Research Reports

IWMI’s mission is to improve water and land resources management for food,livelihoods and nature. In serving this mission, IWMI concentrates on the integrationof policies, technologies and management systems to achieve workable solutionsto real problems—practical, relevant results in the field of irrigation and water andland resources.

The publications in this series cover a wide range of subjects—from computermodeling to experience with water user associations—and vary in content fromdirectly applicable research to more basic studies, on which applied work ultimatelydepends. Some research reports are narrowly focused, analytical and detailedempirical studies; others are wide-ranging and synthetic overviews of genericproblems.

Although most of the reports are published by IWMI staff and their collaborators,we welcome contributions from others. Each report is reviewed internally by IWMI’sown staff and Fellows, and by external reviewers. The reports are published anddistributed both in hard copy and electronically (www.iwmi.org) and where possibleall data and analyses will be available as separate downloadable files. Reports maybe copied freely and cited with due acknowledgment.

i

International Water Management InstituteP O Box 2075, Colombo, Sri Lanka

Research Report 120

Hydrological and Environmental Issues ofInterbasin Water Transfers in India: A Caseof the Krishna River Basin

Vladimir Smakhtin, Nilantha Gamage and Luna Bharati

The authors: Vladimir Smakhtin is a Principal Scientist in Hydrology and Water Resources;Nilantha Gamage is a Research Officer in Remote Sensing and GIS; and Luna Bharati isa Researcher in Hydrology and Water Resoureces, all at the International WaterManagement Institute (IWMI), Colombo, Sri Lanka.

Acknowledgements: The study forms part of the research project on the assessment ofthe National River Linking Project. The project is funded by the CGIAR Challenge Programon Water and Food. Some of the figures for this report were produced by S. Gunasingheand M. Anputhas (both of IWMI-Colombo). Continuing data acquisition efforts by A. Gaurand K. Anand (both of IWMI-India) are gratefully acknowledged. The authors are grateful toHugh Turral (IWMI-Colombo) and to Tom McMahon (University of Melbourne, Australia) forconstructive comments on this report.

Smakhtin, V.; Gamage, M. S. D. N.; Bharati, L. 2007. Hydrological and environmentalissues of interbasin water transfers in India: A case of the Krishna River Basin. Colombo,Sri Lanka: International Water Management Institute. 35p (IWMI Research Report 120)

/ water transfer / river basins / case studies / water resources development / reservoirs /dams / India /

ISSN 1026-0862ISBN 978-92-9090-682-7

Copyright © 2007, by IWMI. All rights reserved.

Cover photograph: The collage on the front cover of this report was created by Mr. SumithFernando, Layout and Graphics Specialist, IWMI, using two original photographs. Thephotograph on the left shows the left bank main canal of the Nagarjunar Sagar Project onthe Krishna River in Andhra Pradesh State of India (photo credit: Mr. Jean-Philippe Venot,IWMI, Hyderabad, India). The photograph on the right shows the Tungabhadra Reservoiron the Krishna River in Karnataka State of India (photo credit: Mr. Daan van Rooijen, IWMI,Ghana)

Please send inquiries and comments to: iwmi@cgiar.org

IWMI receives its principal funding from 58 governments, private foundations, andinternational and regional organizations known as the Consultative Group on InternationalAgricultural Research (CGIAR). Support is also given by the Governments of Ghana,Pakistan, South Africa, Sri Lanka and Thailand.

iii

iii

Contents

Summary V

Introduction 1

Water Transfers in and out of the Krishna River Basin: A Review 3

How Much Water Is Actually Available for Transfers? 9

Environmental Impacts of Reservoir Construction on theGodavari and Krishna Deltas 18

Conclusions 26

References 29

iv

v

This report examines aspects of hydrological andenvironmental feasibility of interbasin water transfersin India and forms part of the larger research projectwhich deals with multiple aspects of the NationalRiver Linking Project. The study uses the watertransfer links in and out of the Krishna River Basinas examples. It reviews the hydrological andenvironmental sections of existing national feasibilityreports, analyzes the methodology used for theassessment of surface water availability for eachtransfer and illustrates the potential environmentalimpacts of the transfers in the deltas of the Godavariand Krishna rivers. It is shown that the planningprocess, as presented, has not considered thevariability of flow within a year, which is high inmonsoon-driven Indian rivers. As a result, muchmore water may be perceived to be originallyavailable at a site of transfer. The use of alternativetechniques, such as a low-flow spell analysis and astorage-yield analysis, to reevaluate the availability ofthe surface water at proposed transfer sites isadvocated. It is shown that water transfer planning is

based on the maximum projections for futureirrigation adopted by each state which falls withineach river basin. This boosts irrigation requirementsand serves as the driver for future water resourcesdevelopment. It is emphasized that environmentalwater demand needs to be calculated (using thedesktop technique developed earlier) and explicitlyincluded at the planning stage—similar to thedemands of other sectors. This “contingency”demand would reserve some water for environmentaluse in the future, while more detailed nationalapproaches for environmental flow assessment arebeing developed. Environmental impacts of reducedwater and sediment inflows to the Godavari andKrishna deltas are examined in the context of themost downstream link from the Godavari (Polavaram)to the Krishna (Vijayawada). It is shown that theKrishna Delta has retreated noticeably during thelast 25 years. Environmental flows need to beprovided to at least delay this “shrinkage” whichthreatens agricultural production and mangroveecosystems.

Summary

vi

1

Hydrological and Environmental Issues of InterbasinWater Transfers in India:A Case of the Krishna River Basin

Vladimir Smakhtin, Nilantha Gamage and Luna Bharati

Introduction

The National River Linking Project (NRLP) wasproposed as the solution to water-related problems inIndia. It envisages transferring water of the Ganga,Brahmaputra and Meghna rivers through the Mahanadiand Godavari river basins, all normally referred to as“water surplus” basins, to the “water-deficient” basinsin the south and west (e.g., http://www.riverlinks.nic.in/). The NRLP is a contentious issue in Indian societyand the media and amongst academics. Manyscholars argue that the needs assessment of NRLPis inadequate. Others are of the view that theassessment of water surplus/deficits in Indian riverbasins, conducted as part of the NRLP proposal, hasignored environmental issues. Yet, others think thatdefinitions of surplus and deficient basins need to bemade more explicit and that alternative watermanagement options, those that are less costly,easier to implement and more environmentallyacceptable, have not been considered.

Extensive work has been done in India onvarious aspects of water transfers related to theNRLP. However, the project as a whole has notreached implementation which, to a certaindegree, mirrors the fate of some other large-scalewater transfer projects in the world. At the sametime, some individual NRLP links are about to beconstructed. Perhaps, one of the major reasons forthe slow development of the project is the lack ofclarity and transparency in technical design,justification of transfers and decision making on theone hand, and the enormity of both the challengeand the scale of the transfer on the other. In anideal world, any water transfer project may be

justified if it satisfies the following broadly definedcriteria (Interbasin Water Transfer 1999):

1. The area of delivery must face a substantialdeficit in meeting present or projected futurewater demands after considering alternativewater supply sources and all reasonablemeasures for reducing water demand.

2. The future development of the area of originmust not be substantially constrained by waterscarcity; however, consideration to transfer thatconstrains future development of an area oforigin may be appropriate if the area of deliverycompensates the area of origin for productivitylosses.

3. A comprehensive environmental impactassessment must indicate a reasonabledegree of certainty that it will not substantiallydegrade environmental quality within the areaof origin or area of delivery; however, transfermay be justified where compensation to offsetenvironmental injury is provided.

4. A comprehensive assessment of socioculturalimpacts must indicate a reasonable degree ofcertainty that it will not cause substantialsociocultural disruption in the area of origin orarea of water delivery; however, transfer may bejustified where compensation to offset potentialsociocultural losses is provided.

5. The net benefits from transfer must be sharedequitably between the donor area and thereceiving area.

2

The International Water Management Institute isconducting a research project, which aims tohighlight, discuss and – where possible – resolvesome of the controversial issues pertaining to theNRLP thus stimulating the debate on India’s waterfuture. This report is one of the multiple outputs ofthis research project. The primary focus of the reportconcerns the hydrological feasibility andenvironmental impacts of NRLP, which are reflectedby criteria 1, 2 and 3 above. It is not the objectiveof the report to analyze all NRLP links from allpossible angles of technical and environmentalfeasibility. The authors rather aim to a) identify andexamine those technical and environmental aspectswhich may have been underappreciated in previousdiscussions on NRLP and need to receive furtherattention, and b) illustrate their importance on oneor several (but very few) links. More specifically, first

this report briefly describes the proposed links inand out of the Krishna River from/to adjacent riverbasins (Figure 1). Krishna is a major river basin,spanning three states in peninsular India.1 This isfollowed by the discussion, using some links asexamples, on how water transfer planning may beaffected by the resolution of the hydrological data.The report further focuses on the environmentalaspects of one of these links: Godavari (Polavaram)-Krishna (Vijayawada); Figure 2. This link is the mostdownstream one in the Godavari-Krishna systemand one which is currently being constructed. Acompanion report by Bharati et al. (n.d.) discussesthe multiple aspects of water management of thePolavaram- Vijayawada link and examines theimpacts of water management options andscenarios using an Integrated Water ResourcesEvaluation And Planning (WEAP) model.

1The Krishna River Basin is one of five “benchmark basins” in which IWMI conducts research, where the intention is to integratevarious strands of biophysical, socioeconomic and institutional research around the world.

FIGURE 1. A schematic map of India, showing the boundaries of the major River Basins/drainage regions of the country.1, 2 and 3 are Godavari, Krishna and Pennar Basins, respectively.

11

22

33Vijayawada

PolavaramPolavaram

3

Water Transfers in and out of the Krishna River Basin: A Review

In order to assess the degree to which criteria 1,2 and 3 above are satisfied in planning of individuallinks in and out of the Krishna, the relevantchapters of the technical feasibility reports(Hydrology, Environment) produced by theNational Water Development Authority (NWDA) ofIndia have been reviewed. Most of the reports areavailable on the NWDA site in HTML format(http://nwda.gov.in/indexab.asp?langid=1). A briefsummary of each link with the authors’ commentsis given below, starting from the most “upstream”link on Figure 2.

Bedti-Varada Link (Link 14)

This is the only incoming link in the upstream partof the Krishna Basin for which no feasibility reportis available at present. Salient features are listedon the NWDA web site, and some very limitedanecdotal information is available (Dams, Riversand People 2004). This proposal envisages thediversion of 242 million cubic meters (MCM)of “surplus” water of the Bedti Basin (in WesternGhats–flowing west into the Arabian Sea;not shown in Figure 2) to the water-“deficient”

FIGURE 2. A schematic diagram of the Krishna River Basin, showing all proposed interbasin water transfers in and out of thebasin (black lines with numbers) together with flow measuring points (stations) for which some observed flow data were

available for the study. Link numbers are circled and correspond to the overall NRLP numbering system. Station numberingis for identification purposes only. Due to the low quality, short records or inappropriate locations relative to the link points,

only a few of the shown stations are usable. These include records at station 3 (Krishna at Agraharam) and part of therecord at station 1 (Krishna at Vijayawada).

4

Tungabhadra subbasin in Krishna (Figure 2). Thewater will be used to irrigate some 60,200 hectares(ha) of land and for hydropower generation. Twonew dams in the Bedti Basin will be constructedwith a combined total (live) storage of 98 (85.5)MCM. The larger reservoir will be connected by alink canal to a tributary of the Varada River.

So far, no environmental studies have beenconducted around this link. The small tributariesinvolved in this project, however, may be verysensitive to flow changes. Also, located in thetropical humid forests (75% of the area) anddeclared by the International Union for theConservation of Nature and Natural Resources(IUCN) as a biodiversity hot spot, the basins to beaffected host 1,741 species of flowering plants and420 species of birds and other wildlife. Thesenumbers exceed those in the whole Kerala State,where the Bedti Basin is located. The flow will bedischarged into the Varada without a receivingreservoir, which may increase channel erosion inlocalized parts of the river. Altered flow patternsmay cause riparian zone degradation and createhabitats for invasive species. The proposed projectis expected to generate 3.6 megawatts (MW) ofpower but it may take over 61 MW to lift the waterto the Varada.

Krishna (Alamatti) – Pennar Link (Link 05)

This is one of the several links effecting watertransfers from the Krishna Basin to the PennarBasin (Figures 1 and 2). The link starts from theexisting Alamatti Reservoir on the Krishna River(upstream catchment area 33,375 squarekilometers [km2]). This link is seen as a partialexchange for the Godavari water brought into theKrishna (links 02, 03 and 04 in Figure 2).However, since all the inward links from theGodavari bring water to downstream parts of theKrishna, and since the inflow from the Bedti link(if constructed) is minor, this link effectivelytransfers the existing “surplus” water from theupstream reaches of the otherwise “deficient”Krishna Basin into another “deficient” basin, thePennar. The purpose of the link is to satisfy enroute irrigation needs. A volume of 1,980 MCM of

water will be transferred through a 587-km canalwith an outfall into a tributary of the Pennar. Anew (balancing) reservoir with a total (live) storageof 83 (73) MCM is to be constructed at thereceiving end of the Pennar Basin at the Kalvapallivillage, with an upstream catchment area of 5,616km2. The need for this new reservoir may need tobe better justified as there is another dam (theupper Pennar) which commands the catchmentarea of 5,245 km2 – just upstream of theproposed new one.

All water transfers in NRLP are planned from“surplus” basins or parts thereof to “deficit” basins.The basin is declared “surplus” if the balance ofwater “naturally” available (assured) in a river, 75%and 50% of the time on the one hand and the totaldemand for the next 25-50 years upstream of thepoint of a transfer on the other, is positive. If thisbalance is negative, the basin is perceived as a“deficit” one (the details of the methods used toestablish whether a basin is surplus or deficit aredescribed and discussed later in this report). AtAlamatti, the “surplus” water at 75% and 50%assurance (“dependability” – in Indian terminology)is estimated to be 5,611 and 8,247 MCM,respectively, while the corresponding values for thereceiving point of the Pennar at Somasila are -3,820 and -3,590 MCM, respectively. Such a largedifference between surpluses and deficits of thedonating and receiving basins is the majorjustification for the transfer.

The major feature of this link is the long canal,and a lot of attention is paid to the justification of itsdesign and cost. It will pass through reserved forestsand a bear sanctuary, where 17 wildlife species arereported including four endangered ones. Losses of,and disturbances to, habitat due to the lined canal,representing an obstacle to wildlife migration routes,are programmed into the project. It is suggested thatwildlife “will migrate to surrounding forests,” and thusimpacts will be minimal. Possible measures tomitigate the disturbance to the sanctuary includerealigning it, including the establishment of a“minimum protected area.” The Kalvapalli Reservoir isanticipated to provide a waterfront for wildlife. Theequivalent of about US$35,000 (in 2006 dollar terms)is allocated in the project for the improvement of theenvironment.

5

Water pollution in the Kalvapalli Reservoir isanticipated through silting and sedimentation,nutrient leaching and agricultural runoff containingfertilizers and pesticides while common mitigationmeasures, such as contour bunding, are planned.A beneficial aspect of the project is an anticipatedincrease in fish production. The link canal is seenas a facilitator of cross-migration of fish specieswhich will increase fish population overall, althoughno justification for this, or evidence from othersimilar cases, is provided. Most ecological issuesconsidered in this feasibility report are related tothe link canal rather than to the donor or thereceiving rivers per se. It is possible to suggestthat no “ecological” releases from the AlamattiDam are made or planned because there is nomention of such releases.

Krishna (Srisailam) – Pennar Link(Link 06)

This is one of the several links effecting watertransfers from the Krishna Basin to the PennarBasin. The link starts from the existing SrisailamReservoir on the Krishna River (with an upstreamcatchment area of 211,657 km2) at its confluencewith the Tungabhadra River (Figure 2). This link,similar to the Alamatti-Pennar link upstream,effectively transfers the existing “surplus” waterfrom the otherwise “deficient” Krishna Basin intoanother “deficient” basin, the Pennar. This mayresult in less water downstream of the SrisailamDam, and the reach between Srisailam andNagarjuna Sagar dams will become even morewater-deficient. The 75% and 50% assured annualflows at Srisailam are estimated to be 57,398 and66,428 MCM, respectively, although the finalsurplus at 75% assurance at the site after alldemands are satisfied is 6,017 MCM.

A volume of 2,310 MCM of water will be divertedthrough the existing Srisailam right main canal,which will operate 6 months a year from July toDecember (monsoonal and post-monsoonalseasons). The water will be discharged into theNippulavagu, a natural stream, and will reach thePennar River through the Galeru and Kunderutributaries. No new infrastructure is required and no

en route irrigation is planned: the transfer targetsexclusively the destinations of Pennar and Cauverybasins (it has to be noted however that oldertransfers of this nature have resulted in thedevelopment of irrigation along the canal andcapture of that water). As with other links,no provisions exist for environmental releasesdownstream of the Srisailam Dam. Some commonimpacts of water diversions (e.g., sedimentation ofreservoirs, changes in hydrological regime due toflow regulation, waterlogging and salinity caused byirrigation and drainage) are discussed in generalterms.

The major point made with regard to this linkis that since there is no new storage and water isto be transferred through partially concrete-linednatural streams, there are no new submergenceareas, waterlogging, or adverse impacts on floraand fauna. It is suggested that the conveyancestreams can easily carry additional 163 cubicmeters per second (m3/s) of water (the amount ofwater transfer for 6 months in a year) in addition totheir own “natural” discharges. It remains unclearhow these streams will react to extra water during6 months, what the riparian conditions are or howembankments will affect fish spawning.

Krishna (Nagarjuna Sagar) – PennarLink (Link 07)

This is a major transfer of 12,146 MCM of waterfrom, and to, existing reservoirs: Nagarjuna SagarDam on the Krishna (upstream area of 220,705km2) and Somasila Dam on the Pennar. The 75%and 50% assured “natural” annual flows are 58,423and 67,346 MCM, respectively. The purpose is toimprove irrigation en route (where irrigation facilitiesare not adequate) and then to transfer water furtherto the south, where water shortages are said to beseverer (a deficit of 3,820 MCM is envisaged at75% assurance in the Pennar River with allirrigation plans in place). A new 393-km lined linkcanal and an existing right-bank canal fromNagarjuna Sagar will run in parallel over 202 km,because the latter can carry only 3,979 m3/sannually while the proposed transfer is for threetimes more water. Such massive transfers may be

6

possible only due to the chain of transfers fromfurther north. The restructuring of the existing right-bank canal is not possible and therefore theconstruction of a new one is seen as a necessaryoption. Because no new storage is associated withthis link, the feasibility report envisages noenvironmental impacts, and no costs areanticipated for mitigation of such impacts. This linkis effectively part of the much longer water transferline from the north to the south. Additional watertransfer to the Nagarjuna Sagar Reservoir isplanned through the Inchampalli-Nagarjuna Sagarlink (see below).

Godavari (Inchampalli) – Krishna(Nagarjuna Sagar) Link (Link 02)

This link involves the transfer of 16,426 MCM ofwater and the construction of a new major storagereservoir on the Godavari at Inchampalli. Theupstream catchment area at this point is 269,000km2 and the gross (live) storage of the future damis 10,374 (4,285) MCM. A low ratio of a livestorage to gross is noteworthy. The water yieldsof the Godavari at Inchampalli at 75 and 50%assurance are estimated to be 66,193 and 76,185MCM, respectively. The proposed irrigation plansare huge and, in all states involved, they exceedthe sum of existing and ongoing irrigationprojects. These plans are effectively thejustif ication of the transfer. The irr igationrequirement projected for the year 2025 on thebasis of the states’ irrigation plans is 40,723MCM and the balance of all demands (irrigationplus others) at 75 and 50% assurance is 20,327and 29,987 MCM, respectively. The Krishna Riverat Nagarjuna Sagar is estimated to have a deficitof 1,525 MCM at 75% assurance, which isanother justification for the transfer. This watertransfer is justif ied by a large irr igationdevelopment, which in itself will probably takemany years to complete, and its feasibility willdepend on the cost of water provided.

From the environmental side, the majorimpacts are perceived to be related to thesubmergence area of the new reservoir, whichleads to major resettlements. It is suggested

however that aquatic life will develop in the newreservoir and that, for example, the loss ofbreeding grounds of crocodiles in the river due tosubmergence is negligible. The report indicatesthat the project will have an impact on theSingaram sanctuary and submerge 65 hectares ofthe Indravati National Park. It lists the knownpresent fauna and birds in the area, whichindicates no endangered species. No adverseimpacts on aquatic life are identified, but nostudies done to this effect are cited. Afforestationis proposed to compensate for the loss of foreststo submergence.

Godavari (Inchampalli) – Krishna(Pulichintala) Link (Link 03)

This link will divert 4,370 MCM from the Godavariinto a new reservoir on the Krishna at Pulichintala,with a gross storage capacity of 1,296 MCMthrough a new, 312-km link canal. The yields at75% and 50% assurance are estimated to be66,193 and 76,185 MCM and surplus surface waterbalances after satisfaction of all projectedrequirements at Inchampalli are 20,327 and 29,987MCM, respectively. Similar estimates are done forMuneru, Palleru and Musi tributaries of theKrishna.

The feasibility report explicitly suggests that allrequirements of the Godavari downstream ofInchampalli can be met by the water available fromthe incremental catchment area located betweenInchampalli and the Dowlaiswaram Barrage andwith the surplus water transferred from Mahanadi.Therefore, no water is likely to be released fromInchampalli downstream and all water atInchamapalli will be used for diversion to theKrishna. The feasibility report refers to simulationsof the Inchampalli Reservoir at a monthly step overthe period of 1951-1981 supplying both Pulichintalaand Nagarjuna Sagar links (4,370 and 16,426MCM). Simulations suggest that all requirementswill be satisfied with a success rate of 76%. Theenvironmental issues associated with this link arethe same as those with the Inchampalli–NagarjunaSagar link, as they are for a common storage(Inchampalli).

7

Godavari (Polavaram) – Krishna(Vijayawada) Link (Link 04)

This is the most downstream link in both theGodavari and the Krishna basins, and the onewhich is scheduled for construction. It is plannedto divert 1,236 MCM of water from the newPolavaram Reservoir on the Godavari (with a livestorage of 2,130 MCM) to the existing PrakasamBarrage on the Krishna through a new 174-km linkcanal. The transfer is designed to substitutereleases for the Krishna Delta from the NagarjunaSagar Dam and to allow “saved” water to be usedfor other projects in the Krishna. The canal,operating throughout the year, will discharge intothe Budameru, a river which flows into the KolleruLake (which is now effectively a large collection ofaquaculture ponds), and from there the transfer willgo through the Budameru diversion canal,discharging into the Krishna 8 km upstream of thePrakasam Barrage. There is already considerableinfrastructure in the Lower Godavari below theproposed Polavaram Reservoir. Lift irrigation stationsalong the river provide irrigation in the LowerGodavari Delta. This may decrease the total areaclaimed to benefit from the Polavaram link. Thereis also no mention of how, and if, the existingcanals will be integrated into the new canalsystem.

Approximately US$600,000 (0.2% of theproject cost) is allocated a) to study the“environmental and ecological” aspects of theproject by various organizations, and b) forprotective measures as may be necessary. Sinceboth donor and receiving points are nearly at theoutlets of the Godavari and Krishna rivers,environmental impacts may only be felt in bothdeltas and en route the canals, where newirrigation, domestic and industrial requirements aretargeted. Possible adverse impacts mentioned inthe report include resettlement, submergence offorests, waterlogging and salinity in the commandarea. Planned mitigation measures include drainagesystems in the command area to mitigate salinity,fish ladders through the Polavaram Dam to allowfor movement of migratory fish, and studies of thenature of existing aquatic weeds in the submergedarea and some others.

The National Council of Applied EconomicResearch (NCAER), New Delhi, India, wasentrusted with the studies of socioeconomic andenvironmental implications of six interbasin watertransfers including this one (Agricultural FinanceCorporation Ltd. 2005). Their report indicates thatthe wild sanctuary in the proposed PolavaramReservoir area will be marginally affected by thesubmergence, and the list of fauna in the areacoming under submergence is given district bydistrict. It is also suggested that wildlife conditionswill actually improve due to the broad expanse ofwater in the new reservoir which is conducive tobreeding of wildlife. The scientific basis for theseconclusions is however unclear from the report. Itis envisaged that endangered species (tiger,panther) will move to deeper forest areas away fromthe submerged areas.

It is indicated that after the Dowlaiswaramanicut has been constructed on the Godavari, fishmigration (e.g., hilsa) from the sea to inland hasbecome obstructed. It is stated that dams converta river to a more placid lotic environment withreduced velocities, which impacts fish species andcomposition and size. However, no quantitative,link-specific conclusions are presented. Genericstatements are also made about phytoplankton,seasonal flow pattern changes, etc. It is alsoadmitted that the entire command area lies in thecoastal belt with high rainfall, enhancing the riskof malaria, while a few general statements aremade about vector breeding and a possibleincrease in waterborne diseases.

The Environmental Management Plan describesa variety of relevant measures including catchmentarea treatment through vegetative measures andstructures (to reduce inflow of extra sediments intothe reservoir), development of flora and faunathrough compensatory afforestation, enhancingaquaculture through stocking of the new reservoirwith exotic fish species, relocating somearcheological structures and disaster management(concluding that there is no possibility of damfailure because probable maximum flood will bepassed by the structure). The report however doesnot address deltas – relevant environmental issuessuch as reduced flow and sediment to deltas dueto dam construction, resulting in stunted delta

8

growth, seaside erosion or degradation ofmangroves.

General Observations

Overall, all NWDA feasibility reports are succinctsummaries of the proposed interbasin watertransfers. They have similar structures and levels ofdetail, and represent, effectively, the only source ofpublicly available technical information on theproposed transfers. As such, they are veryvaluable.

At the same time, they have similarshortcomings. The information presented remainslimited and it is not possible to judge about thequality of the data used. Environmental aspectsand impacts of the proposed projects are onlygenerally described and are primarily related to thesubmergence area associated with new reservoirsand to resettlement of the population affected. It isclear that no provision is made for in-streamecological releases from either existing or plannedreservoirs. If a proposed link is to flood orotherwise affect existing wildlife sanctuaries, thelatter are expected to be relocated/compensated,implying their relatively low importance. The generalcomments on environmental impacts make noreference to the link/site in question and cite nosupporting studies. Technical aspects of some

links need more clarity. For example, Bedti-Varadalink does not seem to be justified from thehydropower angle (as it will produce far less energythan that used to get the water to it). Links startingfrom the Lower Godavari include the construction ofa new Inchampalli Reservoir, which is designed tohave a very low ratio of live to gross storage,making it a huge evaporation tank. The entirecomplex of interbasin water transfers is driven bysignificant irrigation expansion which extends to2050. At the same time, it is not entirely clearwhere this new land for irrigation expansion islocated because most of the proposed “new”irrigated land in the Krishna and Godavari basinsis likely to be irrigated already (H. Turral, IWMI,pers. comm.). The approach can benefit morefrom a more integrated, basin-wide waterresources planning. At present, water is plannedto be transferred from the upper parts of theKrishna Basin, while at the same time otherl inks wi l l del iver water into the Krishnadownstream. The reported low Benefit-Cost (B/C)ratio of some projects is also noteworthy. Forexample, Alamatti-Pennar and Polavaram-Vijayawada links have a B/C ratio of around 1.2each, which makes the effectiveness of theselinks questionable. Finally, the methods by whichwater availability for the transfers was calculatedrequire some comment and are discussed in thenext section.

How Much Water Is Actually Available for Transfers?

A Summary of the “Official” WaterResources Planning Method

The methodology that the NWDA is using inplanning water transfers is essentially the same forall links and is described in abbreviated form inevery individual feasibility report. It is important toattempt to spell it out here because the NRLP hasbeen criticized for not describing the basis onwhich the assessment of water availability and

identification of surplus and deficient river basinshave been made. This is a misconception becausethe issue is not so much that it is unclear, butrather whether it is entirely appropriate given thescale of transfers. The overall planning approachincludes several sequential steps.

The catchment upstream of the diversion point(donor) or receiving point (receiver) is split intoseveral smaller subbasins to cater for spatialvariability of rainfall and runoff over large areas.

9

The number of subbasins varies with linksdepending on the size of the catchment areaupstream of the link point. For smaller links,like Bedti-Varada, such separation is notrequired and one subbasin may be used.Observed annual flows at one hydrologicalmeasuring station or many (e.g., in everysubbasin) are calculated using original flowrecords. Observed records for different linksvary in length. For example, a period of 100years (1901–2000) is used for the Alamatti linkwhereas the corresponding period for theSrisailam link is 32 years.

Since the observed flows are normally affectedby various water abstractions, all theseabstractions are calculated and “added back”to the observed flows. It is not entirely clearfrom the feasibility reports how this is donesince types of abstractions differ, they haveincreased over time, especially during the last20 years, and there is no inventory of thevarious abstractions in India (the latter ispartially due to the competitive nature ofinterstate water management, where eachstate tends to leave its abstraction dataundisclosed to its neighbors). Regardless ofthe methods used, procedural attempts takeplace to “naturalize” observed river flows, asthese flows form the reference condition forassessing water availability for the transfer.

Annual time series of weighted areal rainfall foreach gauged basin is then calculated usingthe data from available/selected rainfallstations. A regression relationship betweenannual naturalized flows and annual arealrainfall is established.

This regression analysis is then carried out forthe entire subbasin (which is ungauged) usingmonsoonal rainfall time series as input. Thisallows monsoonal-period flows to be calculatedfor each year. The non-monsoonal portions offlow are then added to the monsoonal portionfor each year thus building the annual timeseries of naturalized flows. It is not clear fromthe feasibility reports how the non-monsoonalportions are calculated, but the perception is

obviously that these flows do not provide asignificant contribution to the overall annualtotal flow volume.

The calculated annual flow time series forindividual subbasins upstream of the donor/receiver site are then summed up to producethe annual time series of naturalized flows atthe link point. This time series is thenpresented in the form of a cumulativedistribution (a type of a flow duration curveanalysis), which shows the probability ofexceedence of every annual flow in a record.This probability is termed “dependability” inIndian practice (an alternative term “assurance”is often used in other countries). This exerciseallows flows occurring at the site to bevisualized and interpreted all at once. The lowerthe flow the greater its “dependability” becauseother flows frequently exceed it. The higher theflow the lesser the dependability: floods aredifficult to capture because they occur lessfrequently.

The cumulative distribution function of annualflows at the donor/receiver site is used toestimate flows (“gross yields” in Indianterminology) with “dependabilities” of 50% and75%. The selection of these assurances ofsupply is rather arbitrary but is not the mostcritical issue, since many different levels ofassurance of water supply larger than 50% areconventionally (and similarly arbitrarily) used inwater resources engineering practicesworldwide (e.g., Smakhtin 2001)

The annual flows at 50 and 75% assurance(further denoted as Q50 and Q75) are themajor components of the water supplyestimates. Other components includeregeneration and known imports from otherriver basins. Regeneration (most likely anequivalent of “return flows”), is estimated as10% of the net utilization from all present andfuture irrigation schemes and as 80% of thedomestic and industrial uses to be met fromsurface water sources. The total water supply(WS) is calculated by summing up theassured flows with regeneration and imports

10

and deducting exports if any:WS

p% = Q

p% + Imports + Regeneration –

Exports (1)

Where, p% denotes the assurance (50 or 75%).All calculations so far are performed at theannual time step. Most of the further decisionsare based on the estimates performed at 75%assurance.

Various demands are then estimated andprojected for either the year 2025 or the year2050, depending on the link. Agricultural waterdemands are estimated based on the stateplans for irrigation development. Industrialrequirement (assumed to be met entirely fromsurface water sources) is not known and istaken to be equal to domestic needs, which isbased on population Figures. Hydropowerrequirement is taken to be equal to totalevaporation from all hydropower projects.Environmental water demands are notaccounted for. When “downstream”requirements are mentioned, they normallyimply the requirements of downstreamagriculture, industry or domestic needs, but notof aquatic ecology or recreation.

The difference between the total availablesupply at 75% assurance (equation (1)) and thetotal projected demand at the same site (donoror receiver) becomes the basis for declaringthe basin (or part thereof) as “surplus” or“deficit.” If the above difference is positive thebasin is “surplus,” and if negative it is “deficit.”

As a rule, each link includes at least onereservoir – either at the donor or at the receiverpoint or at both. The last step in themethodology is therefore a reservoir simulationmodeling with current observed flows in placeand with all future demands included. Thisstep is performed with a monthly time step.Annual flow data for the available period areused as the basis for calculations. All grossannual current upstream water requirementsare subtracted from the gross annual flow timeseries. This gives time series of annual actualinflows to a reservoir whether existing or new

(e.g., to Alamatti, Inchampalli, etc.). These netannual inflows are distributed into monthlyvalues using weights obtained from the actualmonthly flow data at one of the nearby flowstations. The records used to calculate theweights may be short (e.g., 10 years in thecase of the Srisailam). It appears fromfeasibility reports that average monthly weightsare used for this, i.e., monthly flow distributionis assumed to be the same in dry and wetyears. Monthly irrigation requirements are thencalculated based on crop needs. Initial storage(initial condition for reservoir simulation) is oftenassumed to be the dead storage (this istypical for India, where it seems to be acommon practice to assume full drawdown ofthe stored water every year and no provisionfor interannual storage). A reservoir simulationis carried out to identify whether the proposedtransfer can be managed with the proposedstorage and, if yes, then with what level ofreliability - how many successful years out ofall years simulated. A successful year isnormally defined as a year in which 95% of alldemands are met (which is quite aconservative [good] measure of success).

The Issue of Data Resolution and ItsImpact on Planning Estimates

It is clear from the above summary that flow datawith an annual time step resolution were used asthe basis to derive the estimates of dependable(assured) flows at link points. This approachrequires comment. The existing literature onwater resources systems suggests that althoughannual t ime step data may be used forpreliminary (crude) planning of water supplysystems, the preferred data type for this ismonthly flow time series (e.g., McMahon andAdeloye 2005). The issue of data resolution isnot a superf luous one: data resolut ionsignificantly affects the information content ofhydrological time series. Figure 3 illustrates thispoint with the three most widely used flow datatypes – daily, monthly and yearly. The differences

11

between daily and monthly flows in low-flowmonths are negligible due to minor variability ofdaily flows during low-flow months. However, thedifferences between the mean flow for the “year”and mean monthly flows in different months arepronounced: 8 months out of 12 have flowssignificantly lower than the yearly mean. Annualdata resolution therefore does not capture“enough variability” in flows and can lead tooverestimation of available water throughout theyear.

Figure 4 further illustrates the impact of dataresolution on the calculation of “highly dependable”flows. The Figure shows flow duration curves(FDCs) constructed, using annual and monthly flowtime series for the same arbitrarily selected site onthe Krishna River, for which some observed flowdata were available. The flow exceeded in 75% ofall years (75% dependable flow - in Indianterminology) is much higher than the flowexceeded in 75% of all months. NWDA feasibilityreports use annual flow values at 75%dependability as a measure of surface wateravailability at the points of transfer (both donor and

receiver). However, if more, monthly, information-“rich” data are used instead, the flow available at75% dependability becomes an order of magnitudeless than that determined using annual dataresolution.

To obtain an FDC at Vijayawada, which isrepresentative of more natural and less regulatedconditions, the curve at Vijayawada (station 1 inFigure 2), established from the observed record of1900–1965 (which retains more unregulated flows)has been scaled up by the ratio of mean annualflow for the above period to the “official” estimate ofthe mean annual flow at the Krishna outlet of 78BCM (cited also in Smakhtin and Anputhas 2006).

To obtain an FDC at Srisailam, the “naturalized”duration curve at Vijayawada (Station 1 in Figure 2)has been multiplied by the factor of 0.84 – the ratioof catchment area at Srisailam (221,657 km2) to thecatchment area at Vijayawada (251,360 km2). Thedata period used was 1900–1965 (despite theavailability of more recent observations) to avoid theimpacts of observed significant reduction of theKrishna flow in the last 50 years and ensure amore or less “unregulated” record.

FIGURE 3. An illustration of different temporal data resolution: yearly, monthly and daily flows recorded in the Krishna Riverat Agraharam town during March 1990–February 1991.

12

The implications for the assessment of wateravailable for transfer at link points are clearly verysignificant, if such assessment is made by simplyreading off the 50 and/or 75% assured flows from“annual” or “monthly” FDCs. The very limited dataavailable for this study did not allow reliablecalculations to be carried out for all link points.Only a very few data sets, primarily from theInternet, were available. It is not possible toascertain the accuracy of these data, but it is stillpossible to il lustrate the abovementioneddifferences for some links. The link points for whichdependable flows have been calculated are listed inTable 1. They are the only ones which can beeffectively simulated with the limited data available.

To construct an FDC at Inchampalli, theduration curve at Polavaram (both in the GodavariBasin) has been multiplied by the factor of 0.874– the ratio of the catchment area at Inchampalli(269,000 km2) to the catchment area at Polavaram(307,880 km2). Despite the availability of morerecent observations the data period used was1910–1960. This was to avoid many missing data

FIGURE 4. Flow duration curves for the Krishna River at Agraharam town based on 15 years of monthly flow data andconstructed with annual and monthly aggregation levels.

at both ends of the record, particularly after 1960and to ensure that less-impacted, more natural flowtime series was used. This record gives a long-term mean annual flow estimate at Polavaram ofapproximately 105 billion cubic meters (BCM),which value is close to the “official natural” flowestimate of 110 BCM (cited also in Smakhtin andAnputhas 2006).

To obtain an FDC at Alamatti, the durationcurve at Agraharam (station 3 in Figure 2 – thenearest to Alamatti with usable data) has beenmultiplied by a factor of 0.25 – the ratio ofcatchment area at Alamatti (33,375 km2) to thecatchment area at Agraharam (132,920 km2). Thedata period used was 1983–2000 – the onlyperiod for which data at Agraharam were available.Since neither systematic data on waterabstractions upstream of Agraharam nor “natural”flow estimates at Agraharam from alternativesources were available no corrections to theoriginal flow data at Agraharam were possible. Thismay lead to underestimation of means anddependable flows. Observed data at Agraharam

13

are historical data and are affected by upstreamdevelopments. The mean flow volume calculatedat Agraharam from these data is 19,270 MCMwhich is very small compared to the 50% or 75%flows in Table 1 taken from NWDA. It is clear thatsuch mean flow is not accurate and the error istransferred to the estimates of dependable flowsat Alamatti.

Also, flows do not always have a linearrelationship with the basin area. However, the abovesimplifications are unlikely to lead to majorinaccuracies compared to the differences in estimatesfrom annual and monthly time step data, for example.It has to be noted that should more reliable data beavailable the estimates in this study can be revisedto ensure better compatibility with the data used inthe feasibility reports.

Table 1 is presented for illustrative purposes –to show the remarkable differences between thetwo estimates in every case. It is noteworthy that,for example, the official estimate of the “natural” flowat the outlet (Polavaram) is around 110 BCM (acorresponding estimate obtained from the data asdescribed above is 105 BCM, which is rather close).However, the 75% dependable flow at Polavaram isestimated to be 80.17 BCM (80,170 MCM in Table1), which value is around 73% of the total long-termmean flow. While this estimate makes sense in thecontext of the annual time step used, it is virtuallyimpossible to assume that such an enormousamount of water may be a reasonable estimate ofwater available 75% of the time, given the highvariability of flow within a year in the Godavari, witha large number of low-flow months (the case similarto that shown in Figure 3).

TABLE 1. Estimates of surface water availability (MCM) at 50% and 75% dependability from annual (NWDA) and monthly(IWMI) data resolution for selected link points in and out of Krishna.

Donor/Receiver point Dependability 50% Dependability 75%

Annual data Monthly Annual data Monthlyannualized annualized

Krishna - Alamatti 24,041 958 21,405 326

Krishna - Srisailam 66,428 8,626 57,398 1,684

Godavari - Inchampalli 76,185 10,546 66,193 4,497

Godavari - Polavaram 96,549 12,155 80,170 5,132

Krishna - Vijayawada Not available 11,808 Not available 1,964

The Use of Spell Analysis for theReassessment of Surface WaterAvailability

The two different data resolutions (annual andmonthly) used to assess water availabilityeffectively represent two different ways of thinkingabout the level of possible flow regulation. Annualflow data ignore within-year flow variability and,therefore, indirectly suggest that the river may bealmost completely regulated for water supply. Theuse of monthly data (to assess water availability)implies that almost no future increase inabstraction is possible. Both approaches representextreme cases. The “annual” one unjustifiablypushes up water availability estimates while the“monthly” one significantly reduces them. Theseapproaches and their results are entirelyacceptable. They may rather be thought of asrepresenting the top and the bottom limits ofassured water availability at a site.

It is perhaps more appropriate to use someform of water resources storage-yield analysis toestablish maximum possible draft (reservoir yield) atthe donor point of each transfer. This analysis isused to establish either what reservoir yield ispossible, if a given/planned storage is constructed,or what reservoir storage is necessary with therequired yield. In the context of estimating wateravailability (including water availability for transfers),a reservoir (or a system of reservoirs) could besome feasible maximum storage which will be usedto make the water actually “available.” Assessmentof surface water availability then becomes equivalentto the assessment of the yield (draft) of the reservoir

14

FIGURE 5. An extract from the monthly flow time series at the Srisailam site on the Krishna.

0

5,000

10,000

15,000

20,000

25,000

30,000

0 12 24 36 48 60 72 84 96 108 120 132 144 156 168 180 192 204 216 228 240 252 264 276 288 300

Months since July 1901

Mon

thly

flow

vol

ume

(MC

M)

Flow Threshold

explained earlier) suggests that every year, thereis a significant continuous flow deficit below thisthreshold (Figure 5). The deficits range from theminimum of 27,500 to the maximum of 40,100MCM. The latter, maximum deficit may serve asa crude indication of the storage required tomaintain the NWDA estimate of the water yield atthe Srisailam site.

Even given that the above estimate is rathercrude, it is unlikely that without significant storageincrease, water at the above high threshold can bemade available. Also, while this storage is notimpossible to construct in principle as it is onlyapproximately 60% of the long-term mean annualflow at the site, and dams with larger percentagesof that are known, it is hardly practical because:

The cumulative dam storage upstream ofSrisailam at present is already 17.1 BCM.More storage will not only be detrimental tothe upstream basin but also inefficient in analready heavily regulated system.

The dead storage of such a dam (or acombination of dams) in a flat basin like theKrishna is likely to be a large proportion of thetotal storage.

No major additional storage construction isactually planned.

A cumulative storage of 20 BCM (which is

with the above maximum feasible storage. Theapproach still needs to be based on monthly datato capture the seasonal flow variability.

Storage-yield analysis is a discipline of civilengineering and its description is beyond the scopeof this study but it can be found in text books (e.g.,McMahon and Adeloye 2005). In this study, we usethe approach of spells (runs), which may be seenas a component of storage-yield analysis. A spell(run) is a hydrological event when river flowcontinuously stays below (above) a certain thresholdflow. Each spell is characterized by the duration andexcess or deficit flow volume. Deficit flow volume isa characteristic of a low-flow spell. Depending on atype of flow regime and a threshold there may beone low-flow spell or several in one year. Twoexamples of transfer sites from Table 1 are usedbelow to illustrate the alternative assessment ofwater availability: Krishna (Srisailam) and Godavari-Polavaram. Other points were not, or could not be,considered due to lack of some data, unreliabledata or closeness to other points.

In the case of the Srisailam site, the NWDAestimate of the annual yield which will beavailable for the transfer is 57,398 MCM or aconstant flow volume of 4,783 MCM per monththroughout the year. Placed in the context of thespell analysis, it becomes the flow threshold,which needs to be satisfied. Analysis of themonthly flow data at Srisailam (generated as

15

slightly higher than the already existing storageupstream of Srisailam) has been used here as anarbitrary but feasible value in order to estimate howmuch water can be realistically made available. Toachieve this, several runs with different flowthresholds have been carried out until themaximum deficit in the Srisailam time series hasdropped to 20 BCM. The corresponding thresholdflow is 2,700 MCM per month or 32,400 MCM onthe annual scale.

A similar exercise has been carried out usingthe monthly flow time series at Polavaram. Thetotal cumulative storage in the entire GodavariBasin (existing and planned as part of the NRLP)of 18.8 BCM has been increased to 20 BCM toallow for some limited additional, but feasible,storage growth in the future. The correspondingthreshold flow in the Godavari at Polavaram hasbeen estimated as 3,000 MCM per month or36,000 MCM on the annual scale.

Tables 2 and 3 include the above twoalternative estimates of surface water availability,which are stil l significantly lower than thecorresponding NWDA estimates (obtained usingannual time step data). These estimates have beenused with the data on various demands presented

by the NWDA in order to determine the impacts ofreduced surface water availability on the overallbasin water balance. The various demands havenot been revised and are taken in all cases as isfrom the relevant NWDA reports. The environmentalflow requirements have however been estimatedand added to the Tables (these estimates havebeen made using the method developed bySmakhtin and Anputhas [2006] for the leastacceptable environmental management class Dwith minimum possible environmental waterdemand). It has to be noted that this managementclass is, effectively, the “last resort,” the one inwhich there is a large loss of natural habitat, biotaand basic ecosystem functioning. This is asituation that responsible governments would beexpected to avoid.

As the above Tables i l lustrate, aftersignificant reductions in surface water availability,which is the starting point in planning forinterbasin water transfers, the overall waterbalance of each basin has changed dramaticallyfrom being essentially “water surplus” to beingseriously “water deficit.” It is important to notethat this change would occur regardless ofwhether environmental flow requirements are

TABLE 2. Surface water balance (MCM) at the Srisailam Dam site, the Krishna (211,657 km2).

NWDA IWMI

Surface water availability 57,398 32,400

Surface water import (+) -

Surface water export (-) 7,848 7,848

Regeneration (+)

Domestic use 2,624

Industrial use 3,748

Irrigation use 2,773

Subtotal 9,145 9,145 9,145

Overall availability 58,695 33,697

Surface water requirement for (-)

Irrigation use 43,559 43,559

Domestic use 3,278 3,278

Industrial use 4,687 4,687

Hydropower 1,154 1,154

Environmental use n/a 5,300

Subtotal 52,678 (-) 52,678 (-) 57,978

Surface water balance (+) 6,017 (-) 24,281

16

Environmental impacts of Reservoir Construction on the Godavariand Krishna Deltas

included as the component of the demand or not.It is acknowledged that the estimates suggestedhere may not be very accurate due to severe datalimitations in the first place. However, the changecannot be attributed to data inaccuracies orlimitations but, clearly, to the approach used forassessment of surface water availability. It isenvisaged that if the original data used by NWDAwere available, it would result in a similar

TABLE 3. Surface water balance (MCM) at the Polavaram Dam site, the Godavari (307,880 km2).

NWDA IWMI

Surface water availability 80,170 36,000

Surface water import (+) 3,888 3,888

Surface water export (-) 13,318 13,318

Regeneration (+)

Domestic use 1,512

Industrial use 2,402

Irrigation use 3,138

Subtotal 7,052 7,052 7,052

Overall availability 77,792 33,622

Surface water requirement for (-)

Irrigation use 47,541 47,541

Domestic use 1,890 1,890

Industrial use 3,002 3,002

Hydropower (evaporation losses) 6,380 6,380

Consumptive use from Polavaram 3,808 3,808

Environmental use n/a 8,200

Subtotal 62,621 (-) 62,621 (-) 70,821

Surface water balance (+) 15,171 (-) 37,199

change. The points made here attempt to attractattention to the need for increased accuracy inthe overall planning process and to the need torevise the estimates of water availability andwater balance using more advanced planningtools, more transparent processes as well asaccepting environmental water requirements as alegitimate demand.

Interbasin water transfers are associated with theconstruction of new storage reservoirs. A lot hasbeen said and written about submergence andresettlement (upstream) and impacts of changingflow pattern on fish (downstream) – all associatedwith reservoirs. At the same time, all in-streamstorages anywhere in the basin have impacts onriver outlets. Given the number of reservoirs already

constructed in both basins (the Krishna andGodavari) as well as the planned massive storageconstruction associated with NRLP, it is onlynatural to highlight the issues of upstreamdevelopment impacts on deltas and estuaries.These issues have not been considered in theNWDA reports. They also have a general tendencyto be ignored in water resources planning

17

worldwide. At the same time, depending on theriver and the magnitude of upstream constructionsuch impacts may be significant.

Coastal Erosion: The Godavari Delta

Malini and Rao (2004) examined the recentchanges in the Godavari River Delta, called the“rice bowl of Andhra Pradesh,” using remotesensing images. They discovered that the deltahas regressed landward with a total net land lossof 1,836 hectares over the period of 1976–2000 (ata rate of 73.4 ha/year). It was suggested thatreduced inflow of sediments, associated withupstream reservoir construction, is the main causeof reduced vertical accretion at the delta. At thesame time, coastal subsidence, probably promotedby neo-tectonic activity and consequent relativesea-level rise has continued leading to shorelineretreat. Figure 6 illustrates the dynamics of flowand sediment load at the outlet of the Godavari (atPolavaram) and reservoir storage growth in theentire Godavari Basin since the beginning of the1970s. The flow time series has been taken fromInternet sources, the sediment load data have beenread off similar graphs published by Malini and Rao(2004) and the storage data are from the ICOLDdam register. The flow time series has missingdata during 1980–1990 and neither flow norsediment data have been available after 1998.Cumulative dam storage (including large andmedium dams) increased significantly in the early1970s and has remained relatively constant for thelast 30 years. However, it will increase abruptlyagain after the construction of the PolavaramBarrage and the major Inchampalli Dam (thegrowth of the total storage in the basin after thedam construction is shown in Figure 6 for anarbitrarily assumed Inchampalli Dam completiondate of 2010).

While trends in the Godavari River flowcannot be ascertained from the avai labledisrupted flow time series, the decreasing trendin annual sediment loads is clear from thesediment data (Figure 7, also shown by Maliniand Rao [2004]). The mean annual sediment

load has decreased from 100 million tons in 1978(effectively an ending point in noticeable reservoirgrowth in the basin, Figure 6) to 46 million tonsby the end of the 1990s. The current cumulativereservoir storage in the Godavari Basin remainsrelatively low (6.3 BCM, i.e., approximately 6%of the mean annual flow at the outlet). Thestorage growth is not the only one ofsignificance as much water is also diverted frombarrages, i.e., structures without any storage. Arelatively small storage in the basin and a stillnoticeable decreasing trend in sediment loadimply that the basin sediment regime is verysensitive to reservoir growth, if the latter remainsto be seen as the main source of the problem.More sediment inflow reduction may therefore beexpected after the construction of the Polavaramand Inchampalli storages, which will increase theratio of storage to 19% of the natural flow in thebasin.

Coastal Erosion: The Krishna Delta

In this study, an attempt has been made toexamine whether similar trends exist in theKrishna Basin, concerning the proportion ofstorage: annual flow is much larger than in theGodavari. The observations on sediment loads atthe outlet of the Krishna at Vijayawada over thelast 30–40 years have however not been providedby the Central Water Commission (CWC) duringthe course of the study. The only available datawere for the period 1991–2000 (CWC 2006),which is rather short for any meaningfulconclusions on trends. The comparison of thetwo short time series of sediment loads, atAgraharam (upstream of major reservoirs, Figure2) and at Vijayawada (downstream of all majordams), has revealed a significant decrease insediments downstream of the reservoir system(Figure 8). The differences are particularlynoticeable in high-flow years (1994, 1999), whenmore sediment has reached Agraharam from therelatively unregulated upstream basin but allsediments were likely being trapped by theexisting reservoir system (Srisailam, Nagarjuna

18

FIGURE 6. Time series of annual flows, sediment loads and cumulative storage in the Godavari Basin outlet at Polavaram.

0

50

100

150

200

250

1971 1976 1981 1986 1991 1996 2001 2006

Sed

imen

t lo

ad (m

ill to

ns)

an

d fl

ow

(BC

M)

0

4

8

12

16

20

Cu

mu

lati

ve d

am s

tora

ge

(BC

M)

Sediment load Flow volume Cumulative dam storage

FIGURE 7. Time series of sediment load at Polavaram with a decreasing trend line.

0

50

100

150

200

250

1970 1975 1980 1985 1990 1995

Sedi

men

t loa

d (m

illio

n to

ns)

19

Sagar) upstream of Vijayawada. The absence ofsediment data prior to 1991 does not allow forfurther conclusions about sediment regimechanges to be made. However, these changesare most likely very significant due to the markedreduction of river flow at the Krishna outlet(Figure 9) over the last 70 years. This reductionis due to various water diversions, groundwaterdevelopment and increased cumulative reservoirstorage in the basin, which has grown fromalmost zero in 1960 to 28.5 BCM at present.This present cumulative storage represents 36%and 132% of the natural and present-day Krishnamean annual flow, respectively.

To examine the potential impacts of reducedsediment inflow on the Krishna Delta, severalremote sensing images of the area were analyzed.The images were obtained from Earth Science DataInterface (ESDI) at the Global Land Cover Facility(GLFC) on http://glcfapp.umiacs.umd.edu:8080/esdi/index.jsp and were selected from theperiod of 1977 to 2000 to form a “time series.” Theimages included:

Landsat 2 Multispectral Scanner (MSS) imagedated 1 June 1977 with a spatial resolution of57 meters (m).

Landsat 5 Thematic Mapper (TM) image dated10 November 1990 with a spatial resolution of28.5 m.

Landsat 7 Enhanced Thematic Mapper plus(ETM+) image dated 28 October 2000 with aspatial resolution of 28.5 m.

Three basic layers were used to detectmorphological changes in the delta: band 4 (nearinfrared [NIR]), band 2 (red) and band 1 (blue).These layers have characteristics that are suitablefor coastal mapping, differentiation of vegetationfrom soil, reflectivity of vegetation vigor anddelineation of water bodies. The first, “oldest” imagewas assumed to be the reference condition againstwhich changes in the other two images weredetected. The entire delta shoreline was examinedto demarcate the zones of erosion and depositionusing ERDAS 9.0 software. The areas of depositionand erosion between two consecutive dates (i.e., in1990 and 2000) were identified and calculatedusing ArcGIS software. The areas around selectedpoints (primarily the mouths of the maindistributaries), where significant changes wereexpected to occur were closely examined,highlighting the zones of erosion and deposition ateach. The image of the Krishna Delta showingselected areas where detailed assessment oferosion and deposition has been made ispresented in Figure 10. Figures 11 and 12 displaythe sequence of images for years 1977, 1990 and2000 for some of the selected areas circled inFigure 10. The black lines in each image representthe reference position of the land mass at the startof the period, in 1977. Figure 13 shows areas ofpredominant erosion and deposition during theperiod between 1977 and 2000 for the entire deltashoreline, while table 4 summarizes the calculatedcharacteristics of these processes for the entiredelta over the same period.

TABLE 4. Areal extent of erosion and deposition in the Krishna Delta over 23 years (1977–2000).

Point no. Erosion Deposition Net loss Rate of loss/(ha) (ha) (ha) gain (ha/yr)

1 598 483 115 5.0

2 478 178 300 13.0

3 275 31 244 10.6

4 326 74 252 11.0

5 79 98 -19 -0.8

6 894 3 891 38.7

Total (23 years) 2,650 867 1,783 77.5

20

FIGURE 8. The time series of sediment loads in the Krishna at Agraharam and Vijayawada.

FIGURE 9. Time series of annual flows, sediment loads and cumulative storage in the Krishna Basin outlet at Vijayawada.

0

20

40

60

80

100

120

140

160

180

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Sed

imen

t lo

ad (m

illio

n t

on

s)

Sediment load at Agraharam Sediment load at Vijayawada

0

10

20

30

40

50

60

70

80

90

100

1930 1940 1950 1960 1970 1980 1990 2000 2010

Sed

imen

t lo

ad (m

illio

n t

on

s) a

nd

flo

w (B

CM

)

0

4

8

12

16

20

24

28

32

Cu

mu

lati

ve d

am s

tora

ge

(BC

M)

Sediment load Flow volume Cumulative dam storage

21

FIGURE 10. The image of the Krishna River Delta indicating the areas where a closer inspection of erosion anddeposition was made.

FIGURE 11. The changing morphology of the selected area 2 in 1977, 1990 and 2000. The top and bottom rows of imagesshow the dynamics of the right and left banks of the distributary, respectively.

1

2 3

4

5

6

22

FIGURE 13. A contour of the Krishna Delta showing areas of erosion and deposition during the period between1977 and 2000.

FIGURE 12. The changing morphology of the selected area 4 in 1977, 1990 and 2000. The top and bottom rows of imagesshow the dynamics of the southern and northern parts of the area, respectively.

23

to suggest that upstream basin storagedevelopment leads to the said retreat of deltas.The Krishna River is already effectively a “closedbasin” as only occasional high flows “spill” intothe delta with almost zero sediment contributionto it (Figure 8). Therefore, the storage that isalready constructed in the Krishna will have along-lasting detrimental effect on the delta and itsagricultural productivity (the situation in theGodavari Delta will also most likely deteriorateafter the construction of the additional storagesplanned as part of the NRLP).

Detailed sedimentation modeling studies wouldbe useful in all major deltas of India in order todevelop a better understanding and quantification ofthe links between upstream water and sedimentflow reduction, upstream storage growth and man-induced changes in deltas, on the one hand, withdeltas’ erosion and retreat, on the other. Suchstudies could allow the specification of necessaryenvironmental flow releases to be made for themaintenance of delta sediment regimes.

Coastal erosion may be seen as a slowprocess. However, there are a few aspects whichpromote negative environmental impacts associatedwith it. One is the saltwater intrusion. Bobba (2002)conducted a numerical modeling study of theGodavari Delta and showed that saline intrusionmay become a major factor of reduced agriculturalproductivity in that delta due to increasedgroundwater pumping and reduced freshwater inflow(the authors could not identify a similar publishedstudy for the Krishna Delta). Coastal erosion,caused by similar factors facilitates saltwaterintrusion deeper in the delta adversely affecting theproductivity of land. Additionally, although highlyuncertain in quantitative terms, there is thepotential sea-level rise in the next 50 years due toclimate change, although the limited availableobservations have not detected it so far. This risecan lead to even more coastal erosion and deepersaltwater penetration, accelerating deltadegradation. This research was not the scope ofthe current study and needs to be carried out asa separate and detailed project. While quantificationof the above impacts will be developed, evenlimited environmental flow releases from existingreservoirs in the Krishna and Godavari will delay the

The results suggest that while areas ofpredominant erosion and deposition interchange,the overall tendency is towards the regressionlandward with losses of land to the sea, thesituation similar to that in the Godavari Delta. Theannual net loss rate of 77.4 hectares is almostthe same as that in the Godavari Delta (73.4ha/year; Malini and Rao 2004). One noticeablefeature of the Krishna Delta is also its higher ratioof erosion to deposition (3.05 versus 1.6 in theGodavari) over the same period, which suggeststhat coastal erosion is more “effective” in theKrishna Delta than in the Godavari, despite theslightly smaller area (4,700 km2 versus5,100 km2) and shorter shoreline of the former(134 km versus 160 km). Erosion is also adominant process through most of the coastline,while deposition is limited to certain sections only(Figure 13).

Possible Causes and Implications ofCoastal Erosion

The regression of both deltas cannot be explainedby the sea-level rise. Analysis of the available sea-level data in the region for the period 1970–1996(measurements at Visakhapatnam and Chennai)and for the period 1990–2001 (calculations fromdaily tide gauge data at the Kakinada to the northof the Godavari Delta) did not reveal any significantrising or falling trends (Malini and Rao 2004).Therefore, coastal erosion in the Krishna andGodavari deltas can only be explained by theabove-illustrated reduced sediment supply that, inturn, is due to upstream flow regulation. In addition,human activities in delta regions (e.g., conversionof cropland and mangrove swamp areas intoaquaculture ponds) may also be responsible forsea transgression leading to coastal erosion andshoreline retreat of the deltas (e.g., Sarma et al.2001).

Analysis of the longer sediment load dataseries for the downstream parts of the Krishnaand the use of more recent and more resoluteremote sensing images would result in moredetailed quantification of delta erosion. However,even with the existing limited data, it is possible

24

All NRLP transfers are justified based on thepremise that a “natural” annual flow volumewhich has exceeded 75% of the time (e.g., 30out of 40 years) is available for water utilization.This does not consider the variability within ayear, which is extremely high in monsoon-drivenIndian rivers. As a result, more water isperceived to be originally available at a site oftransfer. Alternative techniques, based on a low-flow spell analysis and, more importantly, astorage-yield analysis, may be used toreevaluate the surface water availability atproposed transfer sites.

All NRLP transfers are further justified based onthe maximum plans for irrigation (for 2025 or2050) adopted by each state within each riverbasin. This boosts irrigation requirements andserves as the driver for future water resourcesdevelopment. Maximum irrigation developmentis therefore effectively programmed into India’sWater Future for the next half a centurywithout alternatives or much discussion of itstechnical and economic feasibility.

A few points on the Krishna (e.g., Alamatti,Srisailam) are classified as “surplus” and are tobecome “donors.” At the same time, some links(e.g., Bedti-Varada) are expected to bring waterinto the Krishna, upstream of the “surpluspoints.” Some “deficit” points in the LowerKrishna then rely on transfers from theMahanadi through the Godavari, rather than onmore naturally available water from the UpperKrishna. It does not appear entirely logical toisolate subbasins and describe them as“surplus,” since they contribute differentially todownstream water availability. There may be a

adverse environmental processes in both deltas.New storage reservoirs need to be planned so asto allow sediments to reach deltas. Construction of

the most downstream reservoirs, particularly aslarge as Inchampalli, will definitely not serve thispurpose.

Conclusions

need for more integrated water resourcesplanning whereby all future water transfers inand out of the same basin are considered andsimulated together.

The demands which are currently consideredin feasibil i ty reports include irrigation,hydropower, industry and domestic use. It issuggested that at least an environmentaldemand for environmental management classD is also explicitly included at the planningstage – even as a contingency item. Thisclass is the least acceptable from anecological point of view and requires a verylimited environmental water allocation, in therange of 10–15% of the long-term annual flow.This would be a precautionary measure in theabsence of other, more detailed, informationat present. It is envisaged that even such aminimal allocation will make some transferplans less feasible, as was illustrated in thisreport. The main point however is thatenvironmental water demand should beexplicitly considered in water resourcesplanning, similar to the demands ofagriculture, industry, hydropower anddomestic needs.

In this report, for the donor and receiver pointson the Polavaram-Vijayawada link, theenvironmental flow requirements have beencalculated using the planning technique ofSmakhtin and Anputhas (2006). Thesedemands, as scenarios for two environmentalmanagement classes, have been used indetailed water resources modeling of this link.The results of this modeling are described ina companion report (Bharati et al. n.d.).

25

Locating reservoir sites (particularly as large asthe planned Inchampalli Dam) in the mostdownstream, normally flat, areas of river basinsis problematic from an engineering perspective.Such reservoirs have large surface water areas,which drastically increase evaporation and incura large dead volume, which reduces the activestorage and makes it inefficient. It also capturesmost of the sediment supply to downstreamdeltas, which are the “rice bowls” of India, dueto the high land productivity. It has beendemonstrated that the Godavari and Krishnadeltas have been in retreat over the last 25years, which is related, most likely, to reducedflow and sediment flow to deltas. Environmentalflows need to be provided to at least partiallyarrest/delay this “shrinking of deltas” whichthreatens agricultural production and mangroveecosystems, despite the fact that this shrinkingis slow.

It is not possible to properly reevaluate anyplans without having the same startingconditions, i.e., the same hydrological data.Consequently, only cautious statements can

be made at present regarding the quantitativeside of planned water transfers. However, norelevant and detailed hydrological data havebeen made available to this project despite allcontinuous efforts to obtain them. This leadsto two more points. First, if these data areavailable (the actual NWDA flow time seriesfor each donor/receiver point considered), it ispossible to revise the estimates presented inthis report. Second, the continued policy ofhydrological “data secrecy” is not conduciveto good water resources planning anddevelopment in India and will not lead tosocially and environmentally acceptable waterprojects. In fact, it is one of the majorstumbling blocks on the way to scientific andengineering progress in water science in thecountry. India needs a centralized datastorage and dissemination system. Such asystem could be developed within a timeframe of 2–3 years. However, policies of freedata access could and should be reinforcedbefore that. Without such reinforcementdifficulties in resolving water controversies inIndia will remain.

26

References

Agricultural Finance Corporation Ltd. 2005. Indirasagar (Polavaram): A multipurpose major irrigation project.Vol. 1: Environmental Impact Assessment (EIA) and Environmental Management Plan (EMP). Hyderabad,India, 417 pp.

Bharati, L.; Anand, B. K.; Smakhtin, V. n.d. Analysis of the interbasin water transfer scheme in India: A case studyof the Godavari-Krishna link. Journal of Irrigation and Drainage (Under review).

Bobba, A. G. 2002. Numerical modeling of salt-water intrusion due to human activities and sea-level changein Godavari Delta, India. Hydrological Science Journal 47(S): S67-S80.

CWC (Central Water Commission). 2006. Integrated hydrological data book (non-classified river basins). NewDelhi, India, 383 pp.

Dams, Rivers and People. 2004. Vol. 1-2. South Asia Network on Dams, Rivers and People (SANDRP).http://www.narmada.org/sandrp

Interbasin Water Transfer. 1999. Proceedings of the international workshop. UNESCO, Paris, 25-27 April 1999.IHP-V Technical Documents in Hydrology. Document No. 28. Paris: UNESCO, 229 pp.

Malini, B. H.; Rao, K. N. 2004. Coastal erosion and habitat loss along the Godavari delta front – a fallout ofdam construction(?). Current Science 87(9): 1232-1236.

McMahon, T. A.; Adeloye, A. J. 2005. Water resources yield. Colorado, USA: Water Resources Publications, LLC,220 pp.

Sarma, V. V. L. N.; Murali Krishna, G.; Hema Malini, B.; Nageswara Rao, K. 2001. Landuse/land cover changedetection through remote sensing and its climatic implications in the Godavari delta region. Journal of IndianSociety of Remote Sensing 29: 85-91.

Smakhtin, V. U. 2001. Low-flow hydrology: A review. Journal of Hydrology 240(3/4): 147-186.

Smakhtin, V. U.; Anputhas, M. 2006. An assessment of environmental flow requirements of Indian river basins.IWMI Research Report 107. Colombo, Sri Lanka: International Water Management Institute, 36 pp.

Research Reports

109. Costs and Performance of Irrigation Projects: A Comparison of Sub-Saharan Africaand Other Developing Regions. Arlene Inocencio, Masao Kikuchi, Manabu Tonosaki,Atsushi Maruyama, Douglas Merrey, Hilmy Sally and Ijsbrand de Jong. 2007.

110. From Integrated to Expedient: An Adaptive Framework for River BasinManagement in Developing Countries. Bruce A. Lankford, Douglas J. Merrey,Julien Cour and Nick Hepworth. 2007.

111. Closing of the Krishna Basin: Irrigation, Streamflow Depletion and MacroscaleHydrology. Trent W. Biggs, Anju Gaur, Christopher A. Scott, Prasad Thenkabail,Parthasaradhi Gangadhara Rao, Murali Krishna Gumma, Sreedhar Acharya andHugh Turral. 2007.

112. The Impact of Government Policies on Land Use in Northern Vietnam: An InstitutionalApproach for Understanding Farmer Decisions. Floriane Clément, Jaime M.Amezaga, Didier Orange and Tran Duc Toan. 2007.

113. Applying the Gini Coefficient to Measure Inequality of Water Use in the OlifantsRiver Water Management Area, South Africa. James Cullis and Barbara vanKoppen. 2007.

114. Developing Procedures for Assessment of Ecological Status of Indian RiverBasins in the Context of Environmental Water Requirements. Vladimir Smakhtin,Muthukumarasamy Arunachalam, Sandeep Behera, Archana Chatterjee, SrabaniDas, Parikshit Gautam, Gaurav D. Joshi, Kumbakonam G. Sivaramakrishnanand K. Sankaran Unni. 2007.

115. Rural-Urban Food, Nutrient and Virtual Water Flows in Selected West AfricanCities. Pay Drechsel, Sophie Graefe and Michael Fink. 2007.

116. Agricultural Water Management in a Water Stressed Catchment: Lessons fromthe RIPARWIN Project. Matthew P. McCartney, Bruce A. Lankford and HenryMahoo. 2007.

117. Treadle Pump Irrigation and Poverty in Ghana. Adetola Adeoti, Boubacar Barry,Regassa Namara, Abdul Kamara and Atsu Titiati. 2007.

118. Evaluation of Historic, Current and Future Water Demand in the Olifants RiverCatchment, South Africa. Matthew McCartney and Roberto Arranz. 2007.

119. Changing Consumption Patterns: Implications on Food and Water Demand inIndia. Upali A. Amarasinghe, Tushaar Shah and Om Prakash Singh. 2007.

120. Hydrological and Environmental Issues of Interbasin Water Transfers in India:A Case of the Krishna River Basin. Vladimir Smakhtin, Nilantha Gamage andLuna Bharati. 2007.