Calibrations and Performance Evaluations of Metal Oxide Low- Cost Air Sensor for NH3, CO and NO2 Detections: Case study for Laboratory and Field Calibration of across Dar es Salaam City

Air pollution is a critical global challenge, impacting quality of life and public health of vulnerable communities in low- and middle-income countries (LMICs), due to absence of monitoring devices, weak policies and fragmented institutions for effective air quality (AQ) management. Carbon monoxide (CO), nitrogen dioxide (NO₂) and ammonia (NH₃), are among the gaseous pollutants found in urban cities with potential to cause respiratory illnesses or cardiovascular diseases. The metal oxide low cost sensors are among the emerging air pollutants measuring devices for indicative measurements in urban areas, because they provide fast cheap results and allow to attain good spatial coverage at a low cost even in areas with no reference monitors. The low cost of these sensors comes with low performances and low qualities of data recorded as compared to reference grade sensors and thus with requirements for frequent calibrations. Frequent calibration necessitates availability of computational resources with skills for data analysis and modeling. The low-cost sensor systems (LCS) are solution for LMICs in building local capacities for addressing these challenges but strong local and international collaboration to improve AQ management is required.

The study focuses on calibrations and performance evaluations of MICS 6814 based gaseous low cost sensors using datasets collected in laboratory and field settings. The calibration is based on testing the performances of model equation for transforming sensor responses into air pollutants measurements in laboratory and field deployments. The results of linear calibration procedures in laboratory showed the transformation of sensor responses into air pollutant concentration to be significantly good with coefficient of determination R² ranging from 0.55 – 1, for all gaseous pollutants. The performance evaluation results for sensor deployments in field across the city showed varying results depending on site categories and time of day such that the coefficient of determination R² for CO, ranged from 0.82 – 1; for NH₃, R² ranging from 0.23 - 0.98, and for NO₂ , R² ranged from 0.52 – 0.9. Variations of coefficient of determination R² during sensor calibrations in field, poses challenges in developing network wide calibration models for real time air pollutants monitoring. Further challenge, comes from the fact that the trends for individual field observations datasets, showed varying picking levels of pollutions during morning and evening rush hours for all categories of sensor location. Over all, the observed indicative measure of air pollutants across the city were sufficient for public awareness and policy making purposes.