A GIS based assessment of potential for windfarms in India

Abstract

The Wind Energy Potential for India has been assessed, assuming as if the whole of the country (apart from the urban and the Himalayan areas) is covered with windfarms, by an innovative approach using GIS platform, wind speed measurements under government's program and the NCEP/NCAR Reanalysis data.(1) The methodology involves setting up a grid of 1 sqkm resolution over areas other than urban settlements and Himalayan regions, computation of wind speeds at boundary layer level through vertical extrapolation of known or measured mean annual wind speed and interpolation of the extrapolated wind speeds to arrive at a mean annual wind field at boundary layer level and then computation of wind speed at the hub-height of the wind turbine. Power output from a standard wind turbine is computed and only areas showing a Plant Load Factor (PLF) higher than 15% are considered in the potential assessment. The results of this exercise indicate the potential for windfarms in India to be significantly higher than what was assumed earlier. The analysis and its revalidation using data measured at varying heights in different parts of the country establishe this approach as useful and perhaps a powerful tool to undertake wind resource potential assessments. This analysis and the results are discussed in the backdrop of the general energy scenario in India and earlier assessments of wind potential in the country.

Full text

A GIS based assessment of potential for windfarms in India
Jami Hossain
a
,
b
,
*
, Vinay Sinha
c
, V.V.N. Kishore
a
a
Center for Energy & Environment, TERI University, New Delhi, India
b
WindForce Management Services Private Limited, Gurgaon, India
c
Department of Natural Resources Management, TERI University, New Delhi, India
article info
Article history:
Received 8 July 2010
Accepted 11 April 2011
Available online xxx
Keywords:
Wind energy
Windfarms
India
Potential
GIS
NCEP/NCAR
abstract
The Wind Energy Potential for India has been assessed, assuming as if the whole of the country (apart
from the urban and the Himalayan areas) is covered with windfarms, by an innovative approach using
GIS platform, wind speed measurements under government’s program and the NCEP/NCAR Reanalysis
data.
1
The methodology involves setting up a grid of 1 sqkm resolution over areas other than urban
settlements and Himalayan regions, computation of wind speeds at boundary layer level through vertical
extrapolation of known or measured mean annual wind speed and interpolation of the extrapolated
wind speeds to arrive at a mean annual wind field at boundary layer level and then computation of wind
speed at the hub-height of the wind turbine. Power output from a standard wind turbine is computed
and only areas showing a Plant Load Factor (PLF) higher than 15% are considered in the potential
assessment. The results of this exercise indicate the potential for windfarms in India to be significantly
higher than what was assumed earlier. The analysis and its revalidation using data measured at varying
heights in different parts of the country establishe this approach as useful and perhaps a powerful tool to
undertake wind resource potential assessments. This analysis and the results are discussed in the
backdrop of the general energy scenario in India and earlier assessments of wind potential in the country.
Ó2011 Elsevier Ltd. All rights reserved.
1. The perspective
India, with its economy growing at more than 6% and likely to
grow at around 8%, continues to face major power and energy
shortages. The shortages are assessed to be in the range of 8%e12%
in energy and from 12% to 25% in peaking capacity [1,2]. Economic
growth and development, may however, result in a larger, currently
latent demand that would have to be met. The total electricity
generation capacity in the country as on November 30 2010 is
167.07 GW [1] and the per capita electricity consumption is rather
low at 733.5 kWh in 2008e09 [1]. According to the integrated
energy policy of the Planning Commission, Government of India,
the installed generation capacity should be at least 800 GW by 2032
[2]. At present, nearly 64% of the generating capacity is thermal and
based on fossil fuels. The option of a similar proportion of fossil fuel
based generation capacity by 2032 is associated with obvious
global environmental concerns.
From various perspectives including that of environment, energy
security, social & rehabilitation issues and economics, India is seriously
constrained on the options for meeting energy demand in future.
The only manner in which the scenario of electricity shortages
and serious environmental risks can be avoided, without foregoing
development, is through a major shift in generation base to renew-
able energies. Among the various renewable energies, in terms of
historical development, commercial viability as well as wide spread
availability, wind turns out to be the most viable resource.
India today (March 31, 2010) has nearly 11,800 MW wind
generation capacity and ranks among the top five countries. With
regard to additional wind power capacity, likely to come up in the
country, there are many questions on the extent to which wind
energy can contribute to meeting the growing energy and power
demand in the country. The main question is - What is the outer
limit of the extent to which wind energy can be harnessed in India
with current state of technology?
*Corresponding author. Center for Energy & Environment, TERI University, New
Delhi, India. Tel.: þ91 124 4353100; fax: þ91 124 4102980.
E-mail address: hossainjami@yahoo.com (J. Hossain).
1
The NCEP/NCAR Reanalysis Project is a joint project between the National
Centres for Environmental Prediction (NCEP, formerly "NMC") and the National
Centre for Atmospheric Research (NCAR). The goal of this joint effort is to produce
new atmospheric analysis using historical data and as well to produce analysis of
the current atmospheric state (Climate Data Assimilation System, CDAS). It contains
the information of climate variables with different time setups. For the analysis,
data has been collected from different data sources like, COADS surface marine
data, the comprehensive Ocean-Atmospheric Data set includes ships, fixed buoys,
drifting buoys, pack-ice buoys, near- surface data from ocean station reports. Still
some work is in progress to collect all the surface marine data for 1947-79. Aircraft
data, Surface land synoptic data, special sensing microwave imager data, surface
wind speeds, satellite cloud drift winds, and satellite sounder data.
Contents lists available at ScienceDirect
Renewable Energy
journal homepage: www.elsevier.com/locate/renene
0960-1481/$ esee front matter Ó2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2011.04.017
Renewable Energy xxx (2011) 1e11
Please cite this article in press as: Hossain J, et al., A GIS based assessment of potential for windfarms in India, Renewable Energy (2011),
doi:10.1016/j.renene.2011.04.017
Author's personal copy

Given the fact that India is a growing economy with deficit in
power and energy availability, the above question assumes
importance from a national planning perspective. This paper makes
an attempt to address this and some of the related questions.
2. Wind resource assessment in India: history and
development
2.1. Historical background
Evolution till date and experiences with technology, resource
assessment and actual installations form the basis of many
assumptions and analysis involved in wind energy potential
assessment. Hossain, Sinha & Kishore [3] and Hossain [4] have
discussed the historical development of wind energy in India in
great detail. Here we present the essential highlights of the
evolution of wind energy in India:
A project led by A Mani and D A Mooley [5] processed and
compiled the wind speed data from meteorological stations all
over the country. The original data in this compilation was not
measured for assessing wind as an energy resource, rather, it
was part of meteorological measurements at airports, ports,
other coastal areas and major towns and cities.
On the basis of this data, the first demonstration wind energy
projects were setup in the coastal areas of the country in 1985e86.
Economics of wind power generation from the first windfarms
in India compared favourably with conventional power
generation (Hossain et. al) [6]. Subsequent studies [7] further
established wind energy as a viable source of energy in India.
A wind mapping programme was initiated by the Department of
Non-conventional Energy Sources (DNES) which has now evolved
into Ministry of New and Renewable Energy (MNRE). Under this
wind mapping programme, which continues till today,so far nearly
600 wind measurement masts have been set up at heights varying
from 10 me50 m [8e11] . Currently the program is being managed
by Chennai based Centre for Wind Energy Technologies (CWET).
A larger number of sites, not known precisely, have been
covered with measurements at up to 80 m height by private
sector. Industry players indicate that nearly 1500 sites may
have been covered by private sector across the country.
2
3. Literature review
Assessment of potential for windfarms is important from
a national planning and policy perspective. In the past, different
kinds of studies have been undertaken in different countries. A
literature review, shows that quite a few studies of this nature or
somewhat similar have been carried out for a region or a country.
Broadly, these studies can be classified as follows:
a) Presentation of wind speed data and analysis involving wind
turbine characteristics and energy computations, thus making
the case for wind energy. Examples are Jagdeesh [12], Hossain
[13] Ramachandra et al. [14], Golait et al. [15], Chang et al. [16]
b) Extrapolation of wind speeds over a geographical area, without
use of a GIS platform. Examples are Aras [17], Hossain &
Raghavan [18],Buflasa et al. [19]
c) Power System studies to assess the penetration levels. Exam-
ples are Hossain [20] [21,22], & Acker et al. [23]
d) Use of GIS Platform. Examples are Hillring & Krieg [24], Denis
[25], Marc & Raymond [26], Mahmmud et al. [27] & Mahmmud
et al. [28]
The paper in the present issue undertakes a near country level
wind resource assessment using the GIS platform.
4. Review of wind potential assessment in India
In the early eighties, the “official”projection for the potential for
windfarms in India was 20,000 MW, referred by us as Wind Potential
Assessment- I (WPA-1). However, the first systematic attempt to
assess the potential for harnessing wind energy for electricity
generation was made in 1992e93 (Hossain& Raghavan) [18], referred
to as WPA-II. After WPA-II, the official projection was revised to
45,000 MW (WPA-III) and remains till today the “official”figure for
the potential of windfarms in India. WPA-II & WPA-III have been
widely quoted in all policy, regulatoryand industry documents.Many
research studies have been carried out assuming these projections as
the outerlimit of the extent towhich windfarms can be set up in India.
To name a few - 1) Golait et al. [15] have concluded that wind energy
has a bright future for India but this future is limited to the figures in
WPA eI & II, 2). Likewise, Mabel & Fernandez [29] have worked on
forecasting the growth potential of windfarms in different states of
India by plotting the S curve for the growth but the S curve also gets
limited by the upper limit of the potential assessed in WPA II & III.
Since 1992e93, the underlying assumptions, database and
technology have changed almost entirely. The key highlights and
the shortcomings of these assessments, particularly WPA-II are
discussed in detail by Hossain, Sinha & Kishore [3]. The changed
scenario is summarized in Table 1.
5. Theoretical background
5.1. Wind speed height extrapolation
Wind speeds measured at a given height can be extrapolated to
another height within the boundary layer (wind turbine hub
height) using the modified power law [30] or 1/7th power law [31],
given by the relation:
V
2
V
1
¼Z
2
Z
1

a
m
(1)
Where V
2
is the average wind speed at the hub height, Z
2
and
a
m
is
the modified power law exponent and V
1
is the average wind speed
measured at the measurement height, Z
1
. The power law exponent
is given by:
a
m
¼a
m
þblnV
1
(2)
Where,
a
m
¼1
lnZ
g
Z
o
þ0:088
10:008lnZ
1
10;(3)
b¼ 0:008
10:008lnZ
1
10;(4)
Z
g
¼ðZ
1
Z
2
Þ
1=2
;(5)
and Z
o
is the surface roughness length.
2
Personal communication wind industry sources, Indian Wind Energy Associa-
tion and Indian Wind Turbine Manufacturers Association.
J. Hossain et al. / Renewable Energy xxx (2011) 1e112
Please cite this article in press as: Hossain J, et al., A GIS based assessment of potential for windfarms in India, Renewable Energy (2011),
doi:10.1016/j.renene.2011.04.017
Author's personal copy

5.2. Wind speed distribution
The wind speed distribution is known to be adequately
described by Weibull distribution [32]. The average wind speed V
can be expressed as a function of the scale parameter “C”and the
shape parameter “k”, by the relationship,
V¼C
G
1þk
1
(6)
The Rayleigh distribution, a special case of the Weibull distri-
bution corresponding to k¼2 is often used to approximate the
wind speed distribution.
5.3. Wind turbine model
The energy output E of a wind turbine, with a power wind speed
curve P(V) in a wind regime having a frequency distribution f(V) in
a time period T is expressed as:
E¼TZ
V
r
V
in
PðVÞfðVÞdVþTPr Z
V
out
V
r
fðVÞdV(7)
Where, Vr is the rated wind speed Vin is the cut-in wind speed Vout
is the cut-out wind speed
5.4. The normalised wind turbine
In a national level wind potential assessment exercise such as
this, there is a need for a neutral power curve representative of all
predominant and commercially available wind turbines. We
decided not to select any one wind turbine type and size but to go
with a generic type. To achieve this, the power curves of different
wind turbine makes, having a rating higher than 1 MW was used
to arrive at a normalised megawatt power curve. The Wind
Turbines considered are Vestas 1.65 MW, Suzlon 1.25 MW, GE
1.5 MW, Regen Powertech 1.5 MW, SEWIND 1.25 MW, SEWIND
2.0 MW. Amongst the above turbine makes Vestas, Suzlon, GE and
Regen Powertech are major suppliers in India and collectively
account for nearly all the wind turbines installed in the megawatt
segment (i.e., >1 MW). The normalised megawatt power curve is
shown in Fig. 1 Annual Energy Output (AEO) from the wind
turbines is computed based on this power curve using Eq. (7),
wherein time T is assumed to be all the hours in an year i.e.,
8760 h.
For a normalised unit megawatt wind turbine, the Plant Load
Factor (PLF) over one year is given in decimals by:
PLF ¼AEO=ð8760*1000Þ(8)
5.5. Weighted inverse distance square interpolation
The weight W
i
to be assigned to any wind speed data point
(CWET or NCEP/NCAR) extrapolated to boundary layer (Boundary
Layer Wind) is given by
W
i
¼O1
y
i

2
=X
n
i
O1
y
i

2
(9)
Where y
i
is the distance of the data point from the central location
in the grid element and the weight operator Ois defined as:
O¼1 for NCEP/NCAR points &
O¼2 for CWET points.
The boundary layer wind (BLW) speed at the grid element is
then given by:
V¼XW
i
U
i
(10)
Where U
i
is the BLW at each wind data point.
Fig. 1. Unit Megawatt eNormalised power curve.
Table 1
The changed scenario.
Sno. WPA I & II Assumptions Change in Assumptions
1) Only a part of barren lands used In setting up 18,000 MW of wind power forest lands, grazing lands, cultivated and
agricultural lands have been used
2) WTG of 55e250 kW rating WTG of w1500e2000 kW being installed
3) Hub height of 20e30 m Hub Height of 80e90 m
4) Rotor Dia of 20e30 m Rotor Dia 80e90 m
5) Max rotor efficiency of around 40% Max rotor efficiency of upto 50%
6) Individual Windfarms of 5e10 MW capacity Individual windfarms of 25 MWe700 MW capacity
6) Only existing sub stations in rural areas
used to evacuate power
Large new and dedicated sub-stations have been set up to evacuate power to the grid.
Special provisions being made to strengthen grid and to set up sub-stations to
evacuate power from wind at high voltage levels
7) Only existing transmission lines to be used New transmission lines as required are also being set up
8) Systemic constraint of its inability to allow
more than 10e20% penetration of wind
This is being questioned worldwide and in many countries like Spain and Denmark,
parts of North Germany and South Tamil Nadu near 20% penetration already exists.
A recent study by NREL has explored use of Compressed
Air Energy Sysytem (CAES) to enhance penetration levels [39]
9) Data based used from 343 Met Stations Around 600 wind monitoring stations plus NCEP/NCAR data
10) Limited experience with existing
windfarms <100 MW
Rich experience with total windfarm capacity w11800 MW
J. Hossain et al. / Renewable Energy xxx (2011) 1e11 3
Please cite this article in press as: Hossain J, et al., A GIS based assessment of potential for windfarms in India, Renewable Energy (2011),
doi:10.1016/j.renene.2011.04.017
Author's personal copy

6. Reassessment of wind power potential in India
6.1. Methodology
The methodology involves setting up a grid of 1 sqkm resolution
over the regions of interest, computation of BLW through vertical
extrapolation of known or measured mean annual wind speed or
the use of NCEP/NCAR re-analysis data, interpolation of the
extrapolated wind speeds to arrive at a mean annual wind field at
boundary layer and then re-computation of wind speed at the hub-
height level above the ground.
The simplifying assumptions in computations and the rationale
for the same are as follows:
1) Power law index computed using surface roughness through
Eq. (2)e(5) is applicable to mean average values of wind speeds
and is used in the analysis uniformly over the grid element
2) Each 1 sqkm grid element is characterised by a single type of
Land Use Land Cover (LULC), which represents the predomi-
nant LULC for the grid element. In selecting grid element of
1 sqkm for this exercise, we have looked at some of the other
geo-spatial data bases being maintained in the country, for
example National Bureau of Soil Survey and Land Use Planning
which maintains a database at 500 m 500 m resolution. In
this database it is assumed that land use and soil types are
homogeneous in 500 m 500 m resolution. In view of this, we
have a reasonable confidence in assuming that information on
land use, vegetation, terrain etc. within 1 sqkm will not be
heterogeneous.
3) In the region of interest, an altitude difference of more than
1 km from the datum does not happen within a radius of 50 km.
Moreover, it is assumed that influence of the values of wind
speed data points (NCEP/NCAR or CWET) on the wind speeds
computed at any grid element is limited to a distance of
50 sqkm i.e., for any grid element data points only within
50 sqkm are considered. The assumption is very reasonable as
the influence of wind data points becomes negligible with
increasing distance from the grid element (Eq (9) and 10).
4) Areas within certain distance of urban and rural population
centres, as described in Table 2, are not suitable for windfarms
and have been masked out of the potential assessment
computation. The smallest population centre masked out had
a population of about 1000. In rural India, the area under
habitat is generally a very small percentage (<10%) of the total
area of the village. We have mentioned in Table 1, that in India
windfarms have come up on all kinds of village lands, some-
times very close to the habitat, therefore, the assumptions of
Table 2 are reasonable and realistic.
5) The assessment does not consider power evacuation infra-
structure (electricity grid) and it is assumed that grid redesign,
its strengthening and development of such infrastructure will
be part of large-scale windfarm development strategy.
6) It is assumed that windfarms do not pose any environmental
hazard and nor will they create any social issues. This
assumption is justified based on our experience in setting up
11,800 MW in India.
7) Environmentally sensitive areas and water bodies, as described
in Table 3 are assumed to be not suitable for windfarm
development.
8) Himalayan and North-Eastern regions of India are not consid-
ered in this exercise due to inapplicability of our techniques,
particularly with regard to height extrapolation and boundary
layer assumptions, as well as certain data limitations.
9) This assessment is not for the entire geographical area of India.
Specifically, we have not assessed potential for the entire states
of Jammu & Kashmir, Himachal Pradesh, Uttarakhand, Aruna-
chal Pradesh and all other States of North-Eastern Region
including Sikkim and Assam. The total Geographical Area of
India is 3287240 sqkm while we have assessed potential over
a region of interest of 2726717 sqkm.
10) The CWET wind speed data used in the analysis is measured
over periods of 1e5 years at different stations and in different
time frames. It is assumed that this data is in general applicable
in the long-term. This assumption is also reasonable as the
analysis with NCEP/NCAR Reanalysis of mean annual wind
speeds since 1948 at several sites indicates an average standard
deviation of around 6%e8%.
11) It is assumed that the normalised megawatt power curve is the
representative power curve for assessment of AEO from the
wind turbines considered in the analysis.
12) The spatial coverage per megawatt of windfarm is computed
on the basis of prevalent megawatt range wind turbines having
a rotor diameter of 80 m and a hub height of 80 m. It may be
noted that in this paper we have used the normalised mega-
watt power curve only to assess AEO and PLF (as defined in
Table 2
Distance from urban centers for suitability of windfarms.
Sno. Population of Urban
Center (thousands)
Distance from the Urban
Center for suitability of
windfarms (km)
1>4000 100
2>3000 75
3>2000 40
4>1000 25
5>100 15
6>50 5
7<50 2
Table 3
Land use land cover type not suitable for windfarms.
Sno. Land use Land Cover Type
1 Mangroves
2 Coral Reef
3 Water Bodies
4 Snow
5 Sea
6 Swamp
7 Mud Flats
Table 4
Region of interest.
Sno. State Area (Sqkm)
1 Andhra Pradesh 275068
2 Bihar 94163
3 Chattisgarh 176750
4 Goa 3702
5 Gujarat 196000
6 Haryana 44222
7 Jharkhand 79714
8 Karnataka 191791
9 Kerala 38863
10 Madhya Pradesh 308245
11 Maharashtra 307731
12 Orissa 155707
13 Punjab 50362
14 Rajasthan 342239
15 Tamil Nadu 130058
16 Uttar Pradesh 243350
17 West Bengal 88752
Total 2726717
J. Hossain et al. / Renewable Energy xxx (2011) 1e114
Please cite this article in press as: Hossain J, et al., A GIS based assessment of potential for windfarms in India, Renewable Energy (2011),
doi:10.1016/j.renene.2011.04.017
Author's personal copy

Eq. (8)) from prevalent megawatt class wind turbines and not
for calculation of geographical area to set up wind turbines. The
normalised megawatt power curve enables us to compute PLF,
which is the parameter for which the potential is assessed.
13) To assess the geographical area we have considered a wind
turbine having a rotor diameter of 80 m. Typically, wind
turbines with 80 m rotor diameter are rated at around 1.5 MW.
6.2. Region of interest
The region of interestin India identified by us is mostly peninsular
India and flat planes in the north comprising mostly of flat or small
scale terrain features. One reason for this is that the techniques of
wind speed extrapolation and interpolation would not be applicable
in complex terrain such as Himalayas. Therefore, the States that have
Himalayan ranges are not considered in the analysis. The States that
have been considered in the analysis are listed in Table 4. The total
Geographical Area of India is 3287240 sqkm while we have assessed
potential over a region of interest of 2726717 sqkm.
6.3. Data
a) The methodology is implemented on the GIS platform (Arc
Editor 9.3) and a dataset comprising of mean annual wind
speeds at 569 out of nearly 600 [11] wind monitoring stations
under the CWET program has been used. This data is available
in processed, compiled and published form and provides
amongst other parameters mean annual wind speeds and
heights of measurement.
Table 5
Land use land cover categories.
Sno. Land-use Category Sno. Land-use Category
1 Mangroves 21 Dry Woodland
2 Junipers 22 Thorn Forest/Scrub (Northern)
3 Alpine Meadow 23 Thorn Forest/Scrub (Southern)
4 Coral reef 24 Abandoned Jhum
5 Irrigated Intensive Agriculture 25 Bush
6 Irrigated Agriculture 26 Coastal vegetation
7 Slope Agriculture 27 Savannah
8 Rainfed Agriculture 28 Plain Grasslands
9 Water Bodies 29 Slope Grasslands
10 Snow 30 Desert Grasslands
11 Sea 31 Alpine Grasslands
12 Tropical Evergreen 32 Sparse vegetation (cold)
13 Subtropical Evergreen 33 Sparse vegetation (hot)
14 Temperate Broadleaved 34 Desert (cold)
15 Tropical Semievergreen 35 Thorn Scrub/Desert (hot)
16 Temperate Conifer 36 Current Jhum
17 Subtropical Conifer 37 Swamp
18 Tropical Moist Deciduous 38 Barren
19 Tropical Dry Decidous 39 Salt Pans
20 Degraded Forest 40 Mud Flats
Fig. 2. Process of implementing wind potential assessment on GIS platform.
Fig. 3. Illustration of the wind extrapolation-cum-interpolation model.
J. Hossain et al. / Renewable Energy xxx (2011) 1e11 5
Please cite this article in press as: Hossain J, et al., A GIS based assessment of potential for windfarms in India, Renewable Energy (2011),
doi:10.1016/j.renene.2011.04.017
Author's personal copy

b) We havealso us edNCEP/NCAR dataReanalysis data use weather
measurements from a variety of sources as inputs to a numerical
atmospheric model in order to produce a description of the state
of the atmosphere, including wind speed. The National Centre for
Environmental Prediction (NCEP) have produced three dimen-
sional global Reanalysis data for a grid with a spacing of 2.5

latitude and longitude at 28 levels above ground. The data are
available four times daily from 1948. We have extracted Mean
Annual wind speeds from a dataset at 10 m above ground level.
c) For land-use land cover and terrain, we have used Advanced
Very High Resolution Radiometer (AVHRR) [33] data. The Land
Use Land Cover (LULC) data is in 40 categories presented in
Table 5.
6.4. On the use of NCEP/NCAR data
We have looked into the reasonableness and suitability of using
NCEP/NCAR data by carrying out literature survey. The findings are
summarised as follows:
Marc & Raymond [26] have compared observed wind speed
data with NCEP/NCAR data and have concluded that the
Reanalysis values are within 1e2 m/s of the observed value and
that the Reanalysis profile does seem to be reasonable. A
variation of 1 m/s- 2 m/s carries a corresponding uncertainty in
AEO estimates. In this issue, use of a combination of actual
measured observations with NCEP/NCAR data has further
minimised this uncertainty.
Winterfeldt [34] has compared NCEP/NCAR data at 10 m.a.g.l.
with another dataset NCEP/DOE-II and in-situ wind speed
observations in the English Channel area. He has concluded
that in offshore, NCEP/NCAR data is in good agreement with
observed wind speed while closer to the coast with in the
English Channel, NCEP/NCAR value is an underestimate by up
to 2 m/s. A main conclusion in this research is that NCEP/NCAR
10 m wind speed forecast is a best guess for the 10 m wind
speed within the NCEP/DOE-II Reanalysis. It can also be seen in
the results presented that NCEP/NCAR either underestimates
observed wind speeds or is in agreement with it but is seldom
overestimating the observed wind speeds.
Table 6
Roughness classes.
Roughness Classes and Roughness Length Table
Rough- ness Class Roughness Length m Energy Index (per cent) Landscape Type
0 0.0002 100 Water surface
0.5 0.0024 73 Completely open terrain with a smooth surface, e.g.concrete runways in airports,
mowed grass, etc.
1 0.03 52 Open agricultural area without fences and hedgerows and very scattered
buildings. Only softly rounded hills
1.5 0.055 45 Agricultural land with some houses and 8 m tall sheltering hedgerows with
a distance of approx. 1250 m
2 0.1 39 Agricultural land with some houses and 8 m tall sheltering hedgerows with a
distance of approx. 500 m
2.5 0.2 31 Agricultural land with many houses, shrubs and plants, or 8 m tall sheltering
hedgerows with a distance of approx. 250 m
3 0.4 24 Villages, small towns, agricultural land with many or tall sheltering hedgerows,
forests and very rough and uneven terrain
3.5 0.8 18 Larger cities with tall buildings
4 1.6 13 Very large cities with tall buildings and skycrapers
Fig. 4. Percentage reduction in wind speed with increase in extrapolation height.
J. Hossain et al. / Renewable Energy xxx (2011) 1e116
Please cite this article in press as: Hossain J, et al., A GIS based assessment of potential for windfarms in India, Renewable Energy (2011),
doi:10.1016/j.renene.2011.04.017
Author's personal copy

Swail & Cox [35] have concluded that the hindcast wind
climatologies closely resemble those obtained from measured
wind data from buoys and offshore platforms.
Research cited above indicates that NCEP/NCAR data provides
a reasonable estimate, where there is no wind speed data available
and in any case NCEP/NCAR data does not seem to overestimate
wind speeds. In this issue we only used NCEP/NCAR data where
there is no data available and also NCEP/NCAR data has been
combined with actual measurements (CWET data) in interpolation
to arrive at wind speeds at a grid element. As presented in Eq. (9) &
10, the influence of NCEP/NCAR data in computing the wind speed
at a grid element is further reduced due to the application of
a lower weight to NCEP/NCAR data in interpolation.
6.5. Approach
The process of computing mean annual wind speeds at any grid
element is illustrated in Figs. 2 & 3. An example of the manner in
which computations are carried out is shown in Fig. 2. Points A & C
are CWET data points at 20 m height and point C is the NCEP/NCAR
data point at 10 m height. D is the central location on the grid
element of 1 sqkm. All these points are not necessarily at the same
altitude level. Using the LULC data the surface roughness length is
assigned to each point in accordance with the defined surface
roughness classes [36], also presented in Table 6. Power law index is
computed using Eqs. (1)e(5). Using the power law index at A, B & C
the wind speeds are extrapolated to a height of 1 km above the
ground. It is assumed that a height of 1 km is well into the boundary
layer and a nominal variation around this height will have
a miniscule impact on wind speeds. This is further illustrated in
Fig. 4, where based on extrapolation of wind speeds using Eq. (1),it
can be seen that incremental increase with height for every 50 m
falls below 1% at around 700 m above ground as compared to 4%e
6% within 200 m above ground.
The rest of the approach is as follows:
a) Compute mean annual wind speeds on each grid element of 1
sq km area by the process outlined in Figs. 2 & 3.
b) The process of extrapolation and interpolation is implemented
for an area covered by a radius of 50 km around each grid
element
c) A shape parameter of 2 or the Rayleigh distribution and the
normalised megawatt power curve are used to compute AEO at
each grid element. It can be seen in Fig. 5 that with Rayleigh
distribution, the sensitivity of PLF to K values from 1.5 to 2.4 for
different mean annual wind speeds is of the order of 10%.
d) Eq. (7) is used to compute the Gross
3
AEO in kWh from the
megawatt wind turbine.
e) The AEO is corrected with factors
4
listed below to arrive at
corrected AEO
a prevalent mean annual air density values in India. Typically
these values vary from 1120 to 1200 gm/cum. For ease in
computation, we have taken a single value of 1160 gm/cum,
which is generally applicable to the region of interest
b machine availability factor of 0.95
c grid availability factor of 0.95
d transmission loss of 5%
e Array loss of 0.90
f) The realistic Plant Load Factor (PLF) also known as the Capacity
Utilization Factor (CUF) is computed using the Eq. (8) but using
the corrected AEO as above in place of AEO
g) Assuming a Wind Turbine of 80 m rotor diameter, the
megawatt potential for windfarms in each grid element is
computed assuming a separation of 10 D (ten times the rotor
diameter), which works out to be a little more than 3 MW/
Sqkm. This is also a conservative assumption as the world-
wide norm for spacing between wind turbines is around 7D
& 5D.
h) Potential is then assessed for each grid element and then by
select land use categories, by PLF range and by State.
6.6. Revalidation
Several locations across the region of interest where the wind
speeds have been measured but are not part of the database (CWET
and NCEP/NCAR) used in the above analysis were considered for
revalidating the results of the model described above.
Fig. 5. Sensitivity of PLF to K values for different mean annual wind speeds.
3
Gross means at the wind turbine level without taking into consideration factors
such as machine availability, grid availability, transmission loss and array loss.
4
These factors are in accordance with industry practice in India and are based on
personal communication with wind farm operators.
J. Hossain et al. / Renewable Energy xxx (2011) 1e11 7
Please cite this article in press as: Hossain J, et al., A GIS based assessment of potential for windfarms in India, Renewable Energy (2011),
doi:10.1016/j.renene.2011.04.017
Author's personal copy

The results of revalidation presented in Table 7 show a Mean
Absolute Percentage Error (MAPE) from the actual measurements
of 9.59%. The corresponding error in energy generation from the
normalised megawatt wind turbine for shape parameter 2 and for
mean annual wind speeds varying from 4.5 m/s to 8 m/s is shown in
Fig. 4. It can be seen that uncertainty in AEO varies from nearly 30%
at low wind speeds to around 16% at high wind speeds (Fig. 6).
7. Results & discussions
Assessment on a GIS platform as described above shows that the
outerlimit (readpotential) of the extent to which windfarms canbe set
up in India with currently prevalent technologies, assuming as if the
whole of the country (apart from the urban and the Himalayan areas)
is covered with windfarms as shown in Table 4 is around 4250 GW.
Potential in Gigawatts at different levels of PLF is shown in Fig. 7,
the wind speed map in Fig. 8 and the PLF map of the region of
interest is shown in Fig. 9. Urban and rural population centers and
settlements have been masked out of the potential assessment
computation in accordance with Table 2. The masking can also be
seen in the maps. Potential under different PLF levels is summar-
ised in Table 8.
The results of this exercise indicate a wind power potential,
which is larger by an order of magnitude from the previous
assessments (WPA-I, II, III). Earlier in the paper, we have described
the shortcomings associated with the earlier assessments. Some of
the main reasons for such a chasm between WPA-I, II, III and
current assessment are:
- In this exercise, nearly 83% of country’s land area is considered,
with the exception of Himalayas, environmentally sensitive
areas and all settlements urban, semi eurban and rural.
However, in the earlier exercise only a certain percentage of
barren lands were considered.
- Earlier only a few CWET measurements at 20 m heights were
considered and now we have considered NCEP/NCAR data in
addition to the latest additions to CWET data and have carried
out exercise at 80 m height. This has vastly changed the scenario
- The technologies considered for deployment are vastly
improved and scaled up over what was considered in earlier
exercises.
In arriving at results, we have onlyconsidered regions that have
the potential to generate electricity at PLF of more than 15%. Nor-
mally, in India, a PLF of 20% determined at a height of 50 m qualifies
for the tariff determined by the Electricity Regulatory Commissions
[37,38],. However, we have noticed that with the advancement
of technology, increase in rotor diameter and hub height as well as
Table 7
Revalidation of the wind speed assessments.
State Site (A) Measured
wind speed
(m/s)
(B) Height of
Measurement
(m)
(C) Power
Law Index
(E) Wind Speed extrapolated to
80 m.a.g.l from measured wind
speed (m/s) using A, B & C
(F) Wind Speed computed by
GIS Platform at 80 m.a.g.l (m/s)
Mean Absolute Percentage
Error between E & F (%)
Gujarat Guj-I 5.96 50 0.21 6.58 7.20 9.45%
Guj-2 6.85 50 0.18 7.45 7.89 5.81%
Rajasthan Raj-I 6.58 65 0.22 6.89 6.51 5.44%
Maharastra Mah-1 6.77 50 0.12 7.16 6.91 3.48%
Mah-2 6.2 74 0.2 6.30 7.87 25.03%
Mah3 6.41 50 0.12 6.78 6.71 1.13%
Karnataka Karn-I 8.05 78 0.12 8.07 6.57 18.63%
Karn-2 7.7 78 0.12 7.72 6.84 11.46%
Karn-3 6.71 63 0.23 7.09 6.20 12.48%
Karn-4 5.82 63 0.25 6.18 6.31 2.12%
Karn-5 5.97 56 0.25 6.53 6.35 2.69%
Karn-6 8.71 65 0.08 8.86 11.06 24.93%
Tamil Nadu TN-I 7.31 78 0.12 7.33 7.76 5.83%
TN-2 6.8 80 0.2 6.80 7.56 11.18%
TN-3 7.18 75 0.2 7.27 7.74 6.42%
TN-4 6.87 50 0.22 7.62 6.93 8.98%
TN-5 5.27 20 0.23 7.25 7.00 3.41%
TN-6 5.6 20 0.15 6.89 7.87 14.11%
Mean Absolute Percentage Error 9.59%
Fig. 6. Uncertainty associated with AEO corresponding to different wind speeds. Fig. 7. Windfarm potential for India exceeding various PLF levels.
J. Hossain et al. / Renewable Energy xxx (2011) 1e118
Please cite this article in press as: Hossain J, et al., A GIS based assessment of potential for windfarms in India, Renewable Energy (2011),
doi:10.1016/j.renene.2011.04.017
Author's personal copy

further optimization and efficiency improvements, many areas that
fall in a PLF category below 20%, will qualify as viable areas. Since,
the potential assessment exercise should be for a fairly long-term
period and as far as possible accounting for cost reductions and
efficiency and technology improvements, we have taken 15% PLF
value as the cut-off level. In arriving at this assessment we also have
to keep in mind that wind turbines in the range 2.5e7 MW, which
have been set-up in Europe but are yet to be set up in India have
a hub height of more than 100 m. An assessment at 100 m or more
can result in a significant PLF improvement and hence a higher
potential estimate. Also if one was to consider the entire
geographical area of the country as well as off-shore potential, the
assessment is likely to be significantly higher.
The analysis shows that there could be an uncertainty of the
order of 30%e16% in energy computations, which if considered
from a conservative viewpoint, may result in the potential assessed
in the PLF class 15e20% slipping downwards. This would still leave
us with nearly 50% of the assessed potential. Even if we were to
consider potential in the PLF category 20%e45% only, we are left
with a figure of 2075 GW, still a very large figure.
If one was to look at just the energy content in 4250 GW oper-
ating at 15e45% PLF, there is enough of it to meet the energy
requirements that would otherwise require 800 GW of conven-
tional capacity by 2032 [2].
An important aspect, we have to keep in mind is implications of
such a large potential for windfarms on land and the many other
competing requirements of land for society. By making an assess-
ment such as this, the authors are not necessarily saying that such
land area should be under windfarms. Our main conclusion is that
there is a vast potential for windfarms in India, which is signifi-
cantly more than what was previously assessed.
In the past, the conventional electricity grid that is designed to
take power from large centralised power stations to remote load
centres was considered to be a major constraint to integration of
utility scale wind energy. Given the fluctuation in wind energy, it
was believed that only 10e20% of wind penetration was possible.
However, in recent times, the idea of higher penetration of wind in
the power systems of the world is gaining ground. According to
a European Wind Energy Association (EWEA) document [39] by
2020, nearly 56% electricity requirements of Irelend and 46.2% in
the case of Denmark will be met by Wind. The authors, therefore,
feel that when assessing the potential for windfarms, it is not
necessary to get limited by the constraints of grid, which can be
resolved by technological solutions and grid management
Fig. 8. Wind speed map of India at 80 m.a.g.l (m/s).
J. Hossain et al. / Renewable Energy xxx (2011) 1e11 9
Please cite this article in press as: Hossain J, et al., A GIS based assessment of potential for windfarms in India, Renewable Energy (2011),
doi:10.1016/j.renene.2011.04.017
Author's personal copy

strategies. The main issue in harnessing the wind energy potential
is that of redesigning the existing electricity grid in order to makeit
compatible with distributed generation. Similarly, the conventional
manner of operating the grid, its economic rationale of despatching
power etc. needs rethinking and reworking. Work is needed in
identifying viable large scale storage technologies such as
compressed air, hydrogen and pumped hydro that go hand in hand
with higher levels of utilisation and integration of wind energy.
Further research will be required in this direction.
Perhaps, combined with other renewable energy sources such
as solar, biomass, small-hydro, the overall renewable energy
potential is enough to meet the energy needs of the country. The
possibility of 100% Renewables India exists!.
This work has demonstrated that GIS platform can be effectively
used in such research and assessment. This being the first attempt
on this scale and resolution, authors are of the view that the
assessment can be further improved upon.
It may be possible to obtain somewhat different figures by
carrying out analysis for still larger wind turbines (>1.5 MW ),
which also come up on towers higher than 80 m.
One is also hopeful that this will pave the way for appropriate
revision in the official figure of the Potential Assessment (WPA eIII)
and also for adequate policy and planning thrust at all levels to
realise the vast wind resource potential in the country.
Fig. 9. Wind Power Potential map of India by PLF (%)
Table 8
Country-wide potential for windfarms.
NET_PLF % Area MW
10e5% 1755.52 5266.55007
25e10% 127182.32 381546.9489
310e15% 548218.46 1644655.394
415e20% 724902.09 2174706.275
520e25% 484697.72 1454093.146
625e30% 136073.93 408221.7989
730e35% 28990.13 86970.39989
835e40% 3572.55 10717.65661
940e45% 38643.55 115930.6352
Total Area (4e9) 2094036.27 4250639.912
J. Hossain et al. / Renewable Energy xxx (2011) 1e1110
Please cite this article in press as: Hossain J, et al., A GIS based assessment of potential for windfarms in India, Renewable Energy (2011),
doi:10.1016/j.renene.2011.04.017
Author's personal copy

Acknowledgements
The GIS platform of the TERI University GIS lab was extensively
used for carrying out this analysis. The authors acknowledge their
gratitude to the lab staff. The authors also acknowledge valuable
inputs by Rohan Kumar of WinDForce in extracting NCEP/NCAR
data and Manoj Sharma of WinDForce for supporting GIS work. The
authors have interacted with a number of industry players to gain
certain insights crucial for this work and this cooperation is also
acknowledged.
References
[1] Monthlyreview powersector (executive summary). accessed on,http://www.cea.
nic.in/power_sec_reports/executive_summary/2010_11/1-2.pdf; January 5, 2011.
[2] Integrated energy planning. Report of the Expert Committee, planning
Commission. Government of India, http://planningcommission.nic.in/reports/
genrep/rep_intengy.pdf; August 2006 [accessed on 05.01.11].20e21.
[3] Hossain J, Sinha V, Kishore VVN. Potential for setting up grid connected
windfarms in India, Large scale integration of windfarms. In: 9th world wind
energy conference and renewable energy Exhibition, 15e17 June, http://
www.wwec2010.com/dosya/program.pdf; 2010. Istanbul, Turkey, last
accessed on January 5, 2011.
[4] Hossain J. The Story of wind power development in India, wind power
development in India. 1st ed. Pune: World Institute of Sustainable Energy,
ISBN 81-902925-0-1; 2006. 15e33.
[5] Mani A, Mooley DA. Wind speed data over India. New Delhi:A lliedPublishers; 1983.
[6] Hossain J, Thukral K, Kishore VVN, Ramesh S, Pachauri RK. Wind Electric
generation: some economic and policy implications. New Delhi: Seminar on
Wind Energy; Feb 12, 1987.
[7] Hossain J, Thukral K, Kishore VVN. Windfarms in India: minimizing the costs
of generation. Journal of Wind Engineering and Industrial Aerodynamics
1988;27:167e78.
[8] Mani A. Wind energy resources Survey for India-I. pp 347. New Delhi: Allied
Publishers, ISBN 81-7023-297-X; 1990.
[9] Mani A. Wind energy resources Survey for India-II. pp 591. New Delhi: Allied
Publishers, ISBN 81-7023-358-5; 1992.
[10] Mani A. Wind energy resources Survey for India-III. Pp 637. New Delhi: Allied
Publishers, ISBN 81-7023-221-X; 1994.
[11] Centre for Wind Energy Technology. List of Wind Monitoring Stations with
MAWS & MAWPD (MNRE as on 30.11.10), http://www.cwet.tn.nic.in/html/
departments_wms.html; January 5, 2011. accessed on.
[12] Jagadeesh A. Wind energy development in Tamil Nadu and Andhra Pradesh,
India Institutional dynamics and barriers - a case study. Energy Policy 2000;
28:157e68.
[13] Hossain J. Estimation of Specific annual energy generation from wind in
Antarctica. Solar & Wind Technology 1989;6(1):91e5.
[14] Ramachandra TV, Subramanian DK, Joshi NV. Wind energy potential assess-
ment in Uttarakannada District of Karnataka, India. Renewable Energy 1997;
10(4):585e611.
[15] Golait N, Moharil RM, Kulkarni PS. Wind Electric power in the world and
perspectives of its development in India. Renewable and Sustainable Energy
Reviews 2009;13:233e47.
[16] Chang T-J, Wu Y-T, Hsu H-Y, Chu C-R, Liao C-M. Assessment of wind char-
acteristics and wind turbine characteristics in Taiwan. Renewable Energy
2003;28:851e71.
[17] Aras H. Wind energy Status and its assessment in Turkey. Renewable Energy
2003;28:2213e20.
[18] Hossain J, Raghavan K. Potential for windfarms in India. In: Kishore VVN,
editor. Renewable energy utilization: Scope, Economics and perspectives. Tata
Energy Research Institute, ISBN 81-85419-01-9; 1993.
[19] Al Buflasa H, Infield D, Watson S, Thomson M. Wind resource assessment for
the Kingdom of Bahrain. Wind Engineering 2008;32(5):439e48.
[20] Hossain J, Nataraj an B, Sinha CS. Potential role of renewables in the power
sector of India. Brussels: a TER I/Commission of European Community;
1991.
[21] Hossain J. High penetrations of renewables in the power grid of Tamil Nadu.
Wind Engineering 1991;15(6):318e29.
[22] Hossain J. Grid integration of renewables in developing countries: a case study
of high wind penetration in the Tamil Nadu electricity utility. Energy Policy;
1993:868e74. August.
[23] Acker TL, Zavadil R, Potter C, Flood R. Wind integration cost impact study for
the Arizona Public Service Company: modeling approach and results. Wind
Engineering 2008;32(No. 4):339e54.
[24] Hillring B, Krieg R. Wind energy potential in Southern Sweden eExample of
planning methodology. Renewable Energy 1998;13(No. 4):471e9.
[25] Elliot D. Wind resource assessment and mapping for Afganistan and Pakistan,
“A presentation under South Asian RegionaI Initiative. accessed on, http://pdf.
usaid.gov/pdf_docs/PNADO338.pdf; June 1, 2010.
[26] Schwartz MN, George RL. On the use of Reanalysis data for wind resource
assessment. In: 11th Applied Climatology Conference. Texas: American
Meteorological Society; January 10e15, 1999. NREL Report No. NREL/CP-500-
25610.
[27] Mahmmud F, Watson S, Woods J, Halliday J & Hossain J, The economic
potential for renewable energy sources in Karnataka, In: Proceedings of
EuroSun 96 Conference, Freiburg, Germany, September (1996), p. 1370.
[28] Mahmmud F, Woods J, Watson S, Halliday J & Hossain J, A GIS tool for the
economic assessment of renewable technologies, In: Proceedings of 18th
British wind energy Association Conference (BWEA18), Exeter, UK, September
(1997), ISBN: 1 86058 034 3, p. 191e196.
[29] carolin Mabel M, Fernandez E. Growth and future trends of wind energy in
India. Renewable and sustainable Energy Reviews 2008;12:1745e57.
[30] Mikhail AS. Height extrapolation of wind data. Journal of Solar Energy Engi-
neering 1985;107.
[31] N Kishore VV. Renewable energy Engineering & technology: a Knowledge
Compendium. New Delhi: Teri Press, ISBN 81-7993-093-9; 2008. pp.
445e501.
[32] Lysen EH. Introduction to wind energy. The Netherlands: SWD; 1982.
[33] http://edc2.usgs.gov/1KM/avhrr_sensor.php.
[34] Winterfeldt J. Intercomparison of Horizontal wind speed components from
NCEP/NCAR-Reanalysis and NCEP-DOE Reanalysis II. Last visited, http://wcrp.
ipsl.jussieu.fr/Workshops/Reanalysis2008/Documents/G2-342_ea.pdf; July 8,
2010.
[35] Swail VR, Cox AT. On the use of NCEPeNCAR Reanalysis surface marine wind
fields for a long-term north Atlantic Wave hindcast. Journal of Atmospheric
and Oceanic Technology 2000;17:532e45.
[36] http://guidedtour.windpower.org/en/stat/unitsw.htm#roughness last visited
on July 1, 2010.
[37] CentralElectricity RegulatoryCommission Notification No: No.L-7/186(201)/2009-
CERC. dated, http://cercind.gov.in/Regulations/CERC_RE-Tariff-Regualtions_17_
sept_09.pdf; September 16th 2009.
[38] Maharashtra Electricity Regulatory Commission (terms and conditions
for determination of re tariff) regulations, http://www.mercindia.org.in/
pdf/Order%252058%252042/Final_MERC_RE_Tariff_Regulation_2010.pdf;
2010.
[39] Pure power - wind energy targets for 2020 and 2030. EWEA update; 2009.
J. Hossain et al. / Renewable Energy xxx (2011) 1e11 11
Please cite this article in press as: Hossain J, et al., A GIS based assessment of potential for windfarms in India, Renewable Energy (2011),
doi:10.1016/j.renene.2011.04.017
Author's personal copy