Final Paper - Executive Summary
November 2000

1. The World Commission on Dams Global Case Study Program

This case study is one of a number being carried out world-wide with a common methodology and approach to inform the World Commission on Dams about the development effectiveness of dams on a range of issues associated with the planning, design, construction, operation and decommissioning of large dams. This particular study deals with the Tarbela Dam Project (TDP) developed as a follow up to settlement of the river waters dispute between India and Pakistan. It formed part of a wider set of infrastructure projects to further develop the existing Indus Basin Irrigation System (IBIS) and facilitate transfer of stored water from the western rivers to replace water lost from the eastern rivers. The primary function of the dam was irrigation, with power as a secondary objective.

2. Context and Scope

2.1 Indus River Basin

The Indus River basin stretches from the Himalayan mountains in the north, to the dry alluvial plains of Sindh in the south. The alluvial plains of the Indus basin cover an area of 207 200 km², approximately 25% of the land area of Pakistan. The catchment area upstream of Tarbela dam is predominately a barren and glaciated landscape and the main source of inflow to the Indus is snowmelt and glacial flow. The climate in the Indus plain is arid to semi-arid with significant variability both throughout the year and from upstream to downstream. Mean minimum temperatures in the upper plain are 2°C in the winter with mean maximum temperatures in the summer reaching 49°C. Mean annual rainfall is low, ranging from 9cm in the lower plain to 51cm upstream in Lahore. Evaporation is high with mean annual values in the range 165 to 204cm. Agriculture is central to the culture and economy of Pakistan although urbanisation is rapidly increasing. At the time of planning the TDP in the mid-1960s, Pakistan's population was 51 million, which by 1999, has increased to 134.5 million. Although the percentage of rural population to total population has reduced from 82% to 67% during this period, the absolute number of rural population has increased from 41.9 to 90.1 million.

The IBIS is the world's largest contiguous irrigation system and has been developed over the past 140 years. Prior to the implementation of Indus Basin Project (IBP), the IBIS already comprised a major network of river barrages and irrigation canals as a sequel to the water treaty between India and Pakistan. In August 1947, British India was divided into two independent states - India and Pakistan. Shortly afterwards on 1 April 1948, India unilaterally closed the canals with headworks in its territory on the eastern rivers (Ravi and Sutlej) depriving Pakistan of irrigation supplies for about 0.7 million hectares (mha) of the most fertile land in Punjab - part of the IBIS. Pre-independence flows of the eastern rivers in the dry (rabi) season amounted to an average of 3.84 billion cubic metres (bcm), approximately 12% of the total dry season flows of eastern and western rivers combined. India agreed to re-open these canals only after the Inter-Dominion Agreement of May 1948, asserting its right over the entire water of the eastern rivers that would affect an irrigation system commanding 2.9 mha (21% of the total canal command of about 14 mha in the Indus Basin). This issue bitterly strained relations between the two countries.

The World Bank (WB) took the initiative to facilitate negotiations between India and Pakistan that continued for 10 years and culminated in signing of the Indus Waters Treaty (IWT) in 1960. The Indus Basin Project (IBP) was the mechanism to implement provisions of IWT and comprised: three storage reservoirs (Tarbela, Mangla and Chashma); six barrages including a siphon; eight new inter-river link canals and remodelling of three existing link canals. The wording of the Treaty referred to 'a dam on the Indus'. After an evaluation of alternative sites, Tarbela was selected. An Indus Basin Development Fund (IBDF) of $895 million was created to finance the IBP with contributions from the WB and other donors to which India made a fixed contribution of £62 million ($173.8 million).

The most positive outcome of the Treaty from Pakistan's perspective was the attainment of water security through construction of the works that made its irrigation system and hydropower generation facilities fully independent of India. The Treaty was, however, opposed by those who maintained that IBP would only include (ever-depleting) large storage facilities, replacing water that Pakistan would be losing, by surrendering to India the perennial flow waters of the three eastern rivers. India, on the other hand, would get plenty of additional irrigation water for new development projects. They considered the development of IBP to be at the cost of development of other important sectors, especially education, health, and communication infrastructure. It was apprehended that IBP would consume most of the scarce financial resources and leave little for economic development. Finally, President Ayub Khan accepted IWT as a pragmatic solution to a very difficult problem^[1].

The Tarbela Dam Project (TDP) was constructed as part of IBP together with Mangla Dam and associated infrastructure. Besides the basic purpose of replacement flows, the TDP was also planned to provide a substantial degree of integration/regulation of IBIS. Its main predicted objectives were:

• to provide an 11.5 billion cubic metres (bcm) storage dam on River Indus (almost doubling Indus flows in the dry season - rabi) in order to partly replace the water of eastern rivers ceded to India, and provide additional supplies during the low flow period to facilitate further development of irrigated agriculture;
• to increase food production to achieve self sufficiency, especially in wheat; and
• to generate cheap hydropower through staged development of 2100 MW capacity.

Two documents were central to the planning of the project and its evaluation and provide the basis for comparison of predicted and actual outcomes in the study. The feasibility study of the project and proposals for wider development plans in the basin were contained in the Indus Special Study completed in 1967- referred to as the Lieftinck Report. The specific elements to be financed under the TDP, including the dam, associated structures and civil works for the first four power units are described in the World Bank's Staff Appraisal Report (SAR) of 1968.

2.2 Financing and Costs

The project was financed through Tarbela Development Fund (TDF) created in 1968 out of the remaining balance from the IBDF and additional loans and grants from friendly countries. The calculation of cost overrun is dependant on the way in which annual payments are inflated (according to local or foreign inflation rates) and which document is taken as the base cost estimate. Two approaches are used to define the possible range of cost overrun. In the first approach, the total project cost including all 12 projected power units is taken from the Lieftinck Report and inflated to 1998 prices. The original estimated capital cost, including the proposed generating capacity of 2,100 MW, was 1136.4 million ($5 875 million 1998 prices) with a foreign exchange component in the order of 60%. The actual annual disbursements for all civil and power works (3 478 MW), including resettlement and debt servicing, were converted into dollars and inflated to 1998 dollar prices. This resulted in an estimate of actual costs in 1998 terms of $8 800 million, or an increase of about 50% over the estimated capital cost.

The second method makes a comparison of financial flows (in prevailing prices) related to the original cost in the SAR and supplementary loans. The final cost according to the Project Completion Report (PCR), excluding power units (WB 1986), was $1 497 million including a foreign exchange component of $800 million. It covered the cost reimbursed from TDF and related to: dam and associated civil works; power station civil works for four units only; design improvements; and special repairs/restoration. It did not, however, cover the cost of resettlement, additional Tunnel No. 5 and power installations (units 1-14) that were borne by the Government of Pakistan (GOP) and the Water and Power Development Authority (WAPDA), respectively. The original estimate for the same scope of work (excluding power etc) was $828 million. The cost overrun in purely nominal dollar terms was therefore 81%. Further work is required to determine more precisely the project cost over-run which is estimated to be in the range 50-81% from the two methods used.

The WB as administrator of TDF, arranged two supplements to fund the revised project cost. The increase in cost was mainly because of an initial under-estimation; design improvements; and major structural repairs/restoration of some project components damaged after impounding.

2.3 Implementation and Commissioning

Tarbela was the largest earth-fill dam of the day and presented a challenge in dam design and construction, particularly given complex geological conditions. During the initial test filling in 1974, one of the three intake gates of a diversion tunnel jammed while being lowered to achieve closure. This resulted in collapse of the intake structure of Tunnel No.2 thus endangering the safety of the dam. The reservoir was evacuated to avert this crisis. After quick repairs, however, the problem was resolved and the dam was fully commissioned in 1976.

Commencement of the project in May 1968 was almost one year later than planned. The original estimate for construction and commissioning of the dam and reservoir was 7.25 years. As a result of special repairs and restoration, the project implementation period was approximately one year longer at 8.3 years. The originalschedule for full commissioning was September 1974. Against this, partial storage releases started in winter 1975, while first complete reservoir filling was accomplished in September 1976, two years behind schedule. The first four power units started generating in early 1977, on average twenty months later than scheduled. Overall electricity generation potential of 3478 MW (67% more than the planned 2100 MW) was developed in stages through 1993.

2.4 Operation and Maintenance

Detailed O&M procedures for the project were developed before its scheduled commissioning in 1975. These procedures required implementation under intense supervision backed by adequate logistic and funding support. This, unfortunately, did not occur, as, during initial filling, serious problems occurred, including damage to Tunnel No. 2, the appearance of sinkholes in the upstream blanket, damage to the stilling basin floor, and the erosion of spillway plunge pools. This resulted in most of the funds being diverted for special repair/remedial works. This funding constraint negatively affected O&M but was somewhat eased later when $3 540 million was provided by the WB. In the meantime, the essential O&M facilities, including equipment due from the main contractor, also became available.

Subsequently, the project is considered to have been well operated and maintained. Extensive monitoring of the engineering, technical and safety aspects of the project has also been carried out. Furthermore, standing arrangements have been made for constant monitoring of the project features. This has significantly contributed to the state of the art with respect to: design; construction, restoration/repair of seriously damaged infrastructure; and efficient operation and safety monitoring of large dam projects.

Operating rules of the reservoir were originally set for irrigation releases but in the 1990s were amended in response to the need to manage sediment deposition. This is described under Section 3.4 'Predicted and Actual Effects'.

Because of the lack of a formal agreement between the provinces on sharing Indus waters, the operation of the reservoir was conducted on an ad hoc basis during 1976 to 1993. A formal accord was reached between the provinces in 1991 that prescribed water allocation between provinces, and the Indus River System Authority (IRSA) was created in 1993 under an Act of the Parliament. This body with representatives from all provinces and federal government is chaired by one of the representatives on annual rotational basis. Regulation of the Indus River system including storage reservoirs at Tarbela and Mangla is now being handled by IRSA.

3. Predicted and Actual Effects

3.1 Water Resources

The primary goal of the TDP was to ensure that the irrigated areas in the country continued to receive water supplies after India diverted the eastern rivers. Storage was also provided to bring additional areas under cultivation through increased cropping intensity in various existing canal commands. Approximately two thirds of the reservoir storage was considered to be for replacement and one third for additional irrigation. Given the integral nature of Tarbela within IBIS and the influence of other investments and inputs on agricultural production, a comparison of actual releases of water with those predicted provides a simple indicator of how the project performed in terms of its original objectives.

The prediction of storage releases from the reservoir for irrigation in the dry (rabi) season assumed a declining trend from 10.17 bcm in 1975 to 6.95 bcm in 1998 based on the predicted effect of sedimentation in the reservoir. Overall, the actual releases were on average 20% more than predicted (19% higher during 1975-90 and 22% higher during 1990-98) although in 1990-92 releases were 8% lower than predicted.

An analysis of flows along the Indus River over time and diversion to the canal systems, particularly in the dry season, provides an indication of the distribution of additional waters supplied from Tarbela. Three barrages were investigated: Taunsa barrage which lies upstream of the confluence of the Indus and other major tributary rivers; Guddu barrage, which is in Sindh close to the border with Punjab; and Kotri, the last barrage on the Indus before the river enters the delta area. Releases from Tarbela contributed to increases in mean dry season flows at Taunsa (30%) and Guddu (26%) measured over the period 1975-95. Reductions in mean wet season flows of up to 6% were attributed to the filling cycle of the reservoir.

The flow distribution at Kotri barrage provides valuable information about flow conditions in the downstream delta areas. A low flow analysis was carried out covering the period from 1940, before construction of Kotri barrage (1955) and commencement of major irrigation diversions (1960). Rabi flows downstream of the barrage averaged 13.5 bcm from 1940-61 and reduced markedly to an average of 3.6 bcm following diversions of water from Kotri barrage. They reduced further to 1.4 bcm in the period post-Mangla (1967-75). Average rabi season flows increased to 3.1 bcm in the post-Tarbela period (1975-98). Within these years there was significant variability. From 1980-85, for example, the seasonal mean was 1.2 bcm whereas from 1990-95 it was 5.9 bcm.

The extent of change in low flow conditions is also illustrated by a comparison of occurrences where there were zero flows downstream of Kotri during rabi season. In the pre-Kotri period, there were no years with zero flow. Following construction of Kotri barrage and the diversion of irrigation water, the number of years with zero flows was 80% during1962-67, 100% during post-Mangla (1967-75) and 96% during post Tarbela (1975-98) periods. The significant change in downstream flows was therefore initiated after commissioning of Kotri barrage, 16 years prior to construction of Tarbela. Additional rabi flows from Tarbela had only a marginal effect on increasing these low flows, but they were still well below the pre-Kotri levels.

In the Lieftinck Report, the estimation of the economic benefits of water released from the dam was based on ascribed value related to the value-added of irrigation water to agricultural production taking into consideration the losses in the Tarbela canal commands. Using the same value per unit of water applied to actual storage releases, a comparison of predicted and 'actual' economic benefits per unit of water was made. The predicted storage benefits over the period 1975-98 were estimated as $297 million at 1965 prices ($1 534 million in 1998 prices). The computed corresponding economic benefits in 1965 prices are $353 million ($1 825 million in 1998 prices) - an increase of 19%. This analysis identifies the potential value of the water released.

3.2 Sedimentation

The Indus is one of the largest sediment producing rivers in the world. The main source of sediment is from the glacial landscape and erosion from steep sided barren slopes. The predicted rate of sediment inflow was 0.294 bcm per year meaning that the dam would silt up to 90% capacity in 50 years and thereafter continue to provide only about 1.2 bcm of live storage. A number of sediment management measures were examined at the time but considered not to be feasible.

In practice, the actual sediment inflow rate has been significantly lower than predicted, with an average rate of 0.106 bcm, ie 36% of the predicted inflow rate. However the proportion of sediment inflow trapped in the reservoir (the trap efficiency) was slightly higher than predicted. The useful life of the dam is now considered to be 85 years, although, as with the prediction, the usable storage will gradually decline over this period. An unexpected aspect of the sediment deposition, however, is the advancement of the sediment delta, which is now located 14km from the dam. There are concerns that under earthquake loading, the sediment may liquefy and block all low-level outlets, including power intakes.

Measures are being investigated to reduce the risk of liquefaction damage and also to prolong the life of the reservoir. These include physical measures, such as provision of an underwater protection to the low level outlets including power intakes and sluicing tunnels to remove sediment, and management measures to reduce the proportion of sediment deposited and its location. Reduction in sediment load entering the reservoir is not possible due to the altitude and nature of the catchment. In terms of management measures, the operating rule of the reservoir has been changed to raise the minimum drawdown level from 396m to 417m and thereafter raise it gradually every year. This would have the effect of depositing sediment in the upper reaches and would reduce the advance of the sediment delta, but at the cost of reducing live storage with the trade off of reducing water availability in the dry season.

3.3 Surface Water Irrigation

The following analysis of irrigated areas, cropping intensity, yields and agricultural production provides an insight into the actual experience, although Tarbela is only one factor among a range of factors that influenced the change (e.g. agricultural inputs, farmer response, impacts of waterlogging, and other investment projects).

The Lieftinck Report, projected that annual average surface water diversions at the canal heads would be around 125.5 bcm by the year 1985. In actual practice, the IBIS attained a 3% higher average figure of 129.4 bcm over the period 1978-82. This figure closely matched with planned abstractions for IBIS, based on the sanctioned allocations of individual projects in the provinces. Tarbela contributed on average about 9.3% (12 bcm) to the total annual irrigation supplies for IBIS. The contribution has been 21% of rabi abstractions (9.8 bcm) and for 3% of wet season (kharif) abstractions (2.5 bcm). It is therefore concluded that Tarbela, along with Mangla dam, achieved the goal to replace water lost to India and increased diversions for irrigation.

The Lieftinck Report projected average canal head diversions by 2000 to reach 149 bcm backed by completion of: three additional storage reservoirs almost aggregating to Tarbela capacity; a number of canal enlargement projects to withdraw additional surplus river flows; and ground-water developments with attendant drainage. However, none of the surface water projects was taken up in the post-Tarbela period. Thus the canal head diversions attained immediately after Tarbela remained almost constant.

As a comparison of irrigation diversions downstream of Tarbela in the pre- and post-Tarbela periods, the dry season (rabi) diversions increased by: 98% at Taunsa (from 1.12 to 2.22 bcm); by 168% at Guddu (0.99 to 2.65 bcm;) and by 80% (2.15 to 3.87 bcm) at Kotri.

3.4 Groundwater Irrigation

Approximately 64% recharge to the groundwater in the Indus Basin occurs in areas of usable water quality either directly or by mixing with canal water. The canal conveyance losses directly associated with additional Tarbela storage are estimated to contribute about 10% of the overall recharge of groundwater (the issue of waterlogging and salinity is covered in a separate section below). In the last 21 years (1972-97), the contribution of groundwater to irrigated agriculture nearly doubled from 31.6 bcm to 62.2 bcm. In 1997-98 it declined to 49.6 bcm, equivalent to 38% of surface water diversions. The expansion in groundwater irrigation was due in part to initial price incentives in the form of cheap electricity and later on to the availability of relatively low cost diesel technology. Groundwater irrigation provided a level of reliability that was deteriorating in the surface irrigation system. The number of irrigation tubewells increased by almost 500% from 1970 to 1996, with a total number of 484 000 installations.

3.5 Agriculture

The projections related to agricultural production in the Lieftinck report were based on the implementation of the TDP, and other infrastructure projects, some of which have been completed, while others have not. However other projects, identified subsequently by the Lieftinck Report (such as the On Farm Water Management projects and the Chashma Right Bank Canal), have been implemented and groundwater utilisation has expanded significantly. It is therefore not possible to disaggregate the contribution of Tarbela irrigation water separately. Comparison between actual production and predicted values for the IBIS as a whole is, however, considered useful to review the overall assumptions that formed part of the decision making process.

The cultivated and irrigated area increased after the construction of Tarbela dam, although not to the extent predicted. The cultivated area in the country increased by 12% from 19.6 mha in 1974-75 to 22.0 mha in 1997-98. The actual cultivated area in 1997-98 was, however, about 8% less than that predicted for the year 2000 (23.8 mha). The irrigated area served by canals met its predicted increase, although as mentioned above, some areas also received supplementary groundwater irrigation. The overall irrigated area increased from 13.3 mha in 1974-75 to 18.0 mha in 1997-98 - slightly higher than the predicted irrigated area in the year 2000 (17.9 mha). Canal irrigated areas increased 45% from 10.1 mha pre-Tarbela to 14.7 mha in 1997-98, and is mainly attributable to Tarbela and increased canal diversions.

There was a pronounced shift in the cropping pattern in the Indus basin as inferred from secondary data collected for this study in three selected districts of Tarbela command, viz: Rahim Yar Khan (Punjab); Nasirabad (Baluchistan); and Nawabshah (Sindh). Comparison of pre-Tarbela (1970-71) cropping with 1997-98 data shows that irrigated wheat, cotton, rice and sugarcane became predominant crops. Increases in the cropped area were found for wheat (36%), cotton (44%), rice (39%) and sugarcane (52%). The overall increase in cropping intensity was, however, less than predicted. The Lieftinck Report predicted cropping intensity for the irrigated areas of Punjab and Sindh rather than for the whole basin. Against the baseline of 95% for Punjab and 90% for Sindh during 1965-98, the cropping intensity in the year 2000 was predicted to increase to 150% and 137% respectively. Based on district level data, the actual achievement for Punjab and Sindh through 1997-98 was 117% and 132% respectively, that is, 22% and 4% lower than the projected figures for the year 2000.

The achievement in crop yields for the major crops in Punjab and Sindh were far below the predicted levels. The predicted yield of wheat in Punjab and Sindh for the year 2000 was
3 339 and 3 379 kg/ha respectively. Against this, the estimated achievement in these provinces in 1997-98 was 2 327 and 2 374 kg/ha respectively, being 30% less than predicted in both cases. Corresponding cotton yields during the same period were 31% and 10% less than predicted. Similarly, for rice the actual yields were 51% and 9% less than those predicted for Punjab and Sindh.

The actual gross production value (GPV) of crops was 28% higher than predicted for 1975 while it was 3% lower than predicted for both 1985 and 1998. The actual GPV for 1998 was Rs.19.96 billion compared to the predicted Rs.20.57 billion for the year 2000 (prices in 1965 rupees). A series of causes influenced the GPV. Factors potentially reducing GPV were lower than predicted canal diversions as other planned reservoir storage was not developed, lower cropping intensity and lower yields. Factors that increased GPV included a shift to higher value crops from those predicted and higher prices. Actual compound agricultural growth for the period 1965 was 3.18% compared to the 3.9% predicted. The national population growth rate over the same period was approximately 3%.

Tarbela was designed to compensate for loss of agricultural production as a result of ceding water of the three eastern rivers to India and to increase agricultural production to meet the growing needs of the expanding population. This objective has been partially achieved. Food security has been a prime policy goal of all policy makers in the country. Although the gross value of crop production has grown faster than the population rate, the domestic production of basic cereal has been inadequate and the country has had to import wheat for the last 25 years approximately.

3.6 Salinity and Waterlogging

Waterlogging and salinity have been increasing in the IBIS as a result of increasing diversion of water for surface irrigation. Where there is excessive irrigation and inadequate drainage, the groundwater levels rise, and where they are close enough to the surface, capillary action draws up salts from the soil into the root zone and land surface. Approximately 60% of the aquifer underlying the IBIS is of marginal to brackish quality. The problem of waterlogging and salinity became apparent in the late 1950s. In areas with fresh groundwater, the water lost from the canal system can be re-used by pumping (at an additional cost to farmers). However, in brackish groundwater areas this is generally not possible, although groundwater is sometimes mixed with fresh canal water. In 1975 the proportion of the IBIS area with groundwater less than 3 metres from the surface was 42% and the area with groundwater less than 2 metres from the surface was 22%. In Sindh the area with a depth less than 3 meters was 57%. Although groundwater use has increased significantly, about 22% of the command area of IBIS has a water table of less than 1.5m.

Compared to the projected drainage coverage of 7.6 mha by 1980, the actual coverage by that date was 5.2 mha (32% less). Against the gross command of about 16.7 mha in IBIS, about 6.0 mha had been provided with drainage while projects in an additional 2.4 mha area were under implementation. It was further anticipated that over and above this, a remaining area of about 4.5 mha would require drainage. Out of the ongoing projects, The Left Bank Outfall Drain (LBOD), though substantially completed is still only partly operational. Similarly, the Right Bank Outfall Drain (RBOD) is still in the early stages of development. There are apprehensions based on the experience with the Salinity Control and Reclamation Projects (SCARPs) that the high O&M costs of these outfall drains may preclude their sustained utilisation to dispose of the saline effluent to the sea, despite the launch of the National Drainage Programme (NDP) in 1998.

3.7 Hydropower

The original design of Tarbela included staged development of twelve generators of 175 MW capacity providing a total capacity of 2 100 MW by 1980. Of these units, only civil works of the first four were included in the financing package of the TDF. In the early 1980s, a power optimisation study led to the decision to increase the installed capacity to 3 478 MW. By 1980, the installed capacity was 700 MW, which increased to 1 750 MW by 1985 and the final capacity of 3 478 MW by 1993.

The predicted average annual generation at full power development of 2100 MW was 12 600 GWh. To account for staged development, the average predicted generation for the period 1975-98 is 11 300 GWh. The actual average annual generation (1978-98) once commissioning was complete, was 9 255 GWh, or 82% of predicted. After installation of full capacity in 1993, the average annual generation was 14 300 GWh (1993-98), ie 13% higher than the full development prediction of 12 500 GWh. Total actual generation in the period since commissioning (1977-98), was 194 500 GWh, or 79% of the predicted figure of 245 300 GWh (adjusted for the two-year delay).

In June 1998 Pakistan had an overall installed capacity of 15 844 MW, comprising: hydro (4 825 MW); thermal (6 995 MW about equally divided between gas-fired and oil-fired); and independent power producers (4 024 MW all oil-fired). Out of this, Tarbela represented over 22% installed capacity providing 28% of annual generation requirements for the inter-connected national grid system of WAPDA.

The net economic benefit of hydropower from Tarbela was compared to the other alternatives at the time of planning in 1965. It was estimated that present net worth of economic power benefits of the TDP (at 8% discount rate), as compared to the next cheapest thermal alternative, would be $160m. Based on actual generation data from 1977-98, economic power benefits of the TDP have been determined on the basis of savings in generation cost per kWh as compared to the most efficient gas-fired plant in the WAPDA System. This exercise yields a present net worth of electric power from Tarbela (at 1965 prices and a 8% discount rate) of $225 million that is 41% higher than predicted. This is equivalent to a net benefit of $1 160 million in 1998 prices. It should be stressed that this is not the gross economic benefit, but the incremental benefit over the next cheapest option. Quantification of unpredicted dynamic benefits was $60 million and avoided thermal plant emissions was $162 million at 1965 prices ($310 million and $835 million respectively at 1998 prices).

The province of NWFP where Tarbela is located, has been receiving Rs.6 billion ($139 million in 1998 prices) annual royalty from hydropower generation at Tarbela after the promulgation of the 1973 Constitution. Tarbela will provide an unexpected power benefit by contributing 20% (1 353 out of 6 586 GWh) of generation from the Ghazi-Barotha Hydropower Project currently under construction immediately downstream of Tarbela.

3.8 Flood Attenuation

Flood management was not included as an important objective of the dam at the design stage and no predictions were made. However, the impact of TDP on attenuation of actual high flood peaks was significant during the filling period of June through August, for impounding about 12 bcm, or 19% out of the kharif inflow of 64 bcm. Attenuation of peak Indus flows is variable, depending on the timing of the flood in relation to reservoir level that is drawdown prior to the wet season. The peak flows in July 1988, July 1989 and August 1997 were reduced by 21%, 26%, and 43% respectively, whereas a peak flow of similar magnitude in September1992 was attenuated by only 2% as reservoir levels were already considerably higher in readiness for the forthcoming irrigation season.

3.9 Municipal Water Supplies

Augmenting municipal water supplies and providing water for industries was also not a significant objective of Tarbela and no targets were fixed. However, since groundwater in vast areas of Pakistan is saline and unfit for human consumption or use by industry, the progressive increase in population coupled with rapid urbanisation and expansion of industries necessitated the harnessing of canal water. For example, in the Sindh, 89% of groundwater is classified as brackish, which means that canal water is the sole source of municipal water supply, either directly, or as a result of the thin layer of freshwater that overlays the brackish water. Water supply to Karachi has benefited from diversions made through Kotri barrage. The indirect contribution of IBIS and of Tarbela dam to supplement municipal water supply has been an important benefit. It is estimated that about 40% of the population may presently be benefiting to some degree from the waters of the Indus that are supplemented by Tarbela storage during the dry season.

3.10 Resettlement

The project design estimated that 100 villages would be submerged, resulting in the resettlement of 80 000 people. The number of villages actually affected was 120, and the number of people affected closer to 96 000, an increase of 20%. The basic law governing resettlement was the Land Acquisition Act of 1894, and criteria for compensating land owners were set by government in 1967. The main criterion for compensation with alternate land was whether land holdings were greater than 0.2ha of irrigated land or 0.8ha of rainfed land. Approximately two-thirds of the affected population were eligible for replacement land in Punjab or Sindh provinces. Those with holdings of less than these limits were eligible for cash compensation. People with houses in the affected area were to be paid cash compensation at 1968 market values and expected to purchase new housing and commercial lots in five villages to be built in the vicinity of Tarbela.

There has been widespread dissatisfaction with the resettlement process for the following reasons:

• lack of participation in the resettlement process;
• displacement occurred prior to receiving possession of alternate land and compensation;
• the Sindh government's refusal to release 65% of the land initially allocated;
• incompatible social and cultural backgrounds of affected and recipient communities;
• delays in the decision of eligibility;
• delays in making payments resulting in reduced real value of compensation due to devaluation of the rupee in excess of 100%;
• lack of employment opportunities in the new townships;
• corruption of officials seeking 'facilitation' payments;
• villagers who did not receive compensation in time had to moved from the reservoir area by the army during impoundment;
• disputed responsibility between institutions for maintaining infrastructure in the new townships; and
• landless and tenant farmers did not have the appropriate skills for non-agricultural employment.

A survey in 1996 showed that 1953 eligible families who applied for alternate land still await possession. A number of challenges were made to the resettlement and compensation process. The adopted criteria were challenged by local groups and a number of cases were taken to the courts of which 38 are still outstanding. The Government created the Nucleus Clearance Cell (NCC) to investigate claims pertaining to the TDP. More recently, in response to the financing conditionality of the downstream Ghazi-Barotha project, a Commission was formed under the NCC to address the outstanding claims. Of approximately 11 000 claims before the Commission, only 4.4% were considered eligible. Many were rejected due to problems with verification by the Revenue Department, although this is contested by local groups. The number of people affected by the project but not eligible for compensation has not been quantified, although assertions from concerned groups place this number in the same order of magnitude as the number of those that are eligible.

Field surveys conducted as part of this study show that people affected by the dam were not systematically involved in the planning and development process. Many felt the project had problems with adequatelycompensating and resettling those affected. In spite of over 24 years of operation of the dam, this is still an active issue. An opinion strongly expressed is that those directly affected and who made a major sacrifice for the project, have not been fairly compensated. They demand that a part of the earnings from the dam should be earmarked for the development of the communities where the displaced have settled and for those that were affected but kept outside the formal compensation mechanisms. In addition, the gender aspect has been totally neglected in the resettlement process, and women, specifically, have suffered as a result of the disruption of their social life due to migration from ancestral places.

3.11 Social Impacts Downstream

A series of local surveys were carried out as part of the study to obtain an insight into the changes that occurred over the period of project operation. Two sites were selected in the Punjab: an agricultural area in Mailsi; and locations along the river in Muzzaffargarh, DG Khan. In general, respondents stated that agricultural production had increased significantly, confirming the changes in cropping patterns to sugarcane, cotton, wheat and rice and orchards, and that land rental values had increased more than 20 fold, almost double the local inflation rate. Negative changes included the loss of forest land to agriculture and loss of livelihood opportunities to the landless who made products from forest and wetlands along the river, representing marginalisation of poorer sections of the population who depended on the river.

In the Sindh, two areas were selected for survey: Thatta and Badin, an area dominated by river and marine fisheries; and Keti Bandar and Ibrahim Hyderi in the coastal mangroves. From Thatta and Badin, there are two main fishing points, a series of lakes including Kinjhar lake where approximately 400 000 people fish, and the open sea. Overall, approximately 3 million people are involved in fishing. Fish catches were considered to have decreased significantly as a result of reduced dry season flows in the Indus (starting with the construction of the Kotri barrage), loss of mangroves, and the sharp increase in numbers of fishers. (Further information on fish migration is included under section 3.12.1Ecological Effects). In the areas served by the surface irrigation system, however, sugarcane has increased substantially, requiring high amounts of irrigation and increasing disparities in distribution of irrigation water.

During field hearings at Ibrahim Hyderi and Keti Bandar, it was stated by the local residents that the upstream developments of barrages and dams on the Indus River (especially the Kotri barrage and Tarbela dam), have further reduced the flow of river water below Kotri into the Indus delta, especially during the rabi season. This has resulted in a reduced supply of fresh water for drinking purposes and for agriculture. The intrusion of sea water has been gradually increasing and agriculture in the delta has received a severe setback. Mangrove forests and fish breeding grounds have also been adversely affected. There has been a large-scale out-migration of population from the lower Indus delta to Karachi and other places in search of livelihood.

3.12 Ecological Impacts

The ecological impacts of the dam were not considered at the inception stage as the international agencies involved in water resources development had not realised this need at that time. This study has revealed several important ecological impacts, mostly negative, on the physical and biological environment. It must, however, be stressed that these negative impacts on the environment are the result of numerous dams and barrages on the Indus River system including the developments in India, and cannot be ascribed to Tarbela alone. The area downstream of Kotri has been a contentious issue ever since the development of irrigation works on the Indus. Major negative impacts are obvious on the Indus Delta, like reduced mangrove areas, negative impacts on human livelihood, and massive destruction of forests, flora and fauna. While such impacts are primarily due to the construction of Kotri and Sukkur barrages that reduced downstream flow of sweet water into the Indus delta, the perception developed by communities downstream lends them to believe that any irrigation developments on the Indus further aggravate the situation in the Delta.

Tarbela has contributed significantly to the increased storage and abstraction of water from the Indus resulting in an increased frequency of low floods that now cover less than 25% of the flood plain. The sedimentcarried down the Indus has also reduced due to abstractions during high flow period. This has shifted the balance between erosion due to high-energy waves and sediment deposition particularly in the Indus delta.

Lower flows and the construction of flood bunds all along the Indus in Sindh has led to the reduction in riverine flood coverage and caused a change in the habitat of the flood plain. Higher forested areas, islands and agricultural lands have dried out and experienced species changes, from Acacia to the less economic Prosopis. Increased use of pumped irrigation has been required to maintain riverine forests, especially in Sindh. The forested areas in Sindh have reduced to only 50% of the original 240 000 ha and the remaining areas are of variable quality.

Building of the dam has prevented the migration of Mahseer (Tor putitora) into the cooler waters upstream during summer, and the occasional operational peak releases can wash out the substrate and fish immediately below the dam, causing a loss to local fishermen. The reduction in flooding also causes losses in fish populations, and in the downstream area the proportion of wild fish caught has declined in comparison to aquaculture. The catches of migratory fish such as Palla and Baramundi have also decreased but these are principally due to the building of Kotri Barrage, rather than as a result of Tarbela. The catch of Palla has reduced from about 10 000 tonnes during the 1980s, to about 200 tonnes per year in the 1990s.

Tarbela reservoir is a staging point for migratory birds, but is not suitable for over-wintering birds. In contrast, Taunsa and Chashma barrages have become significant wetlands important for migratory birds and fisheries although they are under considerable pressure. Populations of smooth coated otter and hog deer are struggling to survive due to a lack of controls. Changes to the floodplain due in part to flood bunds has reduced the indigenous Acacia nilotica and beds of Typha, and allowed the invasion of exotic species with negative impacts on wildlife. The contribution of Tarbela to the decline in wildlife has been relatively minor in comparison with other interventions in the river system.

A sanctuary for the Indus River dolphin between Sukkur and Guddu barrages has stabilised numbers to about 500. Further downstream, however, the salinity increase in Manchar lake resulting from saline drainage has changed the species distribution and loss of the main commercial species Rohu. Due in part to the flood bunds, and also to reduced flood flows from Tarbela, there has been a reduction in the dilution of salinity in the lake.

Because of extended zero flow periods caused by diversion of water following construction of the Kotri barrage, the river bed below Kotri has become severely braided and the channel can become blocked by sand bars. Seawater intrusion has increased during the rabi season, reaching approximately 25km upstream from the delta. The active Indus delta has been reduced to about one tenth of its original size, largely as a result of the construction of river flood bunds confining the river. The freshwater and sediment discharge during the flood period is restricted to the active delta, but the increase in salinity during the low flow periods has reduced the suitability of the delta for the cultivation of red rice, the production of exotic fruit, and raising of livestock.

The mangrove ecosystem is being degraded and the mangroves are now virtually mono-specific and comparatively stunted with losses of about 2% per year. Degradation of the mangroves is both because of reduced water flows and direct human destruction and over use. The major changes in river flow below Kotri have affected the ecology in lower Sindh and the coastal areas significantly, besides adversely affecting agricultural production. Marine fisheries have been affected due to the decrease in freshwater, sediment and nutrient flows. Although overall fish catches increased from 1950-84 to a total of 221 500 tonnes, the fishing fleet also expanded considerably and the catch per unit of effort reduced threefold. The protective nature of the riverine and mangrove forests has also reduced thereby increasing the risk of storm damage.

The direct impacts of Tarbela are difficult to separate out from the impacts of other changes experienced in the Indus Basin as a whole, including the earlier dams, barrages, and flood bunds leading to abstraction of water and containment of the flood plain. Other factors include population growth and increased urban and industrial pollution particularly from the eastern rivers, such as the Ravi (from Lahore). The main influences of Tarbela on the ecosystem were due to the significant reduction on sediment flows causing transgression of the delta and the lower flood flows. Low season flows did not reduce significantly further as a result of the Tarbela.

4. Distributional Effects of the Tarbela Dam Project

The Lieftinck Report gave a broad indication of the distribution of benefits from the TDP indicating that the benefits would be split, with approximately 75% deriving from irrigated agriculture and 25% from hydropower. The irrigation benefits relate to the protection of water supplies to approximately 1.8 mha, and to providing additional irrigation supplies to 6.9 mha. The electricity generated would be fed into the national grid system and therefore beneficiaries would be widespread urban, industrial and rural consumers. Specific mention was made of electrification of groundwater tubewells. More detailed analysis of benefit distribution was not implicit in the planning documents. As mentioned earlier, the negative social and environmental impacts of the project were not identified at the planning stage and therefore affected groups were not identified. The agricultural benefits and negative impacts are due not only to Tarbela, but also to other developments and investments.

Surface irrigation water made available by Tarbela and hydropower generated by the dam have directly or indirectly benefited a large proportion of the population of Pakistan. The rural farming population directly affected by irrigation benefits is estimated to be in the order of 7 to 10 million. Farms receiving additional water supplies were located in southern Punjab and Sindh. The additional water resulted in increased cropping intensities, a shift from 6 monthly supplies to a perennial supply, and increased availability of water per unit of land. Although detailed studies were not carried out, there is a general perception that the larger landholders benefited more than medium and smaller landholders. Land prices for irrigated areas have increased significantly from Rs.37 000 per ha in 1970 to Rs.620 000 per ha in 1998. In real terms this represents a 300% increase (also see discussion on property values in section 4.2.2.6) Increased water supplies from surface and groundwater irrigation have contributed to price stabilisation of wheat, cotton and sugarcane, benefiting consumers across the country.

Major beneficiaries have been the enterprising industrialists who received cheap electricity thus reducing their cost of production, and large landholders with access to cheap power to pump groundwater. The small farmers and landless tenants have also benefited by being able to buy tubewell water from neighbours. With increased agricultural mechanisation there was some displacement of farm labour who had to find alternative livelihoods in the urban areas. Employment in agro-industries, such as cotton- and sugarcane-related industries have increased significantly, together with forward and backward linkages including the transport sector.

Currently, at about 28%, Tarbela dam has maintained a significant proportion of the overall share of electricity generated in the country and can therefore claim to have helped stimulate urban and industrial growth. WAPDA's urban electricity consumers increased to about 5.6 million by 1998 compared to 1.5 million in 1976. In rural areas, the electricity consumers increased from 0.4 million to 4.6 million The budget of NWFP benefited significantly from the TDP with annual transfers of Rs.6 billion ($139 million equivalent in 1998) in royalty payments. Other unexpected benefits were the availability of water for domestic purposes and livestock production in areas where groundwater supplies are too saline.

The social costs suffered by those displaced by the reservoir were immense and are described above in section 3. Directly affected people numbered 96 000. No compensation or development opportunities were provided for those indirectly affected. Negative aspects were the result of inadequate compensation and the loss of livelihood leading to lowered social status in many cases. In downstream riverine areas, those dependant on flood based recession agriculture and the natural resources of wetlands and forests were negatively affected by reduced flood flows, due in part to reduced flows, and to the impact of flood embankments. Fishermen were also negatively affected due to the reduction of fish in the river. However, this was predominately caused by the construction of barrages in the river and the reduction of dry season flows that commenced with the construction of Kotri barrage in the late 1950s.

In the coastal areas, the reduced dry season flows and the reduction in sediment flows, led to saline intrusion, and a reduction in mangrove and forest areas and consequent loss of livelihoods. The contribution of Tarbela dam to this loss cannot be desegregated from other causes as the major changes in flow regime started 16 years before the dam was commissioned. Although dry season river flows in the Indus directly downstream of Tarbela have increased as a result of the dam, this has not led to increased flows downstream of Kotri barrage, but rather to additional abstraction of irrigation flows from the river. Increased population pressure also plays an important but unquantified role in deterioration of natural resources in the area.

General development indicators have improved since construction of Tarbela although this cannot be attributed to the project but rather to other development activities. Life expectancy has increased from 45.8 in 1970 to 55.8 in 1998 and there has been an increase in literacy rate form 18.4% to 34.9% in 1990 although female literacy remains almost half that of male literacy.

During the peak construction period, a labour force of about 15 000 was employed at Tarbela. This offered jobs at high rates, helped to train highly skilled manpower for further national development and for lucrative international employment. Even during the current phase of O&M, Tarbela employs a workforce of about 4 000 thus benefiting about 20 000 members of their families.

5. Decision Making Process and Options Assessment

Discussions relevant to diversion of eastern rivers by India were carried out for a decade (1951-60) within the GOP and tough negotiations with India were mediated by the World Bank. The final decision of Pakistan to accept the Indus Waters Treaty (IWT) was made by the President of the martial law Government, Field Marshall Ayub Khan. During this period, interim arrangements for water distribution were agreed. The final decision to accept the IWT, although quite controversial at that time, was a follow-up of the Inter-Dominion Agreement of May 1948 between India and Pakistan, and the World Bank's plan for division of the Indus waters in 1956. The IWT included provision for a dam on the Indus. Investigations into alternative dam sites had already commenced after independence and reconnaissance studies for Tarbela started in 1954. Three sites were initially considered: Tarbela and two sites 4km and 24km upstream, of which Tarbela was favoured because of higher capacity and cost effectiveness.

Following the signing of the IWT, the World Bank initiated the Indus Special Study (ISS) headed by a vice president of the Bank, Pieter Lieftinck. In the resultant Lieftinck Report, three alternative sites were considered: Tarbela and Kalabagh on the Indus downstream, and Gariala on the Haro river. The other alternatives to either raise the height of Mangla dam or pump underground water to replace the supplies lost to India were not considered feasible As far as records show, no other options such as groundwater development were investigated.

The first phase of ISS confirmed the selection of Tarbela as the most appropriate site for development. In May 1968, a meeting of donors of the original Indus Basin Development Fund agreed to provide supplementary finance through the creation of Tarbela Development Fund (TDF) to cover the shortfall of funds necessary to proceed with the project. Problems with geology were identified and allowance made in the design. Technical problems encountered during impoundment required emergency repairs and remedial works. Two supplementary financing arrangements for the TDF were made in 1975 and 1978 with the latter involving a major increase in contribution from the Government of Pakistan requiring funds to be diverted from other projects to ensure completion.

6. Criteria and Guidelines

Soon after the commissioning of the dam in 1976, it was realised that the overall efficiency of irrigated agriculture was unsatisfactory and needed to be improved substantially to derive maximum benefit from major irrigation projects like Tarbela and Mangla. A Revised Action Programme (RAP) to improve O&M and reform institutional structures was prepared by WAPDA in 1979 with UNDP assistance and administered by the WB. This plan was comprehensive in scope.

The Water Sector Investment Planning Study (WSIPS) of 1990 reviewed the achievements of RAP and concluded that little headway had been made to bring about institutional reforms to effectively co-ordinate the activities of irrigation and agriculture departments. WSIPS also emphasised the need to rehabilitate the deteriorating irrigation system of the Indus basin and improve the agricultural production system through improvements in research and extension.

The issue of transforming the irrigation bureaucracies in the provinces in order to make them more autonomous and sensitive to the economic and biological aspects of irrigated agriculture, as compared to the prevalent emphasis on engineering aspects, was taken up by the WB during the processing of the National Drainage Programme (NDP) in 1997. As a result of extended negotiations with GOP and provincial governments, the semi-autonomous Provincial Irrigation and Drainage Authorities (PIDAs) were established through acts of the provincial assemblies. Although the enactment to establish PIDAs has fulfilled the WB requirement for disbursement of the loan for NDP, it still requires considerable improvement to achieve its aims of institutional reform and improved accountability and performance.

Another major issue impacting on irrigated agriculture has been cost recovery and taxation of agricultural income. Major aid donors to Pakistan have insisted on realistic water taxes to generate income to meet the huge operation and maintenance (O&M) costs of the irrigation infrastructure. This item has been included as a covenant in most foreign-aided development projects in the irrigation sector. However, this has been a highly controversial issue with bitter opposition by the farmers' lobby. Although some increases in water rates have been made, the rates are still far below what would be required to recover O&M for the irrigation supplies. The issue of agriculture taxation, though agreed in principle, has not been faithfully implemented. This is mainly because of collusion between large landholders and the lower level revenue administration that helps conceal the actual landholdings to evade agricultural tax.

On the power side, the government is moving towards the creation of a competitive market through restructuring and privatising the existing thermal generation and transmission and distribution functions. In 1997 the autonomous National Electric Power Regulatory Authority (NEPRA), was established for licensing generation, transmission and distribution. Unfavourable experience with the high cost of generation from Independent Power Producers in the early 1990s, led to a policy in 1998 that gives preference to utilising hydropower, but with the active participation of the private sector.

Environmental impact assessments were introduced as a requirement of all major projects from 1975 and included as a separate chapter in feasibility studies. Comprehensive assessments, however, were not included until later when multilateral banks adopted policies on environmental impact assessment. From 1990, environmental considerations were incorporated in regional and sectoral planning studies. The National Environment Protection Act of 1997, established an Environmental Affairs Division at federal level with cells at provincial level. A more participatory approach to environmental issues is demonstrated by the National Conservation Strategy published in 1998.

7. Lessons Learned

The following main lessons were presented for discussion at the meeting of stakeholders held in Karachi in January 2000. Some participants declined to take part for various reasons including concerns that their response could in some way compromise discussions over any future project in the basin. The responses are therefore not fully representative. Of those that did respond there was general agreement on all except lessons 5, 8, and 10 where comments are noted.

1. The sensitivity of fluctuations in future production due to climatic factors and response to pricing signals and changing circumstances needs to be given more attention in the planning stage.

2. The benefits of irrigation for agricultural production need to be targeted. Unequal distribution of benefits generates inequality amongst groups in specialised crop production systems.

3. Technical improvements through research and extension have a major pay-off through increased productivity and net incomes. This component needs to be incorporated in the plans at planning stage of major irrigation projects.

4. Planning of power production should take greater account of variability of production and its effect on the power system. Power benefits vary during dry and wet cycles, whereas for appraisal the average generation is used. Though in case of Tarbela the predicted output was considered to be broadly stable from year to year, it has been variable in actual practice due to a variety of reasons.

5. Hydropower has a distinct comparative advantage in reducing CO₂ and other greenhouse gases emissions. This has substantial benefits in the regional/global context. (This lesson was challenged by a number of participants who considered that there could be significant emissions from reservoirs and that any benefits on the emissions side should not be used to balance other negative environmental impacts of dams).

6. Pre-project base-line knowledge of the river ecosystem and environmental parameters is essential to draw any meaningful inference for project planning and to design in-stream flow requirements.

7. It is desirable to undertake an independent review of large projects to confirm feasibility and a full options assessment before the start of implementation.

8. For large projects like Tarbela, phasing is sometimes necessary due to funding and other constraints. However, preferably, it should be avoided lest a changing context may lead to dropping of subsequent phases especially during implementation. (There were contrasting views on this point but most agreed that planning stages need careful phasing to ensure that social and environmental concerns are adequately addressed).

9. Developments within a basin context must recognise the broader implications of upstream project development. The quest to harness water resources must be sensitive to impacts on lower riparians. This can have far reaching socio-economic and political ramifications. Water requirements of the Indus delta must be viewed in a wider policy framework to mitigate massive ecological imbalance

10. Some downstream water quantity and quality concerns especially below Kotri, can be addressed by maximising storage releases from the dams under favourable inflow conditions. Besides improved hydropower generation, it would help reduce cumulative impacts of the dams on the riverine ecosystem. (Some disagreed as it might compromise irrigation releases).

11. Compensation and resettlement policies must be discussed with the targeted affectees through some process of participatory consultation. Care should also be taken to identify and adequately benefit the genuine affectees, especially the landless tenants, traders and artisans, etc.

12. Major disputes (international/national) on distribution of river water should be resolved before embarking on large water harnessing projects.

13. Project O&M activities must be established at the onset of project along with adequate funding. These operations must be integrated with overall national/provincial planning systems to accommodate differences between predicted and actual conditions and changing circumstances and priorities.

14. Monitorable targets are required for project objectives including rural and national development and an appropriate monitoring system established. Normally, macro project objectives such as "improving the rural economy" and "promoting national development" are impossible to evaluate post-hoc unless targets are set, and a deliberate monitoring system is designed and implemented.

15. Even when no formal post-project evaluation is intended, periodic in-house performance reviews are helpful for ongoing planning and decision-making.

16. Post-project impact monitoring can help identify, assess, and rectify unforeseen adverse impacts of the project that may occur over the longer term, or to identify opportunities to enhance positive impacts of the project. Such monitoring also provides a credible basis for new operational criteria, standards and regulation.

Subsequent to the meeting additional fieldwork gave rise to an additional lesson namely, "Developments within a basin context must recognise the broader implications of upstream project development. The quest to harness water resources must be sensitive to impacts on lower riparians. This can have far-reaching socio-economic and political ramifications. Water requirements of the Indus delta must be viewed in a wider policy framework to mitigate massive ecological imbalance".

Further lessons were suggested by the participants including the need for a new national resettlement policy, greater participation in planning and design, more attention to social and environmental impacts, internalising of external costs, evaluation of distributional aspects, greater attention to risk, and attention to monitoring and resolving impacts during operation.

8. Stakeholders' Views

Many of the stakeholders contacted during the study consider Tarbela a useful project in terms of development effectiveness for the country since it has helped stabilise and expand agricultural production. The contribution of Tarbela to power generation and averting the severe power shortages prevailing in the country until recently, is also widely recognised. However, it is also recognised that the displaced population was not involved in the planning process and the resettlement process resulted in considerable hardship to these communities. Furthermore, the interpretation of affected people was rather narrow, and many ineligible groups were adversely affected without receiving any compensation.

Reduction in river flow below Kotri, initiated 16 years prior to Tarbela, also adversely affected some communities, especially the fishermen, who were not included in any compensation package. Stakeholders also emphasised the adverse environmental effects of the dam and other interventions especially in the downstream areas. Some of these communities faced accelerated migration to other areas. These communities have strongly expressed the view that upstream developments on the River Indus impacted on them negatively. Such stakeholders feel that while royalties are provided to those regional governments where projects are constructed and direct and indirect costs are generated, project designs must be sensitive to incorporate "damage compensation" for those affectees (especially downstream) who are normally not considered as "legitimate affectees".

There was no overall consensus on whether positive impacts outweighed negative impacts, which points to the difficulty in determining trade-offs among widely different communities, some of whom benefit and some who lose. There was, however, general agreement that there was inadequate participation in the decision-making processes.

Regarding water use efficiency for agricultural production, most stakeholders felt that there was considerable room for improvement. More efficient management of the complex agricultural production system can also significantly increase crop yields with emphasis on water use efficiency.

9. Summary of Predicted, Actual and Unexpected