<![CDATA[Abstract Air temperature sensor is installed in an enclosure with a relative humidity sensor at a height of 1.25 meters from the ground. This sensor is routinely calibrated every year through a field comparison mechanism using standard portable. However, air temperature sensor has potential to experience troubles due to technical or non-technical factors. Estimation of temperature data can overcome loss data. This study aims to design an estimation model for air temperature sensor data using the ARIMA-MLP hybrid algorithm. Estimated air temperature sensor is focused on AWS Cisurupan in Garut Regency, West Java on 2019. The estimated data has an interval of 10 minutes. Outlier detection uses range check and step check methods. Missing data imputation method utilizes multiple linear regression-based interpolation. ARIMA-MLP hybrid algorithm is able to detect elements of linearity and nonlinearity of air temperature measurements. This algorithm meets WMO requirements regarding air temperature measurements, if training data is used with a minimum percentage of 85% of the entire modeling dataset. The resulting RMSE value is less than 0.20C. Keywords: air temperature, Automatic Weather Station, hybrid ARIMA-MLP]]>
Copyrights © 2023
