INTERPRETATION OF CARBON MONOXIDE CONCENTRATION AND METEOROLOGICAL PARAMETERS USING SUPPORT VECTOR REGRESSION (SVR) METHOD AND WAVELET TRANSFORM ANALYSIS CASE STUDY OF BANDUNG CITY, MAJALENGKA DISTRICT, AND SLEMAN DISTRICT

• Main Author: Halawa, Erniwati
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/64862
• Institution: Institut Teknologi Bandung

Air pollution is one of the problems in urban areas caused by many human and industrial activities that produce air pollutant substances that cause air quality to decline. In general, the main source of air pollution in urban areas is transportation. One of the air pollutants produced by transportation is carbon monoxide (CO). An understanding of CO concentration is very important because if the concentration exceeds a certain permissible limit, it will hurt human health and the environment. In this study, the support vector regression (SVR) method was used to predict CO concentration by varying the kernel parameters to obtain the best modeling and prediction accuracy. A review of meteorological parameters is very important because it greatly affects the concentration of air pollutants. Thus, in this research, interpretation of the relationship between meteorological parameters is also carried out using wavelet transform analysis, namely continuous wavelet transform (CWT) and wavelet transform coherence (WTC). This scheme is applied to data on temperature, humidity, rainfall, duration of sunlight, and wind speed measured at the Bandung geophysical station, the Kertajati meteorological station, Majalengka, and the Sleman geophysics station. In this study, the interpretation of the influence of meteorological parameters on CO concentrations in the city of Bandung was also carried out. The results of the CO concentration prediction show that the best predictive accuracy value is 97.68% with kernel parameter values ? = 0.02, ? = 30, and C = 0.006. The prediction results show that the predicted CO concentration trend is in line with the actual CO concentration data, indicating that the prediction results are very accurate. The predicted CO concentration showed a decrease, possibly due to the shift from the dry season to the rainy season that occurred in September. This is illustrated in the WTC spectrum between CO concentration and rainfall. The results of the CWT and WTC analysis explain that air temperature and humidity are detected to have a strong intensity from September to December 2019, which is the transition from the dry season to the rainy season with a period of 20-30 for temperature and a period of 5-10 for humidity. At the Kertajati meteorological station, a significant temperature intensity can be seen in the period 170 – 200 in November 2019 – March 2020, occurring during the transition from the rainy season to the dry season. The strong temperature intensity at the Sleman geophysics station is shown in the wavelet transformation spectrum with periods 5 and 10 which occurred in the dry season in July – August 2018. The correlation between temperature and humidity is negative with out-of-phase coherence. Temperature and rainfall have out-of-phase coherence so they are negatively correlated. The greatest intensity took place at the Sleman geophysical station with a period of 128 occurring from August 2019 to June 2020. The temperature coherence with the length of irradiation was in phase and positively correlated. The eleman geophysical station recorded the highest period in July 2018 – June 2020 with a period of 128–256. At the Bandung and Sleman geophysical stations, the correlation between temperature and wind speed is positive, and at the Kertajati meteorological station, it is negative. Air humidity is more influenced by the length of sunlight than rainfall and wind speed. Rainfall has a strong correlation with the length of irradiation compared to wind speed. Rainfall is not in phase with the duration of sunshine and has a negative correlation. The duration of irradiation with wind speed has a positive correlation. The meteorological parameters that most influence the CO concentration are air temperature, humidity, and rainfall. Air temperature is positively correlated with CO concentration in June but negatively correlated from March to April. Air humidity is negatively correlated with CO concentration in June. Rainfall has a negative correlation with CO concentration in April – May which is the dry season. The duration of solar irradiation and wind speed only at some time of observation showed a negative and insignificant correlation. This study states that the increase in CO concentration is not only influenced by one parameter but the collaboration of several meteorological parameters and several other factors.