GIS and MATLAB modeling of criteria pollutants: a study of lower onitsha basin during rains

The study of air pollutants SO 2 , NO 2 and PM 10 in lower Onitsha basin, a densely populated city was performed using GPS and Matlab modeling. The pollutants were studied in nine specific locations for 3 months of rains over 3 consecutive years with each georeferenced. The Matlab pollution model was generated by integrating the spatial database and measured pollution attributes database using a polynomial expression. SO 2 highest concentration (141 µg/m 3 ) peaked in Upper Iweka at sampling point 1 before dispersing to lower concentrated regions in Awada and Resthouse. NO 2 peaked at 207 µg/m 3 in Upper Iweka at sampling point 3 and driven by wind towards Borromeo area to very low concentration of 38 ug/m3. The PM 10 peaked in Upper Iweka (180 µg/m 3 ) and driven by rains towards Borromeo before increasing again in concentration levels at Awada. The AQI showed that SO 2 pollutants had acceptable air quality at all sampling points while NO 2 and PM 10 air quality affected sensitive groups. SO 2 concentration levels exceeded

artificial neural network assemblage (ANN), for outdoor air quality. The modeling provided reliable predictions for NO2 but unreliable data for PM2.5 (Challoner et al. 2015). Another recent study in Baoding, China effectively used MATLAB grey model for near accurate prediction of ambient air quality (Ying et al. 2017). Similarly, the use of mapping to isolate and pinpoint certain air pollutants has seen more accurate pollution maps generated for densely populated cities (Rohde and Muller, 2015). While, GPS-assisted data collection has also enabled the successful assessment of vehicular exhaust emissions linked with activity travel-based data for assessment (Beckx et al. 2010). Additionally, GIS, GPS and sensors have measured air pollutants (CO, SO2, NO2) from vehicular emissions with greater accuracy for regulatory decisions ). Hence, air quality monitoring will continue to play a major role using GPS, GIS, MATLAB, Artificial Neural Network (ANN) and internet coupled devices due to seasonal variations (Challoner et al. 2015; Wei et al. 2015), tremendous increase of vehicular emissions ) and the need for real time pollution assessment (Jiayu et al. 2018). Moreover, nations are gradually moving towards reliance on published air quality monitoring data. Subsequently, researchers adapted a photochemical monitoring station for ozone monitoring in Mexico City with successful results (Palomera et al. 2016). In Jaipur city, India, studies performed on spatio-temporal analysis to evaluate relationship between air quality and local weather parameters provided a quick view of criteria areas within the city (Ankita et al. 2017). While a GIS personal and population exposure to PM10 and PM2.5 have been studied in Dublin and Beijing respectively by Pilla and Broderick (2015); Zhao et al. (2017). In the USA, Los Angeles has gone a step further by utilizing longwave infrared hyperspectral imaging sensor (LWIR-HSI) to tracking and quantifying gaseous chemical plumes over a 530 km 2 region. The results supported routine regulatory activities and also has the capacity to provide identification and monitoring after environmental hazard occurrence (Buckland et al. 2017). Thus, spatial processing and time series analysis are necessary to test local compliance to standards, study environmental impact of new industries or emission changes associated with traffic and vehicular movement (Puliafito et al. 2003). To this end, researchers in a nation like Nigeria with low level of industrial compliance to emission standards, poor regulatory monitoring, excessive harmful vehicular emission (Nkwoada et al. 2016) would certainly need to key into this well-established area of atmospheric pollutant monitoring. This will chart the course for national ambient air quality compliance data as evidence for frequent regulatory monitoring, sanctions, fines and penalties were necessary. However, researchers have evaluated air pollutants in Niger delta, Nigeria using remotely sensed satellite. The result confirmed that the area needs urgent environmental remediation (Omotosho et al. 2015). Similarly, Yorkor et al. (2017) performed a study in Eleme, Port-Harcourt city; Nigeria using ANN attributed the pollutants to be vehicular and industrial source pollutions. On the other hand, a study of air pollutants by Balogun and Orimoogunje (2015) in Benin City, Nigeria, concluded that seasonal variation is a determinant factor to concentration of pollutants in the city. Comparatively, Onitsha is the commercial hub of southeast Nigeria and a densely populated city like Benin City. The city boast of similar heavy vehicular activities, more manufacturing industries, but deficient of environmental regulatory and compliance officers. Thus, this study will utilize GPS and MATLAB modeling to study for the first time SO2, NO2, PM10 pollutants in lower Onitsha basin during the rains.

II. Materials and Methods
The materials and method therein described the GIS/GPS and MATLAB assisted study of pollutants concentration densities in Onitsha lower basin. New model scripts were designed and applied into the General finite line model with ARCGIS 9.3; this improved the modeling approach and less time spent on GIS workflows. The software ARCGIS 9.3 was used to create specific scripts through workflow coding and commands in successions. Hence, the tool can be applied by the user for various adaptive studies such as a tool for calculating concentration of SO2, NO2, PM10. The correlation was integrated to also evaluate peaks in relative humidity, wind speed and wind direction. The determined pollutant concentration levels were fitted into ARCGIS to determine the total concentration levels of the named pollutants at specific locations and measuring times. MATLAB 7.9 fitting software was used for plotting the graph of weighted coordinates against the mean concentrations in each location in Onitsha lower basin (Pilla and Broderick, 2015;Yorkor et al. 2017).

Data acquisition:
The acquisition of data was achieved by in-situ ground level measurement of SO2, NO2, PM10 in Onitsha study areas. Within each sampling station 4 points were selected which were 500 meters apart. The points were marked and georeferenced. The obtained readings were carried out for 3 months with 72 hourly interval ranging from May 1 st to July 1 st which constitute rainy season peak period in Nigeria. Therefore, for the Onitsha study area with selected nine (9) sampling stations there are 6 x 9 experimental units. Experimental units mean six parameters to be tested x nine locations. Readings were taken at 3 months of rainy season which resulted to 54 experimental units. Each station has 4 points for sampling so that the Onitsha selected area has a total of two hundred and sixteen (216) determinations. Hence, in Onitsha there were obtained 1296 experimental units. This amounted to 5184 data for the rainy reason over a period of 3 years from 2013 to 2016 (Pilla and Broderick, 2015;Yorkor et al. 2017). A clear distribution of sampling locations was illustrated in Figure 01 (Google, 2018). Equipment and calibration: Gas and Particulate monitors was carried out using Crown on gas monitor Model CE 89/336/EEC obtained from the Imo State Environmental Protection Agency. The equipment was used for NO2, SO2 and PM10. PM10was monitored by switching to Crown on Particulate Monitor Model No. 1000 with serial no. 298621. The Wind speed and direction were determined as windrose using a digital meter. Relative humidity and temperature were determined with the same Environmental Meter Model AE.09605 by Rumsey Environmental LLC, from Mechanical Engineering Department; Federal University of Technology, Owerri which is located in the Weather/Erosion monitoring unit in the institution. The sensors were recalibrated and stabilized by exposure for several hours in a sealed vessel at room temperature prior to measurement. NO2 and SO2gas were used as applicable. NO2 gas concentrations ranging from 0-1000 ppb was used to recalibrate the SnO2 sensor connected to evaluation circuit board which records sensor responses. This calibration was repeated for SO2 gas sensor. The protocol for recalibration of particulate matter at PM10 to establish the sensitivity was applied and recorded. Recalibration of the monitors were done at Imo State Environmental Protection Agency

MATLAB assisted modeling:
The pollution characteristics (model) of the study area was generated by integrating the spatial data base and measured pollution attributes data base using the polynomial expression (Raju et  yj= k + k1x1 + k2x2 +k3x3 … +knxn Where, yj represents the coordinates for points 1, 2, 3, 4, in each location which constitutes the spatial data base, the pollution index at any given sampling station can be represented by a function y which depends on the contributions of the various concentrations of the identified pollutants, the windrose and the meteorological conditions such as relative humidity, temperature etc. So that at a given sample station with four sampling points, four simultaneous equations can be written to represent the air pollution index at that station. where y1 = Pollution index at a given coordinate such as point 1, k is an empirical constant k1, k2, k3, are constants which modify the empirical pollutant concentrations and are the constants for the variables SO2, NO2 and PM10 respectively. The application of matrix algebra was used to solve the set of the simultaneous (  Where, G is the variable that outputs the inverse of the matrix X.

Air Quality Index (AQI):
The air quality index (AQI) is an index system of number grading that indicates the level of pollution in the atmosphere. AQI determination is carried out by calculating the IAQI (Individual air quality index) for each pollutant. Where the formula is given below as The IAQIP is the individual air quality index for pollutants P (PM10, SO2, NO2) and CP is the daily mean concentration of the pollutant P. BPLO and BPHI are the nearest and lowest values of Cp The IAQILO and IAQIHI are the individual air quality indexes in terms of BPHI and BPLO as shown in table 04. From the table 04, the IAQI maximum is 500. After the calculation of individual air quality index (IAQIP) for each pollutant, the AQI would then be determined by selecting the maximum IAQIP as follows: Hence, the equation (4) demonstrates that AQI calculation is not the sum of all the pollutants involved but is the maximum value of IAQI obtained. Although NAAQS-2012, PM10, SO2, NO2 are included in the calculation, however, the air pollutant with a maximum IAQI when AQI is larger than 50 is then termed the principal pollutant. Whereas daily AQI less than 100 is supposed to be qualified using NAAQS

SO2 concentration
The concentrations of the SO2 gas was plotted using the box and whiskers chat as seen in figure 02 below.   . The appearance of the figure seems to connote a kind of undulating wind driving the NO2 molecules. These would have cause the molecules to experience dispersion at Upper Iweka and Awada. While aggregation is experienced at other sampling points. CKC and Borromeo experienced the lowest wind action, hence the highest level of aggregation. This idea is supported by the fact that at upper Iweka was high concentration of NO2 whose concentration will reduce as it travels within the wind. These molecules will eventually get dispersed to other regions at small concentration. However, on the converse, the CKC and Borromeo and PH road may be point source pollutions due to agglomeration and low wind action.  The PM10 results obtained were between 70 -190 ug/m 3 and even a similarly lower concentration levels were seen when compared with study carried out in Port Harcourt during rains with maximum in Oginigba as 34 ug/m 3 and minimum in Omuanwa as 2.9 ug/m 3 . These showed a correlation that PM10 is affected by rains which causes a decrease in its concentration levels. hence the scavenging effect of atmosphere is paramount which dissolves the PM10 pollutants (Akinfolarin et al. 2017). However, the values of studied Onitsha lower basin exceeded the concentration in Imo state (Opara et al. 2016), Aba and Orlu.

Air Quality Index
The calculated AQI for the individual pollutants where plotted in figure 05 below. The bar columns were presented in respective colour codes shown in table 04. A closer look and comparison from table 04 shows the simplicity in using colour codes in representing the bar columns for AQI values. The SO2 AQI (yellow code) plot showed that only Rest house, C.K.C and Upper Iweka had AQI greater than 50 hence, a moderate health concern for unusually sensitive people within those areas. However, the SO2 air quality was satisfactory at all other sampling areas as indicated by their green bar columns. The AQI for NO2 at Upper Iweka and Awada demonstrated acceptable air quality but presents possible health effects to members of sensitive group. This is demonstrated by the Orange and yellow codes and several points exceeding AQI 50 value. The other sampling areas had yellow codes (> 50) that indicated moderate health concern to sensitive group.

Effect of meteorological parameters
The effect of meteorological parameters on the studied variables were plotted in the figure 06 below. The diagram showed that the wind speed was well below 10ms -1 and had little or no effect on the SO2, NO2, and PM10 determinants. The average plot of NO2 and SO2 were almost superimposable at each sampling point. The highest NO2 was at CKC at 155ug/m 3 while SO2 at CKC was 57ug/m3. Hence, this may be emanating from a similar source. On the other hand, the PM10 was highest at upper Iweka (159 ug/m 3 ) but adjusted to similar undulating movement at CKC, Borromeo and Awada. This similarity is a pointer that the gaseous pollutants were affected by similar metrological parameters. This pollutant movement was seen as an active undulating wind or slopping region from a point source as found in SO2 concentration in section 4.1, or low wind action from point source as found in NO2 section 4.2 or active points of dispersion by PM10 in section 4.3. Consequently, almost remained equal except at Borromeo and Awada. This high humidity may also be a contributory factor to low levels of gaseous pollutants at Borromeo. However, it was observed that Awada had high humidity but experienced greater wind director. This wind direction may be the force driving higher levels of PM10 and NO2 at Awada sampling point. Hence, the results suggest that SO2, NO2, PM10 dispersal are significantly affected by wind direction and humidity. In addition, it would be noted that PM10 was the least affected by meteorological parameters. In addition, it also demonstrated the importance of rainfall in the scavenging of SO2, NO2, and PM10 criteria air pollutants (Ravindra et al. 2003).  (Said et al. 2016). Thus, the determined concentrations were below WHO standards as similarly described using the NAQS of Nigeria (50 µg/m 3 ).

Data application and implication
The major sources of SO2 pollutant are combustion power plants, fossil fuel and petroleum refining (Popp, 2006). Consequently, upper Iweka, CKC and Fegge/Nupe should be areas of local monitoring concerns for SO2 because these emitted gases will eventually form acid rain due to its low acidic pH value and affect the environment through corrosion of materials, damage to crops and forests, nutrient leaching and contaminating drinking water. Additionally, since SO2 residence time is 2 to 4 days and its main dispersion is by oxidation (Griffin, 2006). The enforcement of standards and map transport within areas of upper Iweka, CKC and Fegge/Nupe and gas transport are required for SO2 pollutant control in Onitsha Lower basin.

NO2 Pollutant:
The maximum values of NO2 determined was 109 µg/m 3 at Upper Iweka, while the minimum value was measured at Borromeo to be 19 µg/m 3 . The values when compared with FEPA (stationary sources) and National air quality standard (NAQS) (ambient limit) for NO2 is 75000 µg/m3 and 1000 µg/m3 respectively (FEPA, 1991). Thus, the levels of NO2 at all sampling points were below Nigeria standard and accordingly, all areas have good air quality with respect to NO2. Additionally, WHO 1h mean is 200 ug/m3, while annual mean period is 40 µg/m 3 (Said et al. 2016) Subsequently, the average concentration of NO2 for all sampling points exceeded 40 µg/m 3 except at Borromeo with 28 µg/m 3 .
Since the average level of NO2 exceeded WHO mean annual level, there is therefore the possibility of reacting with water to form acid rain in all sampled regions. These would lead to material corrosion and damage to crops. Moreover, the NO2 residence time is 2-5 days, while principal sinks occur through oxidation, deposition, photolysis and dissolution in oceans and surface waters. Thus, if the major sources were fossil fuel combustion driving the NO2fluxes, there is the need for monitoring trans-boundary fluxes from stationary and mobile sources. Also, the use of selective non-catalyst reduction technology in such combustion processes will further reduce the release of nitrogen oxides (Popp, 2006;Griffin, 2006).

PM10 Pollutants:
The maximum determined PM10 concentration was 111 µg/m 3 determined at both Upper Iweka and pH road. The minimum determined concentration was at Borromeo at 58 µg/m 3 . Their comparison to Nigeria, FEPA (stationary sources) and National air quality standard (NAQS) (ambient limit) for PM10 is 25000 µg/m 3 and 150 µg/m 3 respectively. Accordingly, the concentration levels were lower than FEPA and NAQs standards. Also, this illustrated the inability of FEPA at such high standard (25000 ug/m3) to effectively evaluate criteria pollutants that are noxious even at low concentrations. The WHO standard for PM10 is 20 µg/m 3 for 24h period and annual mean of 50 µg/m 3 (FEPA, 1991). Thus, the sampled area exceeded the WHO standard with respect to PM10. But since PM10 has particle size < 10um, they are mostly released by combustion of fossil fuels, motor vehicles, agricultural burning and industrial activities.
Such activities can prevent suns radiation from reaching the earth when they act as cloud nuclei. they effect is reduced visibility, depletion of soil nutrients, acidification of surface water and destruction of sensitive ecological forests and farm crops. Consequently, there should be controls for industrial facilities, motor vehicles and use of cleaner burning gasoline and diesel fuels (Akinfolarin et al. 2017;Opara et al. 2016;Popp, 2006;Griffin, 2006).

GIS plot
The GIS plot was shown in figure 07 below. These provided a better and elaborate description of the concentrations of the pollutant relative to their respective positions. The highest concentration of SO2 (A) can be clearly seen to be Upper Iweka and Awada and was similarly observed in NO2 plot (B). Borromeo and Mission road/waterside were the lowest points of concentration for NO2 and SO2 respectively. The contours showed that that pollutants spreads from Awada and Upper Iweka to other sampling regions. Three or more central forces were active in SO2, while about two could be seen in NO2. This demonstrated that SO2 is more quickly dispersed than NO2. A similar dispersion movement was observed in PM10 with Upper Iweka showing highest concentration and spreading towards CKC. One central force was observed in PM10 and hence, dispersed at a speed slower than NO2. Summarily the SO2 was the most active gaseous pollutant and more easily dispersed by metrological forces (wind speed, wind direction, humidity) than NO2 and PM10. These showed a correlation with section 4.5 that the studied pollutants were affected by metrological factors in the order of SO2> NO2> PM10.

MATLAB results
The Matlab assisted plot for PM10 was significant and hence shown in figure 08 below. The Upper Iweka had been described as an area of active pollutants. The Matlab windspeed showed that there was an upward wind in upper Iweka driving the pollutants towards dispersal at some other region. The wind action appears forceful and later a still blowing wind which aided the movement of pollutants. The other Matlab surface plots showed that mission road generally experienced similar concentrations levels of PM10 at most sampling points. The Awada and Borromeo were previously described as lower regions having lower PM10 concentrations. Hence, while Awada would peak at some point before dispersing, the Borromeo region would vary a little in its concentration level with upward increase in concentration when descending/dispersing to other regions. Both the abrupt movement and change in concentration experienced was initially demonstrated in figure 06. Consequently, both the Matlab and GIS plots confirmed that during rains, metrological factors act as scavengers to air pollutants.

IV. Conclusion
The study of air pollutants utilizing spatial processing and time series analysis has seen accurate results for regulatory purposes in densely populated cities. GPS and MATLAB modeling utilized in this study for SO2, NO2 and PM10 pollutants during rains in Lower Onitsha basin. The named pollutants were studied at different measuring times in nine locations for 3 months of rains in three consecutive years. A Matlab model was generated by polynomial equations while GIS coordinates were mapped using ARCGIS 9.3. All the 3 pollutants showed highest concentrations at Upper Iweka and dispersed towards Awada, Borromeo or Rest house. AQI showed that PM10 and NO2 may affect sensitive groups at Upper Iweka, Awada and PH road. SO2 levels were below WHO standard but NO2 and PM10 average concentrations exceeded the WHO standard. Additionally, the study revealed the inability of FEPA at such high standard (26000 for SO2, 75000 for NO2 and 25000 for PM10 µg/m 3 ) to effectively evaluate criteria pollutants that are noxious even at low concentrations The Metamodeling and GIS mapping identified wind speed, wind direction and humidity as effective scavengers of SO2, NO2 and PM10. Hence, the study demonstrated that Upper Iweka is a major point source pollution and scavenged by wind speed, wind direction and humidity under prevailing rains in onitsha lower basin.