Filling missing temperature values in weather data banks

Meteorological data (wind speed, wind direction, rainfall, temperature etc) are an essential parameter for energy applications studies and development. Weather data is subject to different types of errors. The most commonly observed problems in temperature data embrace missing observations, unreasonable readings, spurious zeroes, and so on. Therefore, the data must be cleaned - that is, the errors and omissions must be corrected. In this research, the methodology adopted is to discard certain observed values and treat them as ´missing data´. We then examine and analyse the imputation accuracy of different interpolation techniques and filling methods for missing historical records of temperature data. The performance of these techniques as predictors of the missing values is evaluated using standard statistical indicator, such as correlation coefficient, root mean squared error, etc