Assessing water scarcity : an example from he Gediz basin , Turkey

Müslüm BEYAZGÜL*, Atila GIRGIN*, Göksel GEÇGEL*, Hammond Murray-RUST**, Charlotte de FRAITURE** and Peter DROOGERS**
*Tarimsal Hidroloji Arastırma ve Egitim Merkezi (thaem),
(35660) Menemen, IZMIR, TURKEY, Phone: +90 232 831 10 52 – 831 05 12, fax: +90 232 831 10 51, e-mail: thaem@khgm.gov.tr

**International Water Management Institute, P.O. Box 2075 Colombo Sri Lanka, Phone:+94-1-867404 – fax: +94-1-866854,
e-mail:H.Murrayrust@cgiar.org

Abstract

The Gediz Basin in western Turkey is widely considered to suffer from long-term water scarcity.  By combining data from a wide variety of sources including several Government agencies, LANDSAT and other public domain sources, it has been possible to assess the water balance of irrigated agricultural areas at three different levels: basin, sub-basin and Irrigation Association level. Analysis indicates that surface water deliveries into the 13 Irrigation Associations are adequate to meet not only surface water demand but also the demand from other irrigated areas, and that for the past three years there has been sufficient water to meet all irrigation demand at basin and sub-basin level.  Only in a few Irrigation Association areas does there appear to be a water deficit. 
Key words: Irrigated area,water scarcity, assessing, Gediz basin, Turkey

Introduction

Water scarcity is an increasing threat around the world, and is rapidly becoming the focus of major initiatives to safeguard human development in the next century. Efforts to identify locations where threats of water scarcity are increasing gain momentum, and alternative management strategies are being developed that will help policy makers and managers in coping with these threats. 
The Gediz Basin in western Turkey provides a good example of how it is possible to quickly evaluate whether or not water scarcity exists in a basin, and the extent to which there is a functional scarcity.  By combining readily available public domain data it is possible to make a quick assessment of present water use conditions, and estimate the extent to which a functional water scarcity may occur in the future.

Materials and methods

The Gediz Basin has a total area of some 17,000 square km just north of the city of Izmir in western Turkey.  Because it is agriculturally important and because it is close to major centers of population it has received considerable attention from planners and scientists in recent years. 
The basin is essentially fully developed in terms of water resource management.  There are three reservoirs in the basin, the most important of which is Demirköprü Reservoir.  This was constructed in 1970 as a multi-purpose flood control, hydropower and irrigation reservoir.  For the first two decades of operation flood control was the most important function of the reservoir, with irrigation playing a secondary role.  Hydropower was generated whenever reservoir releases were made and in periods of surplus special hydropower issues were also made.  Average inflow into the reservoir between 1970 and 1977 was 496 million cubic meters per year.  Between 1978 and 1988 higher rainfall in the catchment led to higher inflows, and they averaged 706 million cubic meters annually in this period. 
From 1988, however, a drought started that dramatically decreased the inflow into the reservoir.  While rainfall dropped only very slightly throughout the lower part of the basin, there was a noticeable decrease in rainfall in the upper catchment, and this greatly reduced inflows in to the reservoir in winter months (Kite and Droogers, 1999).  The total annual inflow dropped to an average of 327 million m3 in the period 1989-1997.  In 1998 inflows totalled 620 million cubic meters, and in the first two months of 1999 inflows totalled 425 million cubic meters, enabling the reservoir to completely fill up for the first time since 1988. 
Conventional wisdom is that the Gediz basin is water short: there are restrictions on the use of water for irrigation which appear to have led to reductions in the overall irrigated area during the past decade, and the demand for water for domestic, industrial and environmental reasons continues to increase.  This paper determines the extent to which this assumption is correct by examining water use in the agricultural sector. 
Basis for the Analysis; Because the different data sources use different boundaries for reporting it is necessary to decide on a common basis for undertaking the analysis.  Three levels of analysis have been attempted in this study at basin, sub-basin and Irrigation Association level.
Sources of Data;
The assessment of water scarcity in the context of irrigated agriculture requires information on the availability of water and the use of it by irrigated agriculture.  As a consequence several different data sources have to be combined before a complete assessment can be made.  Because of changes in reporting practices since the creation of Irrigation Associations only data from the period 1996-1998 have been used. 
For the Gediz basin study the following sources of information have been used: 
Water Deliveries ;
DSİ Publishes, Irrigations assosciations annual reports, 
Irrigated Area ;
Irrigation assosciations records, Landsat images (landsat images were used to calculate NDVI values using the ILWIS GIS pakage) 
Cropping Patterns ;
Irrigation assosciations records(DSİ,1998), Landsat images ( landsat data using ILWIS a tentative classification in to the three crops-cotton, grapes and others ), 
Meteorological Data ;
Meteorological data were obtained from the Department of Meteorology (DMİ,1998) in Ankara and MenemenResearch stitute(GDRS,1998). 
Data used in this study include maximum and minimum temperature, relative humidity, average daily wind speed, hours of sunshine and rainfall. 
Calculation of Potential Crop Water Demand;
For each crop the potential evapotranspiration was calculated using CROPWAT (Smith,M.,1991) for each 10-day decade during the irrigation season for each of the years of the study.  Because the Gediz area is normally quite dry during the peak irrigation season of July and August there is little variation from one year to the next. Once the PET had been calculated for each crop the total demand for each irrigation system was determined by multiplying the total cropped area with the CROPWAT results for each crop. 
Determination of Soil Moisture Availability;
The final data used in the calculations was an assessment of the total change in soil moisture from the start to the end of the irrigation season.  This was done using the SWAP simulation model (van Dam et al., 1997).  The best estimate based on several simulations was that there was probably a decrease of 50 mm of available soil moisture during the irrigation season, and a decrease of 100 mm from the start to the end of the growing season (Droogers, 1999).  These values were included as a standard for each year of analysis rather than trying to make separate estimates for each year.

Methodology

The results of this study were obtained from using a simple spreadsheet that determines the net water balance for each area chosen, and also calculates Relative Water Supply, calculated in this paper as: 
    Relative Water Supply = (Irrigation + Decrease in Soil Moisture)/(Water Requirement) 
To understand fully the extent of water shortage in the Gediz Basin during the summer months two time periods have been used: irrigation season and full season.  Three successive years were analyzed because there were significant changes in overall water availability during this period. 
Irrigation Season is determined by the period of releases of irrigation water from Demirköprü Reservoir into the surface irrigation system.  Throughout the three years covered in this study the official irrigation season was from July 1 to August 31.  The reason for undertaking this study was to try to assess the degree of water scarcity or surplus when canal water is available as this will give insight into the extent to which strict water management is necessary 
Full season is defined here as the period 10 June to 20 September.  This is the period when crops normally require supplemental water to give maximum yield.  In the Gediz area the two dominant crops are cotton (50% of the total cropped area of the basin) and grapes (35%).  Cotton is planted in May but is able to use soil moisture for the first few weeks.  Modeling using the SWAP model indicates that typically cotton will become stressed without irrigation by mid-June.  Grapes can access much more soil moisture due to deeper rooting but also benefit if they receive a June irrigation.  Cotton still benefits from irrigation September, but towards the end of the month harvesting has started and demand for water decreases rapidly towards the end of the season.  Grapes, the vast majority of which are processed for raisins, may be harvested in August or September, so only limited amounts of irrigation are required in September. 
. 
The water balance spreadsheet calculates total water demand (pET less effective rainfall multiplied by cropped area and crop coefficient for each 10-day decade) based on a two-fold cropping pattern (cotton and other crops) and compares this with total water available (irrigation water supplied plus soil moisture depletion).  The balance represents the surplus or deficit, expressed in both total volume and in depth in mm over the total irrigable area. 
Relative Water Supply is calculated both for the area reported by Irrigation Associations as being irrigated with surface water (actually the area irrigated using surface water for all or part of the total irrigation application) and for the total irrigated area that was calculated using the LANDSAT images. The difference is important in understanding both overall water scarcity and irrigator behaviour. The RWS for the surface irrigation area represents the overall water environment in which members of Irrigation Associations function.  It determines the ratio between canal water discharges and crop water demand of IA members only.  If RWS values are high then it implies water is quite plentiful and the Irrigation Association will not be under any severe managerial stress.  As RWS values drop then there will be a need for tighter management to spread water among users.

Results and discussions

Results are presented for both irrigation season and whole season periods at basin, sub-basin and irrigation season levels.  The key indictors of water scarcity are the estimated water balance and the Relative Water Supply.  RWS has been calculated for both the areas reported as being irrigated using surface water (RWSsurf), and for the entire irrigable area (RWStia) 
Basin Level Analysis; At the basin level there is no evidence that water is in short supply during the irrigation season.  For each of the three years of data available there is a positive water balance, showing that there is water in excess of total irrigation demand.  In both 1996 and 1998 the water balance expressed in mm was in excess of the assumed contribution of soil moisture, so that even if no soil moisture were accounted for, then there would still have been sufficient water to meet total crop needs.  Only in 1997 was the surplus less than the 50 mm assumed to come from the soil. 
RWSsurf are very high, ranging from 1.90 to 2.52.  This means that there was roughly twice as much water delivered than actually required, or a surface irrigation system efficiency of less than 50%.  However, if it is assumed that excess surface water is available to other water users through pumping or reuse of return flows, and the total delivered amount is spread over the whole area estimated to be irrigated, RWStia values drop within the range 1.17 to 1.42.  These values are indicative of better basin-level water management, but they still indicate there is sufficient water for everyone if it can be spread out reasonably.  There may be a few areas with some stress because not all water can be pumped or reused. If the assumption is made that the same volume of water is required to meet crop water needs for the entire season then the water balance and RWS figures change.  For the water balance 1996 becomes negative: demand in June and September was in excess of the additional 50 mm of soil moisture contribution, and the total area was just negative.  However, an RWStia value of 0.95 indicates everything is more or less in balance. Inevitably there will be stress in some locations but it is not a really widespread water shortage. In 1997 and 1998 the water balance is positive for the entire season, although in both years the soil moisture contribution of 100 mm is required to maintain this positive balance. 
Sub-Basin Analysis; A similar trend can be observed for each of the three sub-basins of Alasehir, Main Valley and Delta.  During the irrigation season the water balance is always positive, again demonstrating that water deliveries are more than sufficient to meet not only the areas scheduled to receive irrigation water but also to meet the requirements of the entire irrigable area. RWSsurf values for the irrigation season are always above 1.5 and frequently above 2.0, suggesting not particularly efficient water management practices.  But if water can be used over the entire irrigable area RWStia values drop to between 1.1 and 1.4 for the irrigation season. With the exception of the main valley in 1996 the assumption about soil moisture providing 50 mm of useable water does not affect the positive water balance. For the entire season the water balance is positive for all sub-systems except the Main Valley in 1996.  Because this is the largest irrigable area, the negative water balance here draws the entire basin into deficit.  Even so, the RWStia is still not a great deal below 1.0 (actually 0.91) suggesting that there may be moderate amounts of stress in certain parts of that sub-basin. 
However, when the entire season is considered, the soil moisture assumption of 100 mm is required to keep all of the sub-systems in positive balance.  If no contribution of soil moisture were assumed, then all parts of the basin would show a negative water balance over the full season. One striking factor emerges at the sub-basin level which was not evident at basin level.  With the exception of the Main Valley in 1969, the RWStia values for the full season are all within the range of 1.0 to 1.1.  This indicates two things.  Firstly, the total volume of water released by DSİ into each sub-basin is just about sufficient to meet all irrigation requirements when supplemented by soil moisture use.  However, for operational reasons the water is delivered only during a 60-day portion of the 100-day season. Secondly, it indicates that pumping is an extremely effective way of utilizing excess surface water in a productive manner.  RWS values of 1.1 indicate sub-system water use efficiency of 90%. 
Irrigation Association Level analysis; The most detailed level of analysis possible using these data is the Irrigation Association level. During the irrigation season almost all areas are in surplus.  Three locations show a negative balance in all three years of the study (Sarigol, Ahmetli and Gediz) while other areas in the main valley (Adala Left Bank, Gokkaya, Turgutlu and Sarikiz) show one or two years of deficit.  The Delta systems are always in surplus. During the irrigation season every Irrigation Association has a RWSsurf greater than 1.0 (and the majority have RWSsurf values greater than 2) for the area identified as receiving surface water. For the whole season the pattern changes.  Three systems were in deficit every year (Sarigol, Alasehir Uzum and Gediz), while only two systems were always in surplus (Sarigol Bag and Adala Right Bank).  All other systems were in deficit for one or two of the study years.  The assumption on soil moisture contribution is critical in determining which systems appear to be surplus or deficit ones, because there are relatively few instances where surpluses are more than 100 mm. None of the systems has a RWSsurf less than 1.0 over the entire season so there is more than sufficient water for the surface irrigated areas even if it is not sufficient for the entire irrigated area. Irrigation Association data on RWS and percentage of systems irrigated using groundwater show a slightly different trend than the sub-basin analysis.  Although there is no significant trend, there is a general increase in the area irrigated using groundwater as RWSsurf increases. 

Long Term Trends in Water Scarcity;

Using long term data it is possible to determine the extent to which there is a trend in terms of water scarcity.at basin and sub-basin level.  There are no data available to undertake this analysis at Irrigation Association level. It is possible that demand may have been slightly higher before the drought due to a higher percentage of cotton, but the difference is unlikely to be significant. The results are clear.  Between 1989 and 1994 basin level water supplies were inadequate to meet the total demand but they more or less matched the demand for the surface irrigated area only because the surface irrigated area shrank in response to reduced water supplies.  Given that it is unrealistic to try to distribute water equitably over a basin the size of the Gediz with an RWSsurf value at or close to 1.0 it is inevitable that there were periods of severe crop stress even in the areas irrigated using canal water. The response of farmers was primarily to start to exploit groundwater reserves.  Unfortunately there are no data available on total irrigated area apart from the 1998 LANDSAT images, but it is clear that when water deliveries started to improve again after 1994 farmers did not revert back to surface irrigation. Demand for surface water is estimated to have only increased from 180 to 250 million cubic meters from 1994 to 1998, but available water increased from 150 to over 400 million cubic meters.  It is clear that the water delivered in excess of the demand from surface irrigation areas is being used through pumping to bring cropping intensities up to the 100% level.  This compares to an average of 80% in the period before the drought started. The drought clearly forced farmers to look for and exploit alternative water sources.  Having made the investment in pumps, and discounted the capital cost, they can pump water for more or less the same price as charged by Irrigation Associations (Gecgel, 1999).  They are now much better equipped to avoid being affected by drought than they were in 1989.

Conclusions

It is difficult to classify the Gediz Basin as water short.  Instead it is better to identify three distinct phases of water conditions. Before the drought there was more than enough water but cropping intensities were only in the order of 80% of maximum.  This suggests that there were conveyance or management constraints that prevented full irrigation despite the abundance of water.  There was no significant pumping of water in this period, so the empty land could not be readily irrigated. Between 1989 and 1994 the area under surface irrigation shrank to match available water, with a system-wide RWS of 1.0, indicating severe water shortages in many parts of the surface irrigated system.  This was truly a time of water scarcity because there was insufficient water to irrigate the surface irrigated area, let alone the total irrigable area.  Given what happened after this period, however, the water scarcity can only be classed as temporary.  Indeed, the scarcity was in part due to lack of technology, not due to lack of water: groundwater was available but without pumps it was inaccessible. From 1995 onwards water is not scarce but surface cropping intensities have not increased greatly.  Looking at areas reported as irrigated by DSİ and Irrigation Associations it appears water is insufficient to meet demand because surface cropping intensities are about 60% of the total irrigable area.  But because farmers, with considerable support from DSİ and GDRS, plus very large private investments, have sunk large numbers of both deep and shallow tubewells, most water lost in the surface irrigated areas is reused and cropping intensities are now higher than they were before the drought. 
The water scarcity of the Gediz Basin has been avoided because of the investment in pumping.  Even though surface irrigation efficiencies are low (roughly 50% of water delivered to surface irrigated areas is lost), basin-scale efficiency of the irrigated agriculture is high, ranging from 90-95% in the past three years. The data suggest the current agricultural water management is sustainable because surface water releases are sufficient through a combination of canal irrigation, pumping and reuse to meet the maximum estimated demand from the irrigable area. For the entire period 1970 to 1999 only in 5 years were surface water releases inadequate to meet the potential demand from the entire irrigable area.  Unfortunately these five years were consecutive. It is not clear that this drought period justifies classifying the Gediz basin as water short, particularly as there have been six successive years after the drought when surface water releases have been greater than total demand.  It may be better classified as a basin that is vulnerable to prolonged drought but average data indicate a relatively healthy future for irrigated agriculture.

Acknowledgement

This study was conducted as part of a collaborative research program between the General Directorate of Rural Services (GDRS) in Turkey and the International Water Management Institute (IWMI) and is funded by the Government of Turkey as part of the World Bank assisted Turkish Agricultural Research Plan (TARP).

References

Anonymous, 1998. Salihli Right Bank and Menemen Left Bank Water Users Association records. Unpublist. Salihli-Menemen,Turkey. 
Geçgel, G., Murray-Rust, H., Gül, S., Ray, I. 1999. A Survay of Pump-Based Irrigation in Salihli Right Bank Irrigation Association. GDRS/IWMI Collaborative Research Project, Unpublist, Menemen,Turkey. 
DMİ, 1998. Salihli climatic Data (1998). State Meteorological Services (DMİ). Archive Records, Unpublist, Ankara,Turkey. 
DSİ, 1998. State Hydraulic Work (DSI) Irrigation Associations Annual Records. Unpublist. Manisa,Turkey. 
GDRS, 1998. Menemen climatic data (1998). Research Institute of General Directorate of Rural Services. Archive Records, Menemen,Turkey. 
Molden, D., Sakthivadivel, R., Perry, J.C., Fraiture, C., Kloezen, W.H. 1998. Indicators for comparing Performance of Irrigated Agrıcultural Systems. IIMI Research Paper No:20. Colombo, Sri lanka:International Irrigation Management research Instıtute. 
Murray-Rust, H., Beyazgül, M., Kayam, Y., Girgin, A., Svendsen, M., Brewer, J.1999. Assessment of O&M Tasks of Irrigation Associations. GDRS/IWMI Collaborative Research Project, Unpublist Menemen,Turkey. 
Smith, M, 1991. CROPWAT Irrigation Management Land and Water Development Division Food and Agriculture Organization. Via Terme di Caracalla, Rome, Italy. 
Van Dam, J.C., J. Huygen, J.G. Wesseling, R.A. Feddes, P. Kabat, P.E.V. van Walsum, P. Groenendijk and C..A. van Diepen, 1997. Theory of SWAP version 2.0. Report 71, Department of Water Resources, Wageningen Agricultural University, Wageningen, The Netherlands.