a fuzzy model in forecasting dissolved oxygen in rivers

amount of dissolve oxygen in a river has a great direct affect on aquatic macroinvertebrates and this would influence on the region ecosystem indirectly. in this paper it is tried to predict dissolved oxygen in rivers by employing an easy fuzzy logic modeling, wang mendel method. this model just uses previous records to estimate upcoming values. for this purpose daily and hourly records of eight stations in au sable watershed in michigan, united states are employed for 12 years and 50 days period respectively. calculations indicate that for long period prediction it is better to increase input intervals. but for filling missed data it is advisable to decrease the interval. increasing partitioning of input and output features influence a little on accuracy but make the model too time consuming. increment in number of input data also act like number of partitioning. large amount of train data does not modify accuracy essentially, so, an optimum training length should be selected.