Sengupta, M. and Dalwani, R. (Editors). 2008 Proceedings of Taal2007: The 12 th World Lake Conference: 851-856 Phosphorous Modelling of Upper Bhopal Lake Using Best Subset Procedure M K Sharma*, V K Choubey and V K Dwivedi National Institute of Hydrology, Roorkee-246 67, Uttarakhand, India *Corresponding author: E-mail: mks@nih.ernet.in ABSTRACT The Upper Bhopal Lake, the largest fresh water lake in M.P., India, is located in the heart of Bhopal town. The lake catchment comprises of a myriad of interconnecting streams and several sub-basins. The upstream portion of the lake is shallow. The downstream portion of the lake is the deepest portion of the lake bounded by urban population all around the lake. It is a typical urban lake, mainly used for providing drinking water for the mammoth population of old city of Bhopal. A number of ephemeral water channels enter the lake from the human settlements around the lake and discharge polluted water in the lake bringing several chemical as well as sediment load in the lake. Eutrophication is one of the major problems in some parts of the lake. Water quality may become so degraded that the lake’s original uses are lost. It may no longer be used for bathing, swimming or fishing purposes. Phosphorous is the limiting nutrient responsible for eutrophication of the lake. In the present study, Best subset procedure based on R 2 and F values is used in model dissemination for predicting the phosphorous concentration. Regression models are developed in explaining variation in phosphorous concentration using routinely measured parameters. The predicted values of phosphorous are compared with observed values and reasonably good matching are obtained. Further, efficiency of the developed model was tested by error estimation using correlation statistics, standard error and mean multiplicative error approaches. The developed model may be used for prediction of missing observed value of phosphorous concentration. Keywords: Water quality parameters, best subset, regression, phosphorous modelling, Upper Bhopal lake INTRODUCTION Lakes and reservoirs provide for the volumetric storage of watershed inflow and may have long flushing time periods as compared with the rapid inflow and outflow for a reach in a flowing stream. Therefore they are of special concern with regard to the retention of certain constituents, especially nutrients including forms of nitrogen and phosphorous which promote eutrophication. Eutrophication is a natural ageing process in which the water content becomes organically enriched, leading to the domination of undesirable aquatic growth, such as algae, water hyacinth and so on. The eutrophic process tends to decrease aquatic life and has detrimental dissolved oxygen effects. The problem of lake eutrophication can be controlled by limiting the phosphorous input to the lake and other management interventions, such as changes in depth of a lake or increases in the rate at which phosphorous is lost. Thus the concentration of phosphorous plays a vital role in controlling the eutrophication of the lake. Both modeling and monitoring are useful in attempting to ascertain a lake's response to phosphorous loading from development. Monitoring provide present conditions but is expensive, time consuming as lakes exhibit high variability in terms of algal response and nutrient concentrations on both an annual and seasonal basis. Therefore there is a need to provide a cost-effective alternative which can help to protect the water quality of lakes. Modelling provides such an alternative to interpretate data with prediction for the future. Wang et al. (2003 a,b) developed a model to study the phosphorous dynamics in aquatic sediments and to conduct dynamics predictions of phosphorous release across a sediment-water interface. The model focuses on the sediment active layer below the sediment-water interface and is based on primary mechanisms regulating phosphorous behaviour in sediments. The developed model was applied to Chesapeake Bay, USA to examine model performance in predicting sediment-water phosphorous exchange and found satisfactory results. Pettersson (2001) studied the proportions of phosphorous forms as well as total phosphorous content in suspended and settling particles during spring, summer and autumn in order to improve the understanding of particle composition and mineralization processes in moderately eutrophic, temperate lakes with summer stratification. However, vast amount of data requirement (rainfall, soil type, land use etc.) places several constraints on the application of these models in Indian conditions. The statistical approach provides a viable alternative over other conventional models and gives