Relative humidity prediction with covariates and error correction based on SARIMA-EG-ECM model

RH is a physical quantity measuring atmospheric water vapor content. Predicting RH is of great importance in weather, climate, industrial production, crops, human health, and disease transmission, since it is helpful in making critical decisions. In this paper, the effects of covariates and error correction on relative humidity (RH) prediction have been studied, and a hybrid model based on seasonal autoregressive integrated moving average (SARIMA) model, cointegration (EG), and error correction model (ECM) named SARIMA-EG-ECM (SEE) has been proposed. The prediction model was performed in the meteorological observations of Hailun Agricultural Ecology Experimental Station, China. Based on the SARIMA model, the meteorological variables that interact with RH were used as covariates to perform EG tests. A cointegration model has been constructed. It revealed that RH had a cointegration relationship with air temperature (TEMP), dew point temperature (DEWP), precipitation (PRCP), atmospheric pressure (ATMO), sea-level pressure (SLP), and 40 cm soil temperature (40ST), which revealed the long-term equilibrium relationship between series. An ECM was established which indicated that the current fluctuations of DEWP, ATMO, and SLP have a significant impact on the current fluctuations of RH. The established ECM describes the short-term fluctuation relationship between the series. With the increase of the forecast horizon from 6 to 12 months, the prediction performance of the SEE model decreased slightly. A comparative study has also been introduced, indicating that the SEE performs superior to SARIMA and Long Short-Term Memory (LSTM) network.