A Statistical Approach for A-Posteriori Deployment of Microclimate Sensors in Museums: A Case Study

Abstract

The deployment of sensors is the first issue encountered when microclimate monitoring is planned in spaces devoted to the conservation of artworks. Sometimes, the first decision regarding the position of sensors may not be suitable for characterising the microclimate close to climate-sensitive artworks or should be revised in light of new circumstances. This paper fits into this context by proposing a rational approach for a posteriori deployment of microclimate sensors in museums where long-term temperature and relative humidity observations were available (here, the Rosenborg Castle, Copenhagen, Denmark). Different statistical tools such as box-and-whisker plots, principal component analysis (PCA) and cluster analysis (CA) were used to identify microclimate patterns, i.e., similarities of indoor air conditions among rooms. Box-and-whisker plots allowed us to clearly identify one microclimate pattern in two adjoining rooms located in the basement. Multivariate methods (PCA and CA) enabled us to identify further microclimate patterns by grouping not only adjoining rooms but also rooms located on different floors. Based on these outcomes, new configurations about the deployment of sensors were proposed aimed at avoiding redundant sensors and collecting microclimate observations in other sensitive locations of this museum.

Keywords: cluster analysis; deployment; microclimate; multivariate approach; museum; principal component analysis; relative humidity; sensors; temperature.

Figure 1

Rosenborg Castle: (a) 3D sketch of the external view; (b) the Great Hall (Room 21) located on the second floor.

Figure 2

Plans of Rosenborg Castle: (a) First-guess configuration of the TRH sensors (yellow circles). (b) Location of those artworks (black parallelepipeds) which were selected by museum conservators in the framework of the CollectionCare project. Room numbers are provided in the circles and also correspond to the associated TRH sensor.

Figure 3

Schematic workflow of the approach: Data collection from several sensors located in different rooms; data pre-processing to select long-term time series in multi-room observations; application of statistical methods to identify microclimate similarities among rooms to be associated with climate-room blocks.

Figure 4

Schematic workflow of the k-means clustering.

Figure 5

Results of the completeness index (CoI) for each room and each year over the period under study. Note that the term “year” here indicates the period (365 days) covering the months from June to May of the following year.

Figure 6

Box-and-whisker plots of temperature (a) and relative humidity (b) recorded in Rosenborg Castle over the period from 1 June 2017 until 31 May 2018. Additionally, outdoor air conditions retrieved from the ERA5 database [24] were also considered (labelled as “out”). The length of whiskers was set equal to 1.5 times the interquartile range (IQR).

Figure 7

Scatter plot of the loadings corresponding to Component 2 versus Component 1 for the hourly data of (a) temperature and (b) relative humidity.

Figure 8

Time evolution of temperature (left panels) and relative humidity (right panels) in each cluster based on outcomes of different k-mean Cluster Analysis: (a,b) for k = 2; (c,d) for k = 3; (e,f) for k = 4; (g,h) for k = 5. Colour code: Cluster 1 in blue; Cluster 2 in orange; Cluster 3 in yellow; Cluster 4 in violet; Cluster 5 in green; outdoor conditions in black.

Figure 9

Plans of Rosenborg Castle. A posteriori deployment of the TRH sensors according to: (a) box-and-whisker plots, (b) principal component analysis, and (c) k-means cluster analysis.