Restoration of missing dry-bulb temperature data with long-term gaps (up to 60 days) for use in building performance monitoring and analysis—Part II

The lack of standard procedures for filling climatic data has the potential to undermine design, monitoring, and control efforts aimed at climate-responsive building design, performance monitoring, and energy efficiency. This article addresses the challenge of long-term missing gaps in dry-bulb temperature data by examining three spatial methods, namely the inverse distance weighting (IDW) method, the spatial regression test (SRT) method, and the substitution with best match data (SSBM) method, as well as two temporal methods, namely the temporal regression test (TRT) method and the temporal substitution with best match data (TSBM) method. Using these methods, missing dry-bulb temperature data with long-term gaps, ranging from 1 to 60 days, are restored for use in building performance monitoring and analysis. Three one-year, hourly datasets were used to evaluate the performance of these approaches. Each method was applied to deal with artificial gaps which were generated randomly and represented different seasons of a year. In terms of the difference between estimated values and measured values, three evaluation indices, namely mean absolute error (MAE), root mean square error (RMSE), and standard error of bias (BIASSTD), were utilized. The comparison results show that spatial methods are better than temporal methods. The confidence level of the SRT method was further investigated by applying this method to existing data and missing data, and examining its performance. The results indicate that the uncertainty of the SRT method can be predicted and at least two neighboring stations are recommended when using it. This is the second part of the research results obtained through the ASHRAE 1413 research project (in press) with a focus on introducing gap-filling methods for long-term gaps in dry-bulb temperature.