good paper_epw irrigation in south india doc.rtf

8/14/2019 Good paper_EPW irrigation in South India doc.rtf

1/16

Irrigation and the Great Indian

Rural DatabaseVignettes from South IndiaThe census offers an irreplaceable dataset to examine local settings and study them on

various scales, from a global perspective contrasting regional trends down to villagewisevariations. This paper seeks to demonstrate that irrigation data from the Indian census has

been severely underutilised in this regard, and also offers a view on the issues

that can be researched using village level statistics.

It is more common in India to lament over the lack ofaccurate sources of in-formation than to deplore theunderuse of existing ones. It is therefore not surpris-

ing to read in a recent assessment of irriga-tion statistics in

West Bengal [Rawal 2001] that these data remain

unreliable and in-complete, in spite of the numerous

sources that are scrutinised. Most data users are familiar

with this type of situation and have to struggle with

incomplete or dis-cordant data to study the dynamics of

agri-cultural change. What looks more puzzling is that this

survey of irrigation statistics leaves out one of the richest

sources for irrigation studies in India, which appeared in

the 1990s. This omission is unfortu-nately quite commonand does not concern only the domain of irrigation

statistics. The rural database derived from the census is

probably the most underused source in the study of

contemporary Indian economy.

This brief paper will endeavour to give a rapid survey of

the wealth of data that lies unused in the census

publications. The salient points of the argumentation will

rely on the examination of the irrigation data, which

constitutes but a small fraction of the database. Materials

used in this paper, collected in the course of a research

project on south India, will provide a sample of possible

applications of these data for description, exploration,

identification, analysis and synthesis. Beforehand, I shall

offer a brief overview of the nature of these vastly ignored

data.

Villages,Irrigationand the

Census of India

Since 1951, the Census of India has been collecting and

publishing data about all revenue villages in India.1Data

include

CHRISTOPH

EZ

GUILMOTO

most notably social

and demographic

par-ticulars

assembled from

individual records

collected at the

household level.

The type of data

tabulated at the

village level have

not changed so

much over the

censuses and cover

some of the most

familiar census

data: households,

population, dalit

and tribal

components,

literates, workers

and occupational

categories. The

1991 Census

published not lessthan 42 such

figures (also known

as Primary Census

Abstract) for the

5,80,781 inhabited

villages of India.

What is

surprisingly less

known is the fact

that information is

also collected at the

village level on

infrastructure and

land use. These

data are collected

prior to the

population census

day and cover avery wide gamut

of information

such as edu-

cational, health,

communication

and mar-ket

facilities, as well

as many other

amenities(water

and electricity

supply, etc).

Moreover, when a

given facility is

not locally

available, the

distance to the

near-est locality

where this facility

exists is given. As

part of this

exercise, the

census also

collects village

level data on land

use and irrigationpattern. These

infrastructure data

(also known as

Village Directory)

constitute another

96 variables that

are available for

all inhabited

villages.

In view of the

size of this single

source, it would

not be exaggeratedto call the village

census data the

Great Indian Rural

Database (hereafter

the GIRD). The

GIRD has thuslong included

decennial estimates

of irrigation down

to the village level.

Though yearly

estimates of

irrigation would be

preferable in view

of fluctuations in

land use in rural

India, the census

data carry the

advantage of

identifying all in-

dividual villages

and offering for

each of them

irrigation

estimates. No other

source provides

such a fine

geographical

picture


2/16

of the rural scene.2Landuse data are detailed in five

categories (forest land, irrigated land, unirrigated,cultivable waste, area not available for cultivation)apart from the village total area. As for irrigated areas,they are subdivided into 11 irriga-tion types (Table 1).Although many useful dimensions of irrigation aremiss-ing from the census (such as cropping intensity or

cultivated crops), no less than 13 irrigation statistics areprovided for all Indian villages.

Before examining these in detail, one may wonder why

these data have not been used more systematically to study

agricul-tural development at the local level. The main

explanation may lie in the fact that these data are found

only in the district level volumes published for individual

states. These volumes (District CensusHandbooks) are

usually published manyyears after the census and it is not

uncom-mon for

some of them to be

published while the

next census is

being launched.

These volumes,

which otherwise

contain invaluable

material on villages

and in-cludedetailed district or

taluk informa-tion,

are not easy to

locate themselves

and with the

exception of patient

PhD stu-dents, they

usually do not

reach out to the

potential audience

(planners,

administra-tors,

NGOs, research

scholars).

Moreover, the

printed data

scattered indozens of volumes

are almost

impossible to

compile to get a

larger picture for

taluks, districts or

states. Or so they

were till the 1991

Census.

In fact, the 1991

Census introduced

a radical change in

the dissemination

set-up by offering

data in a computer

format right from

the mid-1990s.Village data were

actually provided

on floppies long

before most of the

district handbooks

were pub-

Economic and Political Weekly March 30, 2002 1223


3/16

Figure1:IrrigationTypesforIndividualStates,1991CensusVillage

Data

cent of the sample) where the

differencebetween these two totals was greaterthan

T a m il Na

d u

0.1 ha.7

The classification of cultivated land for

P o nd iche rry

south India is given in Table 1. As canbe

seen, some irrigation categoriescorrespond

K e

ra la

tonegligible

areas, while someother

categories (well and tubewell irrigation)

K a rnata ka

have been subdivided according topower

sources. There remain, however, two

main

A nd hra P rad e

s h open questions that our global analysis

cannot answer. First, are irrigation data

0%10 %20 % 30 %40 % 50 % 60 %70 %

80

% 90 %10 0

systematically collected and includedin

(Per cent)

the census records? Overall, theanswer

c an a l

w

e

ll

ta n

k

o

th

eseems to be positive. The state totaldoesmatch available data from othersources

for corresponding years. On the micro

level,

we have also tried to locate villagesreport-

lished. Data from the Primary Censustion data as a specific topic fordisaggre-

ing no irrigation in areas with thehighest

Abstract andVillage Directory are given

gat

ed

analysis.5

Th

ese

irrigation

data

irrigation levels (for example, along

the

at the district level in two separate files,happen to be dramatically under-

exploitedKaveri,Godavari and Krishna deltas).These

with an additional file containinglocation in view of their potential in a country

villages happen to be very few andthey

codes and village names. With a littlewhere irrigation is estimated to havein-

are usually clustered in distinctzones,

patience, it is actually possible to

combine creased fourfold since independence.

suggesting that local factors such as

terrain

these two sets of population and infra- A crucial question about these dataaccount for the absence of irrigationfa-

structure files into a single register of

village

concerns their quality, which dependson

cilities. Nevertheless, closecomparison

data for entire districts. As raw data en-the care with which village recordswere

with other aggregated data (at theblock

tered in computer format are not scruti-collected and compiled during the

census.or district level) would probably allowa

nised, inconsistencies and deficienciesare

It must be stressed at the outset thatlack

more accurate assessment of theseesti-

frequent.3 The infrastructure characteris-

of comparable village data prevents

any mates for specific regions.

tics of villages can then be comparedwith

attempt at assessing the precise qualityof The second question relates to the proper

their social and economic composition.the data. Like other censusdemographic

identification of irrigation categories.The


4/16

Once compiled for entire states, the data-estimates, irrigation data are simplythe

11 available categories for classifyingir-

base becomes a powerful tool to investi-only systematic source of informationat

rigation (shown in Table 1) may berespon-

gate rural issues from social,demographic

the village level. What can, however,be

sible for some amount of overlap orcon-

or economic perspectives. The GIRD hasexamined is the internal consistency of

the fusion. For instance, the difference be-

no match to describe and monitor severaldata and here our experience withthese

tween private and governmentcanals,

features of rural development at the mostdata inspires confidence. Using asample

between tubewells with and withoutelec-

decentralised level.4 of 70,260 villages with cultivated areas6,

tricity, or between lake and tank may

be

we checked the quality of the data andat times spurious. It is probably saferto

DataandTheir Quality

found only a negligible number oferro-

aggregate irrigation figures intobroader

neous or incomplete records. We also

notedcategories as we have done for thestate

In this paper, we will illustrate thethat in some cases, the reported totalir-

totals, as shown in Figure 1. All welland

usefulness of this database by using a

rigated area is different from the total

oftubewell categories have beenclubbed

sample derived from a project on fertilityindividual irrigation categories.

However,together, while lake irrigation hasbeen

decline. This sample consists of the entirein our south Indian sample, weidentified

merged with tank. Similarly, riverirriga-

village database for south India, that is,only 116 villages (that is, less than 0.2

pertion is here combined withgovernment

Andhra Pradesh, Karnataka, Kerala,Tamil

Nadu and Pondicherry union territory. ItTable1:IrrigatedandbyCategoriesinSouthIndia,1991VillageCensus

Data

includes a total of 70,984 inhabited vil-

IrrigationType Total Area Share of IrrigatedArealages. The number of villages actuallyused (ha) (Per Cent)

in the analysis may vary according to theGovernment

canal 30,4,3!" 3#$%

data examined because of the statistical Private canal ##,34 0$#deficiencies encountered (missing orerro-

Tan

& #3,',% #!$%

neous data). There is no doubt that com-

)ell*ithelectric

ity "4,3#,""4 "!$4

)ell*itho+telectric

ity ,'#,#"0 "0$3

ments and examples provided here are

T+e*ell*ithelectric

ity 3,%%,#!0 4$0

probably applicable to similar studies thatT+e*ell*itho+telectric

ity '0,#% 0$'

could be conducted in other areas of the

-iv

er ',0,300 '$!

.a&

e ,"4 "$"country. However, in view of the large

aterfall

',(

% 0$"

variety of informationincluded

inthe /ther o+rce 3,"!,3!! 3$4

GIRD, we shall focus here solely on

irriga-

Tot

al 3,#,'%" "00$0

1224Economic and Political Weekly March 30, 2002


5/16

Figure!:"gri#ultural$or%for#eCo&positionand Density

"##ordingtoIrrigationevels,1991CensusVillage Data

cultivators (per cent)

60%

labourers (percent)

50

%

ag. workers (per

ha)

o

f

cultivators (per ha)

40

%

30

%

ce

20

%Pe

10

%

0

%

0-910-19

20-29

30-39

40-49

50-59

60-69

70-

7980-89

90-99

100

Irrigation level in per cent

graphic density. Irrigated villages support

larger populations, most notably because

the workforce engaged in agriculture (in-

cluding dalits) is significantly larger as will

be shown below. The analysis shows also

that higher irrigation goes along with closer

proximity to urban areas, following the

general pattern of Von Thunens model.

While the dalit is positively associated with

irrigation, mainly through the de-mand for

agricultural labour, tribal settle-ments are

obviously deprived of irrigation facilities.

Further analysis of qualitative data on

infrastructure development would also

show the positive association of ir-rigation

on a large array of collective amenities.

Surprisingly, the positive link between

literacy and irrigation appears to be almost

negligible. Obviously, many factors are

simultaneously at play and a

and

priv

atecan

al

irrigati

on.

An

oth

er

pos

-sibl

edra

wb

ack

isthe

fact

that

have

a

popu

latio

n

belo

w

200

for

whic

h

cens

us

estim

ates

are

likely

to be

fragil

e.9

As

Table

2

sho

ws,

all

indi

cato

rs

test

ed

agai

nst

irri

gati

on

turn

out

to

be

sign

ifi-

cant

ly

corr

elate

d

beca

use

of

the

sheer

size

of

our

samp

le.

How

ever,

this

statis

tical

analy

sis

indic

ates

some

intere


6/16

sting corre-

lates of

higher

irrigation

levels. Thus,

the

percentage

of irrigated

area is

stronglyassociated

with canal-

and river-

based

irrigation.

Conversely,

the share of

tank or well

irrigation is

greater when

irrigation

level is

moderate.We shall

come back

below on

this

dimension

of the

irrigation

system,

which is

crucial in

understandin

g the role of

various

irrig

ation

type

s in

Tami

l

Nad

us

agri

cult

ural

deve

lop

men

t.

As

expe

cted,

irrig

ation

appe

arsalso

to be

posit

ively

asso

ciate

d

with

high

er

dem

o-

more

comprehe

nsive

statistical

analysis

would be

called for

in order

to

disentang

le the

specific

associatio

ns

between

irriga-

tion,

social

compositi

on and

various

di-mensions

of social

developm

ent such

as

literacy,

health,

and

schooling

facilities.

An

important

result of

our

analysis

is the

negative

correlati

on of

irrigatio

n with

the

proporti

on of

cultivato

rs. On

the

contrary,

the

proporti

on of

agricultu

ral

labourers in the

total

workforc

e is

much

higher in

irrigated

villages.

This

phenome

non is

better

captured

on the

data

plotted in

Figure 2.

Here, we

have

divided

villages

into 10

irrigation

classes,

starting

from 0-10

per cent

of

irrigated

areas to

100 per

cent.

Unweight

ed

averageshave been

com-

puted for

each class

in order to

display

the

changing

compositi

on of the

agricultur

al

Sta

tis

ti#

al"

nal

ysi

sof

Irr

iga

tio

n

in%ag

es

Th

e

GIR

D

lends

itself

to all

possi

ble

statis

tical

examinati

ons

in

v

i

e

w

o

f

i

t

s

c

o

m

p

u

t

e

r

f

o

r

m

a

t

.

While

data

acquis

ition

was

previo

usly

diffic

ult

from

publis

hed

sourc

es,

comp

uteris

ed

censu

s data

can

be

easily

proce

ssed.

This

allows

syste

matic

descri

ptio

n of

irrig

ation

featu

res

and

a

mor

e

elab

orate

stati

stica

l

anal

ysis.

In

this

secti

on,

we

have

perf

orm

ed a

corr

elati

on

analys

is of

irrigat

ion

level

using

other

vari-

ables

derive

d

from

the

GIRD

. This

analy-

sis

uses

popul

ation

as

statist

ical

weigh

ts as

avera

ge

villag

e size

is

extre

mely

variab

le in

the

differ

ent

social

and

ecolo

gical

re-

gions

of

south

India.

8 The

avera

ge

villag

e size

is

indee

d animpor

tant

featur

e to take into

account before

any serious

analysis, as

more than 17

per cent of

villages in

India

Table

!:Cor

relat

esofI

rriga

tion

evel'

ithot

herVi

llage

Chara

#teri

sti#s,

SouthIndianVillages,

to*n

1alit

Trie

.iteracy

C+ltivator

Agric+lt+rallao+rer

.ive0toc&,fi0hing,fore0

try

2ining, +arrying

o+ehold ind+try

5on6ho+ehold ind+try

Contr+ction

Trade and commerce

Tranportation

/ther ervice

N

o

t

e

s7

Co

rr

el

at

io

n

co

mp

+t

ed

on

%0

,#

'0

vil

lag

e

*it

h

tot

al

pop

+la

tio

n

+e

d

a

*ei

ght

8

Allcoe

ffi

cie

nt

ig

nif

ica

nt

at

"

per

cen

t

lev

el8

1at

afro

m

t

h

e

S

I

9

P

d

a

t

a

a

e

$

E

c


7/16


8/16

workfor

ce with

rising

irrigatio

n.

More-

over, a

simple

index of

agricultural

den

sity

(tot

al

agri

cult

uralwor

kfo

rce

per

hec

tare

of

cult

ivat

ed

lan

d)

has

als

o been

added.

The data

of Figure

2

demonst

rates thatincreas-

ing

irrigatio

n causes

a

sizeable,

gradual

decline

in the

share of

cultivato

rs

among

the

peasant

ry,

going

from 51

per

cent in

thedriest

villages

down

to 29

per

cent in

comple

tely

irrigate

d areas.

At the

same

time,

the

share of

agricultu

ral

labourers

records a

regular

increaseand is

greater

than that

of

cultivato

rs when

more

than half

of the

cultivate

d areas

are

cultivate

d.

Int

ere

sti

ngl

y,

the

inc

rea

se

of

agr

icu

ltu

ral

lab

our

do

es

not

ent

irel

y

o

f

f

s

e

t

t

h

e

d

e

c

l

i

n

e

i

n

cultiv

ators,

as an

incre

ase in

non-

agric

ultura

l

activi

ties

(indu

stry

and

servi

ce) is

also

notic

eable.

At

the

same

time

,

thes

e

irrig

atio

n

data

indi

cate

that

irrig

atio

n is

stro

ngly

con-

nect

ed

to

agri

cultural

den

sity.

Bec

ause

of

mul

tiple

crop

ping

,

irrig

atio

n

has

a

dire

ct

imp

act

on

the

gros

s

areaculti

vate

d

and

ther

efor

e on

labo

ur

dem

and.

This

iscap-

tured

in

Figur

e 2,

wher

e the

avera

ge

ag-

ricult

ural

workf

orce

is

plotte

d

again

st vil-

lage

irriga

tion

levels. Dry

villag

es

with,


9/16

I

r

r

i

g

a

t

i

o

n

l

ev

e

l

(

P

e

r

c

e

n

t

of

c

+

l

t

i

v

a

t

e

d

a

r

ea

)

0

6

"

0

0

'

0

6

0

4

0

6

'

0

#

0

6

4

0

0

6

#

0

say, 10

hectares

of

cult

ivat

ed

lan

d

sup

por

ts

onl

y

8.5

wor

ker

s in

agri

cult

ure

(cul

tiva

tors

and

labour

ers)

.

The

nu

mb

er

of

wor

ker

s in

agri

cult

ure

dou

ble

s

wh

en

the

irri

gati

on

lev

elincr

eas

es

to

50

per

cen

t

and

alm

ost

trip

lesto

24.5

workers

when the

village is

complete

ly

irrigated.

This

differenc

e is

primaril

y due to

the

increasin

g

agricultu

ral

labour

per

cultivate

d

hectare.The

number

of

cultivato

rs per

hectare,

also

shown in

Figure 2,

increasin

g at a

much

slower

pace

when the

irrigatio

n level is

less than

50 per

cent and

remains

constant

thereafte

r. This

meansthat

when

irrigatio

n has

cros-sed

50 per

cent, the

addition

to the

agri-

cultural

workforc

e isentirely

due to

the

number

of

landles

s

laboure

rs and

mar-

ginal

farmers

. These

observa

tions

in-

directly

confir

m that

irrigati

on

tends to

worsenland

inequal

ity

for

which

we can

use the

ratio of

laboure

rs to

cultivat

ors as a

proxy.

Our

models

also

suggest

that any

in-

crease

in

irrigati

on

levelsmay

determi

ne an

increasi

ng

local

labour

demand

. On

the

contrar

y,

labourabsorpt

ion in

areas

where

irrigatio

n

remains

stagnant

may be

more

difficult.

Using

the data

derived

from this

T

hes

e

fea

tur

es

wo

uld

, of

co

urs

e,

req

uir

e

fur

the

r

an

aly

sis

in

ord

er

to

det

er

mi

ne

the

nat

ure

of

the

ca

usa

l

lin

ks

bet

we

en

vill

ag

e

ch

ara

cte

rist

ics

.

Ho

we

ver

,

agr

icu

ltu

ral

pro

s-per

ity

h

a

s

u

n

d

o

u

b

t

e

d

l

y

m

a

n

y

d

i

r

e

c

t

a

n

d

i

n

d

i

r

e

c

t

e

f

f

e

c

t

s

o

n

t

h

e

v

i

l

l

ag

e

struc

ture

that

could

be

exam

ined

using

such

data.

Con-

trary

to

other

sourc

es on

irriga

tion,

the

GIR

D

inclu

des

both

agric

ultura

l and

socio

-

demo

graph

ic

chara

cteris

tics

of

villag

es.

The

analy

sis

can

theref

ore

focus

not

only

on

select

ed

regio

ns,

wher

e

linka

ges

be-

tween

irriga

tion

andpopul

ation

figu

res

wou

ld

be

adm

itted

ly

mor

e

mea

ning

ful

than

for

the

who

le of

sout

h

Indi

a,

but

also

on

othe

r

spec

ific

issu

es.

For

inst

anc

e,

base

d on

the

sam

e

data

set,

our

own

rese

arch

has

sho

wn

the

size

able

role

play

ed

by

irrig

a-

tion

in

fertility

redu

ction.

11

Dem

ograp

hic

data

used

to

investigate

the

deter

mina

nts of

fertili

ty

decli

ne

rarely

incor

porat

edirect

infor

matio

n on

agric

ultura

l

devel

opme

nt

such

as

irrigation

patter

ns.

(

a

p

p

i

n

g

C

e

n

s

u

s

D

a

t

a

:

Con


10/16

toursofI

rrigation

Mapp

ing

vill

age

-

lev

el

dat

a

repr

ese

nts

a

furt

her

imp

rov

em

ent

in

the

GIRD.

While

most

district-

level

informat

ion is

now pro-

vided in

a

cartogra

phic

format,

there is

almost

no

attempt

to map

local

data.

How-

ever, as

most

census

district

handbo

oks

offer

maps to

locate

village

units, it

is also

possibl

e to

draw

maps

of rural

localities

. As part

of our

original

project

on south

India, we

prepared

such

villagewi

se maps

for south

Indian

rural

localities

.12From

1andPoliti

calWe

eklyMarc

h 30, 2002


11/16

Figure):Irrigation*atternsinTa&il+adu,1991Census Data

Irrigationpattern

:nirrigated canal

2i;ed pattern

Tan&ell

India does not conform to the usual ad-

ministrative classification. This map could

be fruitfully enlarged in order to zoom

inon smaller regions. As the database is

extremely detailed, such mapping could berestricted to areas the size of a district.

While the previous map offered a syn-

thetic description of irrigation in south India,

the next one attempts a more ana-lytical

presentation of a region. It is based on the

classification of irrigation types already

shown in Figure 1 and is restricted to Tamil

Nadu, where irrigation levels are the highest

in rural South India.15 Irrigation in Tamil

Nadu corresponds chiefly to paddy

cultivation, while important irrigated cash

crops are groundnut and sugarcane.

We first performed a thorough statisticalanalysis of irrigation patterns and identi-

fied five categories of villages, namely,

unirrigated (irrigation accounts for less


12/16

than 20 per cent of cultivated land); where canal irrigation accounts

for more than 50 per cent of all irrigated areas; well irriga-tion

accounts for more than 50 per cent of all irrigated areas; tank

irrigation ac-counts for more than 50 per cent of all irrigated areas;

mixed pattern corresponds to areas where no single irrigation type

account for more than 50 per cent of all irrigated areas.

this source,

we have

prepared

two irriga-

tion maps

that we

present in

this section.

As rivers in

south Indiaare mostly

seasonal and

the climate

overall

semi-arid

(except for

the western

coast),

irrigation

has been,

during the

recent

period, acrucial

factor in

agricultural

prosperity.1

3

The first

map (Figure

3) covers the

whole of

south India

and consists

of a contourmap of

irrigation

levels.14

Irrigation is

a highly

localised

phenomenon

, the map of

south India

offers a clear

picture of re-

gional

divisions.The map

shows the

east-ern

gradi

ent of

irriga

tion

in

south

India.

Apart

from

pocke

ts

along

the

weste

rn

coast

of

Karn

ataka

and

Keral

a,

wher

e

rainfa

ll is

other

wise

abun

dant,irriga

tion

is

mainl

y

conce

ntrate

d in

Andh

ra

Prade

sh

andTamil

Nadu

. This

is

chiefl

y

relate

d to

major

rivers

such

as

Godavari

or

Kave

ri, which

flow

eastwards

into the

Bay of

Bengal.

The basins

and delta

determinat

e large

irri-gated

areas that

are clearly

discernibl

e on the

map.

Moreo

ver, old

and new

forms of

alter-

native

irrigatio

n

through

tanksand

tubewell

s allow

irrigatio

n in

other

areas in

the

interior,

most

notably

in entire

Tamil

Nadu

and

coastal

Andhra,

where

irrigatio

n levels

are

rarely

below

20 per

cent ofcultivate

d areas.

The

Deccan

plateau,

which

cuts

across

the

Karnata

ka-

Andhra

Pradeshborder,

appears

as a

uniform

region in

terms of

irrigation.

Most

villages

enjoy no

irrigation

at all,

except in

isolated

tracts

clustered

around

major

dams(Krishnar

aja Sagar,

Nizam

Sagar,

Nagarjun

a Sagar)

or along

some of

the

tributaries

of the

major

rivers.

As the

original

database

is

extremely

fine,

contours

do not

follow

district or

even taluk

boundaries. On the

contrary,

they

divide

administr

ative

boundarie

s, in-

cluding

state

boundarie

s. Many

isolated

pockets

appear on

the map,

iden

tifyi

ng

spec

ific

sub-

regi

ons

with

disti

nct

irrig

a-

tion

syst

em.

This

sugg

ests

that

the

ecolo-

gica

l

zoni

ng

of

irrig

atio

n

patt

erns

in

sout

h

a

t

a

f

o

r

t

h

e

1

5

,

0

0

0

-

o

dd

v

i

l

l

a

g

e

s

i

n

T

a

m

i

l

N

a

d

u

w

e

r

e

f

i

r

s

t

m

e

r

g

e

d into

1,442

spati

al

clust

ers

befor

e the

carto

graph

ic

analy

sis.

For

each

of

these

clust

ers,

the

irriga

tionpatter

n has

been

comp

uted

along

the

for-

mula

given

abov

e.

Unirr

igate

d

tracts

(less

than

20

per

cent

of

the

cultiv

atedarea)

are

show

n as a

blank

on

the

map.

As

mixe

d

areas

are

few

in

numb

er

and

irri

gati

on

typ

es

spat

iall

y

con

cent

rate

d,

the

ma

p

pro

vid

es a

coh

erent

divi

sion

of

Ta

mil

Nad

u

into

co

mp

act

irri

gati

on

zon

es.

The

can

al

typ

e

pre

do

minates

in

the

Kav

eri

delt

a,

fro

m

Tir

uchi

rap

alli

dow

n to

the

ocean

, as

well

as

along

in

other

isolat

ed

pock

ets

along

the

coast

and

in

weste

rn

Tamil

Nadu

. The

mainarea

of

tank

irriga

tion

is

found

to the

south

of the

Kave

ri

river,

in an

area

that

stretc

hes

from

Pudu

kkott

ai to

Rama

natha

puram.

Smal

ler

tank-

irriga

ted

pock

ets

are

also

visibl

e

along

the

coast.

Well


13/16

irrigation

covers most

of the inland

tracts, which

are

characterise

d by red and

black soils.

Several

large

-

scale

pock

ets of

dry

areas

are

also

found in

different

parts of

the state,

be it on

the coast

or along

the

Karnataka

border.

This

patterni

ng of

irrigatio

n as

obtained

in Tamil

Nadu

closely

reflects

geophysic

al

dimensio

ns such

as the

drainage

system,E

c

onom

ic

and

Poli

tical

Week

ly March 30, 2002

1227


14/16

the soil map as well as rainfall distribution.

Moreover, because of the relationship of

irrigation with the social and economic

profile ofrural areas, the irrigation map has

also important consequences on the settle-

ment pattern and on labour absorption

capacity as previously mentioned. Of late,

excessive reliance on groundwater in Tamil

Nadu has caused water levels to fall, while

over irrigation is also responsible for soilsalinisation.This means that both economic

and environmental concerns warrant a

detailed monitoring of the progress of

irrigation.This type of irrigation map,

which cannot be produced from any other

source,16would therefore deserve a sepa-

rate analysis of its own.

Con#lusion

This overview has confirmed that irri-

gation data from the census have beenseverely underutilised so far. Even while

other sourcescomprise important additional

information, the census offers an irreplace-

able dataset to examine local settings and

study them on various scales, from a global

perspective contrasting regional trends

down to villagewise variations. The paper

has tried to offer a general view of several

research issues that can be fruitfully inves-

tigated with the village statistics.

The sheer size of the information col-

lected in the GIRD has prevented so far a

systematic analysis by the Census of India.Analytical publications on infra-structure

data are very rare and limited in their

thematic and geographical scope. Scholars

should contribute to the census enterprise

by attempting a more systematic

description and quality assessment of the

village data. This can be followed by more

detailed studies of various socio-economic

or infrastructural features. In the case of

irrigation illustrated in this paper, the

analysis would probably start with the ex-

amination of irrigation patterns as exem-plified in our presentation of the Tamil

Nadu map. Linkages between irrigation on

the one hand and labour composition,

economic or social development, and local

infrastructures on the other, may represent

further topics for in-depth investigations

based on the GIRD.

The 2001 Census has now been success-

fully completed and the preliminary re-

sults are already available in a variety of

formats. When village level data become

available, there is no doubt that they will

be doubly interesting. Not only will theyoffer the best outline of the unequal impact

of economic development in rural areas at

the beginning of the 21st century, but

combined with 1991 data, they will also

minutely describe the dynamics of change

during a decade that has witnessed some of

the most important developments in Indias

economic history. It may become then

difficult to study the development of

irrigation and the redeployment of agricul-

tural labour in rural areas without using

census data to explore regional trajecto-

ries. It is hoped that this presentation of the

GIRD will revive the scholarly inter-ests in

changing rural contexts while sug-gesting

new tools for analysis.

+otes

[The database and maps were prepared for the

South India Fertility Project, a joint research

programme conducted at the French Institute of

Pondicherry, with support from the Wellcome

Trust. The help from Z Headley, S

Vingadassamy, R Amuda and Allah Pichay is

gratefully acknowledged.]

1 For more information on the Census of India,

see their efficient web site:

www.censusindia.net.

2 As Rawals paper reminds us, data from theagricultural census are often based on

sample surveys and are not available below

the district level.

3 Matching infrastructure and socio-demo-graphic data for villages is not as easy as it

seems. Users will often discover puzzling

errors in village names or identification

codes that prevent any automatic coupling of

the two sources.

4 The large survey conducted by the NCAERin 1994 [Shariff 1999] covered only 1,765

villages and its findings cannot therefore be

directly compared with the census.

5 However, research using the south Indiadatabase has been already initiated on many

other subjects such as female discrimination in

Salem areas, deforestation in Kodagu district,

and fertility decline, etc. See for example the

analysis in Ramakrishnan et al (2000).

6 A good 724 villages, mostly located on forestland, report no cultivated land at all and cannot

be used for the rest of our analysis.

7 Errors among other infrastructural data fromtheVillage Directory are much more frequent

than among acreage estimates.

8 Due to its peculiar ecological and admini-strative features, the average population of

Keralas villages is of 15,000 inhabitants,

that is, 10 times more than in the rest of

south India. Conversely, census villages in

some areas like the Mysore plateau or the

tribal areas of the northern Circars have

often populations lesser than 500

inhabitants.

9 For regional comparisons, see Census of

India (1997).

10To compute this 8.5 per cent value, we first

regressed agricultural density against irrigation

levels. From this model, we deduced the


15/16

irrigation increase corresponding to a 13.2 per

cent increase in the workforce; 13.2 per cent is

the total population increase between 1991 and

2001 for the entire region of south India.

11 Villagewise fertility levels are negatively

correlated with irrigation levels, even after

controlling for the effects of a large number

of other fertility determinants. See

Chakrabarty and Guilmoto (2001).

12 The map and database for TamilNadu have been subsequently published in

CD format. See Guilmoto et al (2000).

However, village boundaries have not been

digitised.

13 Yields in irrigated areas are onaverage twice that of rain-fed agriculture.

See Gulati and Kelley (1999) for a

disaggregated analysis of agricultural growth

in semi-arid areas.

14 As the number of village units washuge, villages lying within a 5-km radius

were aggregated, resulting in a simplified

database of 6,986 spatial units. More detail

on the preparation of maps in this paper isavailable from the author.

15 The map has been prepared inrelation to a research project on tank

irrigation conducted by Olivia Aubriot

(CNRS/FIP). A trend map of irrigation levels

in Tamil Nadu is also provided in Guilmoto

et al (2000). More maps on our web site

(http://members.rediff.com/ sifp/).

16 Detailed maps of irrigation in TamilNadu based on administrative divisions may

be found in Ramesh and Tiwari (1983).

eferen#es

Census of India (1997): Availability of

Infrastructure Facilities in Rural Areas of

India: An Analysis of Village Directory Data,

Office of Registrar General, New Delhi.

Chakrabarty, M and C Z, Guilmoto (2001): An

Analysis of the Determinants of Fertility

Behaviour in South India at the Village

Level, unpublished manuscript.

Guilmoto, C Z, S Oliveau, S Vingadassamy and

R Amuda (2000): South Indian Population

Information System, Volume I: Tamil Nadu

and Pondicherry, CD-ROM published bythe

French Institute with support from the

United Nations Fund for Population,

Pondicherry.

Gulati, Ashok, and Tim Kelley (1999): Trade,

Liberalisation and Indian Agriculture

Cropping Pattern Changes and Efficiency

Gains in the Semi-Arid Tropics, Oxford

University Press,.

Ramakrishnan, P S et al, (eds) (2000):Mountain

Biodiversity, Land Use Dynamics, and

Traditional Ecological Knowledge, Oxford

and IBH, New Delhi.

Ramesh, A and T S Tiwari (1983):Basic Resources

Atlas of Tamil Nadu, Department ofGeography,

University of Madras, Madras.

Rawal, Vikas (2001): Irrigation Statistics in

West-Bengal, Economic and Political

Weekly, XXXVI, 27, July 7, 2537-44.

Shariff, Abusaleh (1999): India: Human

Development Report. A Profile of Indian

States in the 1990s, National Council of

Applied Economic Research, Oxford

University Press, Delhi.


16/16

1228 Economicand PoliticalWeekly March 30,2002

good paper_epw irrigation in south india doc.rtf

Documents