Predicting historical indoor temperatures from available local weather data

In this article, a simple approach to modeling heat transfer between the outside environment to a location inside a building is used to precisely predict indoor temperatures for a large range of historical dates. The data collection, statistical modeling and prediction of inside temperature based on available weather data obtained outside the building of interest are presented. An initial simple linear regression model estimates the heat transfer mechanism between outside and inside which is used to predict historical indoor temperatures. The results of the model show that the inside temperature moderates but follows the outside temperatures with a seasonal pattern. In addition, uncertainty ranges for the estimates and predictions were constructed by calculating empirical confidence intervals for the average daily inside temperature and obtaining the range of observed temperatures within each day (within‐day variability). For the example of predicting the inside temperature of a Los Alamos National Laboratory storage bunker, the modeling approach provides excellent prediction over multiple years.

openalex: W2906062129