State of the Weather and the Crops’ were generated for the monsoon season (June to October), drafted by the Government Collectors in each of the divisions of the Bombay Presidency. The reports contained descriptive comments on monsoon rainfall and the progress of the crops.
Reports of past famines in the Bombay Presidency Etheridge (1868)
This document published in 1868, and consists of collated reports of previous famines. Material has been used only where there is significant overlap between reports from different areas, and/or where contemporary reports are also available.
28
940
941
Table 3: Summary of descriptive rainfall conditions from documentary materials, together with illustrative quotes for the monsoon season of 1853, Mumbai reconstruction area.
Monsoon Month
Description of conditions Illustrative Quotes
May/June First report of rain around 30 May. Occasional storms until two days of heavy rain on 13 and 14 June.Sunshine on 15, then succession of heavy showers.Heavy rain from 17 until at least 21 June. Flooding in Bombay. Reports of one of the heaviest falls in June remembered.‘Abundant’ rainfall in Thane and Colaba during the second half of June, with banks destroyed and crops flooded.
“We have had little rain and no thunder worth speaking of as yet, and we still wait for our opening easterly squall. But we have during the last four and twenty hours had a thick damp atmosphere with an almost constant drizzle.” Bombay Times, BLNC SM 73, 14 June 1853"We seldom remember a heavier fall of rain in June than we have had during the past ten days.” BLNC SM 73, 20 June 1853
July ‘Abundant’ rain reported across all regions in first half of July, with floods reported in some places. One area only described as ‘insufficient’.Very heavy rain until at least 20 July in Bombay (cleared by 25)‘Seasonable’ rain during late July, with a very few areas described as ‘deficient’.
"Heavy showers" Diary of Isabella Bremner, NLS MSS.19224, 8 July 1853"Since we came down from Matheran we have had constant rain 50 inches of water haven fallen in five weeks we are anxiously looking out for a break." Correspondence of F.P. Lester, BL MSS Eur Photo 431, 20 July 1853
August ‘Light’/’partial’ rain during first half of August, with some reports of damage to crops.Rainfall in the second half of the month described as ‘scanty’.
“COLABA-There had been a slight fall of rain in all the Talookas of the Colaba sub-collectorate… the crops had suffered in some instances from want of sufficient rain.” Substance of the Weather Reports for the fortnight ending 15 August, BLNC SM 73, 3 September 1853
September / October
Rainfall in first half of September ‘seasonable’/’good’ throughout the region. Crops revived.Little or no further rain in September or October anywhere; reports of significant damage to crops.
“TANNA-In the Sungam, Mahim and Bassein districts there was no rain at all, during this fortnight, and in others it was so slight and partial that no general benefit was experienced.” Substance of the Weather Reports for the fortnight ending 30 September, BLNC SM 73, 9 November 1853
Seasonal Summary
Very heavy rainfall during June. Up to mid-July rainfall described as some of heaviest ever known. Much less rainfall in August and September, leading to damage to the crops by October. Water drought reported in Bombay in May 1854.
29
942943
Table 4: Indian Meteorological Department terminology to describe localised rainfall at a weekly or monthly timescale, used to generate monthly indices of monsoon strength. The final category (in italics) is an addition for the purposes of this study.
IMD rainfall category
Percentage deviation from long-period average (LPA)
Reconstructed monthly rainfall class
Scanty < 40% -2
Deficient 40% - 80% -1
Normal 80% - 120% 0
Excess 120% - 160% 1
Heavy > 160% 2
30
944945946
947
Table 5: Calibration table of monthly indices of monsoon strength for Mumbai, describing prevailing characteristics of each monsoon classification noted within documentary archive material during the calibration period.
Rank Classification Characteristics Key Descriptors-2 Scanty Widespread drought reported for the entire or
majority of the month, with no reports of heavy rainfall. Reports of low rainfall unequalled in the recent past.Reports of damage to or failure of crops, with a large temporal and/or spatial extent.
Drought, scanty rainfall, fair, clear, deficient, insufficient, suspension of monsoon
-1 Deficient Generally reports of low rainfall, but balanced out with reports of moderate rainfall over the month, or short period of heavy rain.Intermissions of rain and bright conditions.No reports of flooding; reports of heavy rainfall are spatially or temporally limited.No reports of damage to crops, or reports are limited temporally and/or spatially.
Light, slight, partial, moderate, more or less rain
0 Normal Mixed reports, and not universal throughout the region.Either no widespread drought or flood, or roughly equal reports of droughts and floods.
Regular, sufficient, favourable, satisfactory
+1 Excess Periods of elevated rainfall, interspersed with dry periodsLocalised flooding or reports of crop damage due to moisture, but of limited spatial or temporal extent.
Seasonable, plentiful rain
+2 Heavy Reports of flooding with a large spatial and/or temporal extent, and damage to crops due to excessive moisture.Heavy rain or storms.Reports of heavy rainfall unequalled in the recent past.
Torrential, abundant, considerable, superabundant, continual rain, boisterous conditions
31
948949950
951
Table 6: Rainfall categories and percentage of long-period average as used in the IMD 5-Parameter Statistical Ensemble Forecasting system.
IMD rainfall category
Percentage deviation from long-period average
Reconstructed monthly rainfall class
Deficient < 90% -2
Below Normal 90% - 96% -1
Normal 96% - 104% 0
Above Normal 104% - 110% 1
Excess > 110% 2
32
952953
954
Table 7: Calibration table of seasonal descriptors of monsoon strength for Mumbai, describing prevailing characteristics of each monsoon classification noted within documentary archive material during the calibration period.
Rank
Classification Characteristics
-2 Deficient Widespread drought reported, either during the monsoon year or prior to the onset of the following monsoon season.Reports of damage to, or failure of crops towards the end of the monsoon season;Rainfall reported as deficient for at least 1½ months. (“Scanty”, “deficient”, “mild” or equivalent descriptors).
-1 Below Normal
Mixed reports, with no major reports of drought and famine, or localised droughts.Break periods evident within daily reports, or rainfall described as “showers”, “intermittent” or “light rain” where no daily reports are available.No reports of flooding; reports of heavy rainfall are spatially or temporally limited.Monsoon seasons where rainfall is described as very heavy towards the start, but with damage to crops due to drought reported at the end of the season, may be categorised within this class.
0 Normal Mixed reports, and not universal throughout the region.Either no widespread drought or flood, or roughly equal reports of droughts and floods.Description of an “average” monsoon season.
+1 Above Normal
Very heavy rain reported, but for a short period only (≤1 month).Localised flooding, but not widespread.Reports of low rainfall for one month maximum.Rainfall generally described as “sufficient” (or equivalent), with few reports of “excessive” or “copious” (or equivalent) rainfall.Abundant or above average harvest.
+2 Excess Reports of widespread flooding, and damage to crops due to too much rain.Repeated reports of “heavy” and “abundant” rainfall, or equivalent.Reports of continuous rainfall.
33
955956957
958
Table 8: Regional monthly, seasonal and Western India Monsoon Rainfall (WIMR) rainfall categories for 1838, 1818, 1853, 1793 and 1827. See section 4.5 for descriptions of conditions. N.D. indicates no data.
Year/Region “Monthly” rainfall classifications Seasonal classificationMay/Jun Jul Aug Sep/Oct
1838Mumbai 0 -2 0 -2 -2
Pune 1 -1 -2 0 -2
Gulf of Khambat -2 -2 -1 -2 -2
WIMR -2
1818Mumbai -1 -2 2 1 0
Pune -1 -2 1 -2 -2
WIMR -1
1853Mumbai 2 1 -2 -2 -1
Pune 2 2 -2 -2 0
Gulf of Khambat 2 2 -2 -2 0
WIMR 0
1793Mumbai N.D. N.D. N.D. N.D. 1
Pune N.D. N.D. N.D. N.D. 1
WIMR 1
1827Mumbai 2 0 0 0 2
Pune 2 -2 -1 2 1
Gulf of Khambat N.D. N.D. N.D. N.D. 2
WIMR 2
34
959960961962
963964
Table 9: Kerala teak series low-growth years (Borgaonkar et al. 2010) with reconstructed monsoon indices for the three reconstruction areas and Western India Monsoon Rainfall (WIMR). Years in agreement are shaded. N.D. indicates no data.
Low-growth year Mumbai Pune Gulf of Khambat WIMR1791 N.D. N.D. N.D. 1790: -2
1792: -11794 2 -1 -1 21802 -1 N.D. 1 1801: -2
1802: 01803 -2 -2 -2 -21812 -2 -2 -2 -21814 2 -1 N.D. 21818 0 -2 N.D. -11821 0 N.D. 1 01824 -2 -2 -2 1823: -2
1824: -21829 0 -2 0 -11832 2 -2 -2 01833 -1 0 -2 -21837 1 2 1 11838 -2 -2 -2 -21844 -1 2 2 11853 -1 0 0 0
35
965966967968
Table 10: Correlations between the modified GFENSO index (Gergis and Fowler 2009; see text) and reconstructed seasonal monsoon indices for the three study areas and the combined Western India Monsoon Rainfall (WIMR) record. Correlations post-1860 are calculated using degraded instrumental data.
Years Mumbai(1781-)
Pune(1811-)
Gulf of Khambat(1812-)
WIMR(1781-)
Reconstruction period (earliest stated-1860)
0.09 0.10 0.07 0.09
Instrumental period (1861-2002)
0.12* 0.08 0.14* 0.15**
All available years (1781-2002)
0.11* 0.09† 0.09† 0.13**
† = significant at 90% * = significant at 95%. ** = significant at 99%
36
969970971972973
974
975