Water quality modeling in the Nam Pong River, Northeast Thailand

Poor water quality which caused massive fish kills in the Nam Pong River in 1999 was associated with low dissolved oxygen (DO). A dynamic water quality model was thus constructed using data collected for 2 years from 1999-2000 to predict whether there was an algal bloom which subsequently died off and caused low DO and fish kills on the same day in 1999. Flow and runoff were calibrated by using lignin and tannin (LT) as a conservative trace. Root mean square error (RMSE) of the flow calibration with LT was comparable with literature values, using salinity as a conservative trace. Results of correlation coefficients (R²) from the runoff calibration were reasonable (0.64-0.75 for 1999 and 0.62-0.88 for 2000). RMSE values from the model calibration and validation of conventional nutrients were found to be comparable to literature values for steady-state models. The predictive capability for chlorophyll a showed that, with the flow calibrated from LT, the bloom was not overestimated. The model predicted the bloom die-off which lowered DO and possibly caused fish kills on the same day in 1999, and suggested that the accuracy of the dynamic model was on a time scale of days, not seasons like most steady-state models.