The raster data model is commonly used to represent spatially continuous phenomena. When the phenomenon under study is dynamic, time series of raster data are typically used to capture temporal changes. Because of the availability and ubiquity of (remote) sensors, the importance of such raster time series is ever increasing.

Raster time series can be analyzed using a wide variety of methods (e.g. Blok et al. 2011). Also, many methods can be found in the literature to extract spatio-temporal patterns from this kind of data (Zurita-Milla et al., 2013). However, the comparison of patterns extracted from two or more raster time series has received little attention so far. For instance, presently there is not a simple method to compare patterns belonging to related processes like precipitation and the subsequent greening-up of vegetation (especially if the two raster time series have different cell sizes).

The MSc candidate will start by reviewing existing methods to extract and identify spatio-temporal patterns of co-variability in time. After that, the most promising methods will be selected for further evaluation. A likely winner could be the method presented by G. Eshel in his book Spatio-temporal data analysis. This method is based on the singular value decomposition.

As a case study, we propose looking at two vegetation development time series: a time series of temperature-based phenological indices and a series of start of season derived from Earth observation sensors. The spring index raster time series are available for The Netherlands. The time start of season should be created by, for instance, downloading MODIS data from public FTP/servers or from Google Earth Engine (a cloud storage and processing systems).

This MSc research topic links with current work done at the GIP department on the extraction of information from heterogeneous data sources. Several MSc theses have been successfully completed on the analysis and modelling of volunteered phenological observations and two PhD students (Hamed Mehdipoor and Irene Garcia) are currently active in related topics. Given the nature of the topic, an analytical mind and affinity with programming are required to complete the proposed topic.