Filling Gaps in Daily Temperature Data with a CNN-LSTM Model

Abstract

The challenge of missing data poses a significant difficulty in the practical use of recorded meteorological data, a concern that has become apparent in Ukraine in recent years. This paper presents a deep learning methodology for imputing consecutive missing data within weather station records. By integrating convolutional neural network and Long Short-Term Memory (LSTM) layers, both recognized as prominent techniques in weather data forecasting and imputation, this approach provides reliable results for filling gaps in daily air temperature data.