A Study on Interpolation of Missing Values in Multivariate Sand and Dust Data Considering Timeseries Features

The production and survival of humans are threatened by sand and dust storms; yet, the research of sand and dust storms is always beset by the issue of partially absent data. In order to address the missing data issue, we presented the sand-dust data interpolation technique (SDIM) in this study. While maintaining the overall structure of the Transformer model, our model employs an automatic coding and decoding technique for improved interpolation of sand and dust data, along with a timelag decay strategy along with positional coding and masking matrices. The associated results analysis reveals that the three evaluation indexes of mean absolute error, mean absolute percentage error, and root mean square error are the lowest when compared to the interpolation performance of conventional data interpolation methods like mean interpolation.