Nilüfer Çayı, Uludağ'ın kuzey ve batısından gelen sulan tahliye eder. Ana kol, dağdan Bursa'nın batısına doğru akar, sonra doğuya doğru kıvrılır; daha sonra tekrar geri dönerek Bursa'nın 50 km. batısında Simav Çayı ile birleşir. Bursa 'nın kuzeyin de, ana yolların bulunduğu kısımda Nilüfer Çayı, havzanın doğusundan gelen sulan tahliye eden sulama kanalı ile birleşir. Nehir Bursa'nın mansabında, yoğun tarımsal etkinliğe sahip bir ovadan geçer. Bursa şehri ve çevresinde kurulu bulunan endüstriler kuruluşları itibariyle dağınıklık göstermektedir. Endüstri sayısının çok fazla olması ve dağınık bir şekilde yerleşmeleri nedeniyle, endüstriyel yüklerle ilgili olarak kapsamlı bir çalışma yapılamamıştır. Endüstriyel atıksuların yerleri ve yükleri hakkında, 1989 yılında yapılmış `Bursa Kirlilik Envanter Çalışması `ndan yararlanılmıştır. Bu çalışma, Bursa şehri boyunca 335 endüstriyi kapsamaktadır. Bu çalışmada Bursa şehri içindeki atıksular, biri evsel, diğeri endüstriyel nitelikli atıksuların yoğunlukta olduğu iki bölgeye ayrılmıştır. Nilüfer Çayı'nın hidrolojisi ve su kalitesi ile ilgili bilgiler, 1979 yılından beri bu çay üzerinde gözlemlerini sürdüren Devlet Su İşleri (D.S.İ.)'den alınmıştır. Ayrıca nehir akımıyla ilgili olarak D.S.İ. ve E.İ.E. (Elektrik İşleri Etüd İdaresi)'nin akım gözlemlerinden yararlanılmıştır. Eldeki hidrolojik ve su kalitesi bilgilerinin yetersizliği nedeniyle karmaşık bir model yerine BOI(Biyokimyasal Oksijen İhtiyacı) ve ÇO(Çözünmüş Oksijen)'in simüle edildiği, tek boyutlu kararlı hali gözönüne alan bir model kullanılmıştır. Halen Nilüfer Çayı'na özellikle evsel atıksular nedeniyle önemli miktarda azot yükü gelmektedir. Bu nedenle mevcut kirliliğin belirlenmesi için NH3-N'unun simüle edilmesinde yarar vardır. Kullanılan model BOI ve ÇO'in yanısıra NH3-N'unu da modellemektedir. Model hem aerobik hem de anaerobik şartlar altında, noktasal ve/veya yayılı atık yükleri ile kirlenen kollu bir nehir sisteminde, su kalitesi üzerinde meydana gelen BOI, ÇO ve NH3-N'u konsantrasyonlarının değişimlerini modellemektedir. Model de, organik maddenin biyokimyasal ayrışması sonucu BOI'nin indirgenmesi, hava dan oksijen transferi sonucu ÇO'in artması, çökelme ile BOI'nin indirgenmesi, dip tabakalarından su ortamına BOI ilavesi, fotosentez ve solunum olaylan neticesi ÇO'de- ki değişim ve NH3-N'unun indirgenmesi mekanizmalar hesaba katılmaktadır. Kalibrasyonda 1987 yıllık ortalama BOİ, ÇO ve NH3-N'u değerleri kullanılmış tır. Modelin verifikasyonu aynı parametrelerin 1988 yıllık ortalama değerlerine göre yapılmıştır. Atıksuların arıtılması halinde su kalitesinin nasıl bir değişim göstereceği belirlenmeye çalışılmıştır. Mathematical models of dissolved oxygen have been used for water quality ma nagement for many years since first proposed by Streeter and Phelps. The models have been used to assess the assimilation capacity of rivers in regard to discharges from Water Pollution Control Plant (WPCP) and to evaluate various management options of pollution control strategies for improving the water quality. The models have been continuously improved to be more representative of the real world including more independent or state variables that represent the relevant physical and chemical processes found in nature. The first 40 years of river mode ling involved two linear differential equations of biological oxygen demand (BOD) and dissolved oxygen (DO) applied one - dimensionally to rivers and estuaries. This evolved into two - dimensional models with additional state variables to include the effects of different nitrogen compounds. Later, the phosphorus cycle and the effect of phytoplankton were included. The recent advances included three - dimensional models involving carnivores, zoop- lankton, phytoplankton and toxicants in nonlinear interactive `compartments`. These models can involve more than 20 state variables. There is always some error or uncertainty in a model. A mathematical model cannot represent the real process perfectly : either there is some unknown process involved or some part of the representation cannot be calculated due to complexity or economics. The uncertainly has usually been attributed to the randomness inherent in natural processes. This variability has usually focused on the rate constants or pa rameters representing the different processes being modelled. Initial estimates of ra te constants are modified until the model reproduces field measurement data (calibra tion). There is always some error or uncertainty that cannot be eliminated. Mathematical models of water systems allows the water planner to evaluate the effects of various treatment schemes on water quality to predict river conditions du ring periods of extreme climatological conditions. Inherent in these models is a de tailed consideration of the numerous elements defining the hydraulic and water qua lity characteristic of drainage basins. Many of the models in use are extensions of two simple equations proposed by Streeter and Phelps in 1925 for predicting the biochemical oxygen demand(BOD) of various biodegradable constituents, and the resulting dissolved oxygen consentration XI(DO) in rivers. Often used with these BOD-DO models are other fairly simple first- order exponential decay, dilution and sedimentation models for additional noncon- servative and conservative substances. More complex multiconstituent water quality models have also been proposed and have been applied to predict the physical, chemical and biological interactions of many constituents and organizms found in natural water bodies. These multiconsti tuent simulation models generally require more data and computer time, but they al so can provide much more detailed and comprehensive information on the quality of water resulting from various water and land management policies. Water quality models can be used to evaluate steady state conditions, for which the values of the water quality and quantity variables do not change with time. They can be used to evaluate dynamic or time-varying conditions. The latter type of mo del permits an evaluation of transient phenomena such as nonpoint stormwater runoff of spills of pollutants. Steady state models are usually simpler and require less com putational effort than dynamic or transient models and are more relevant to long-term planning than to short-term management and control. Assumptions pertaining to the mixing of pollutants in water bodies dictate the spatial dimensions of the model. Sufficient accuracy may be obtained in many river systems by modelling only one or two dimensions. One dimensional models of river systems assume complete vertical and lateral mixing. The catchment of the Nilüfer River drains the nothhern and western slopes of the Uludağ mountain. The main river flows out of the mountains to the west of Bursa, before turning eastwards along the foot of the steep mountain slopes; the river then turns back on itself, before flowing into the Simav River some 50 km. to the west of Bursa. Just north of Bursa, near the point where the main road to the north crosses the river, the Nilüfer is joined by a network of irrigation drainage canals that collect water from the eastern part of the basin. There are many small creeks and torrents which feed the Nilüfer along its cour ses through the urban area. Some of these are fed by springs that have been used for water supply over many years ; other watercourses flow intermittently after rainfall, or in the spring when they are fed by snowmelt from the mountains. Downstream of Bursa, the river meanders through a flat alluvial plain to the east and north of the city, where there is extensive development. Some of the area is supplied with irrigation from groundwater. At Bursa almost 70 percent of the annual rainfall falls in the wet season from December to June ; at higher altitudes the per centage of winter rainfall approaches 80 percent of the annual total. The hydrologi- cal datas are held at the Bursa office of DSI and other data are available from publis hed yearbooks. Quality control of the data was not be possible, which were accepted in published form. In the study area, the main DSI gauges are located in the upper parts of the catc hments. Since the main river was modelled, the DSI flow gauges have not been suffi cient. The EIE gauge at Gecit is used for the downstream computations. The Gecit 2 has a catchment area of 1 290 km, out of the total catchment to the tributary with the 2 Simav of over 1900 km. It appears that there are no stream flow gauges downstream of Gecit, either on the Nilüfer itself, or on tributaries. Maintaining good gauges on xnthe downstream reaches of the Nilüfer is difficult for several reasons. Firstly, the bed is very muddy and unstable, so it is difficult to obtain accurate crossectional measu rements for the derivation of rating curves. Secondly the river water is considered to be highly polluted and there are dangers associated with current metering. The E1E gauge at Gecit was the main source of streamflow data for the study. DSI initiated a programme of water quality monitoring at locations in and aro und Bursa in 1984. The sampling points fall into two categories ; points upstream of the Doğancı Dam which are used to monitor water flowing into the reservoirs, and points downstream of the dam, located throughout the urban and agricultural area. At first the sampling frequency was about six times a year, but this has fallen in recent years. A range of chemical and biological determinants are measured ; flows are measured at the same time as the samples are taken. So these flow measurements were used at uppermost reaches. The study demonstrates that the increasing pollution of surface watercourses as they pass from their upper catchments through the urban and agricultural areas befo re joining the main river within the plain to the north of Bursa. The sampling points of the Nilüfer river are divided into three groups : mounta in sample points, upstream sample points, and downstream sample points. The ups tream sample points are used by DSI to monitor the quality of reservoir inflows. The data indicate that these mountain watercourses are relatively unpolluted, with high dissolved oxygen (DO) and low biochemical oxygen demand (BOD). However some high E.Coli measurements indicate that domestic pollution from upstream vil lages can sometimes be a problem. Sampling points in upstream group are at a lower altitude, but are all upstream of the main urban area of Bursa. Nevertheless there is a decline in water quality, parti cularly in the dry season when dilution is small. BOD is higher and there are samp les with high E.Coli counts. The sampling points in the third group are further downstream where rivers, channels and drains flow out of the urban area. The effects of untreated outfalls are clear; dissolved oxygen is reduced and in some cases falls to zero. BOD and chlori de have increased in general, all the sample points indicate gross pollution from se wage outfalls upstream. The quality of these surface water courses does change over the seasons. Hig hest concentrations of BOD and lowest DO levels tend to occur at the end of the dry season (October/November) through to January/February. During this period flows are initially low so that there is little dilution of the effluent discharges. Higher rain fall results in increased flows but this causes residual pollution in many of the dry water courses to be flushed into the system. DO concentrations tend to decreas when moving downstream, together with the zero DO concentrations. It should be remembered that only spot samples are taken, at a frequency of every two months or so. It is likely that river water quality will deteriorate dramati cally within the space of just a few hours when particular industrial processes are operating and discharging effluents at a high rate. Such pollution events could be completely missed by the present sampling procedures. Consequently, it is likely that extremely poor water quality can occur at any time of the year. xmAt present many of the surface waters downstream of Doğancı Dam would fall into the Class IV or very polluted category. A first stage objective would be to redu ce pollution from the major outfalls so the reaches of the river can be assigned to a less polluted class. During the analyzes öf water quality datas, it is observed that Nilüfer River has lost its ecologic characteristics and has become an open channel. At microbiologi cal analyses it is found that some coliform bacteria levels are even at billion levels. Therefore, it is definite that usage of these waters for agricultural purposes will cause very serious results for human health which can not be recovered. In this study, the present water quality of the Nilüfer River has been modelled and the solution alternatives have been investigated The model is a steady-state one- dimensionel water quality model which was developed by Wu J. S. and Ahlert R. C. [9]. It has the capability of handeling river system and computing BOD- NH3-N-DO profile under both aerobic and anaerobic conditions. Firstly, a diagram of the system to be analzed was drawn. Secondly the reaches were numbered starting with the up permost and carefully determined the values of the control variables. The model has three control variables that can be defined below: NCON 0 uppermost reach 1 all other reaches, except 2 encountered a branch 3 end segment of a branch joining main river NSTOP 0 for last downstream reach 1 for uppermost reach joined by a branch 2 a non-branched reach joining main river 3 otherwise NCASE 1 no waste input 2 point source input 3 uniform loads input 4 point and uniform loads The river system is divided in to the reaches which have the same physical, bio logical and hydrological conditions. The reaction constants do not change in the sa me reach. Kd, Ka, Ks, K^, B and PR are used in the model as the reaction constants. But B and PR are neglected in the model. In this study, 1987 mean water quality measurements are used for calibration and 1988 mean water quality measurements are used for verification. The model inclu des biochemical oxydation of organic substances, reduction of dissolved oxygen by this oxidation, elevation of dissolved oxygen by reaeration process, removing of BOD by sedimentation, benthic oxygen demand and photosynthesis/respiration ef fects. The model can simulate the BOD, DO and NH3-N concentration in a river un der both aerobic and anaerobic conditions. xivThe development of water quality simulation models requires an understanding of the external forces acting on the stream system. These inputs to the system are waste loads, effluents, water withdrawals, tributaries or all of these of the main stre am system. Detailed investigations of these inputs were not possible in the study and `Bursa Pollution Inventory Study` was used. So it should be remembered that this study could not characterize the real system because of the insufficient flow data. But it can be usefull for understanding the pollution level of the stream system and it can be a base for better investigations. xv 125