Use of combined receptor modeling technique for prediction of possible sources of particulate pollution in Kozhikode, India

PM 10 (particular matter having size 10 µm or lesser) was sampled at three sites in Kozhikode district, India, and was found to vary between 20.83 and 83.8 µg m −3 at Site 1; 20.78 and 99.76 µg m −3 at Site 2; and 45.6 and 123.7 µg m −3 at Site 3. A dual-channel dust sampler was used for sampling. Average concentration (87.21 µg m −3) was found highest at Site 3. Among the species identified, Fe, Ca, SO 4 2− and Na + were the predominant ones in all the sites. A coupled receptor model used for source apportionment was initially tested using synthetic data and then used for experimental values. Sources predicted at Site 1 were paved road dust, marine aerosol, garden waste combustion, vehicular exhaust, and diesel generators, and their percentage contributions obtained were 56, 10.2, 27.2, 0.5, 6.1 from chemical mass balance and 50.8, 11.8, 22.2, 4.5, 10.7 from positive matrix factorization, respectively. In Site 2, fertilizer, vehicular exhaust, LPG combustion, marine aerosol and dust were the main sources with percentage contributions of 43.4, 9.6, 22.3, 0.5, 24.2 from chemical mass balance and 31.4, 24.7, 11.7, 2, 30.2 from positive matrix factorization, respectively. The model suggested four sources at Site 3: diesel generator, incineration, construction and vehicular exhaust with percentage contributions of 80, 0.5, 13.5, 6 from chemical mass balance and 74.8, 1.5, 9.4, 14.2 from positive matrix factorization, respectively. Stringent rules and regulations from the policymakers can curb source emissions to a great extent. Results from the present study can be used as a base to develop the same.