To enhance the accuracy and reliability of the chemical pollutant concentration prediction model, data assimilation technology based on the ?local inverse distance-Kalman fusion algorithm? was introduced for optimization. The local inverse distance algorithm was employed to ?interpolate observation data?, thereby increasing the density of observation points. Subsequently, the Kalman filter was applied for ?data assimilation? between the observed data and simulated data to achieve chemical pollutant concentration prediction. Experimental results demonstrate that: the ?relative error? of pollutant concentration decreased at all seven selected monitoring points after assimilation. Specifically, monitoring point 2 exhibited the ?largest reduction in relative error?, decreasing from 8667% to 4445%. Monitoring point 5 showed the ?smallest reduction?, decreasing from 176% to 171%. These results ?validate the feasibility? of this technique.