Authors:
Mohamed Redha Sidoumou
1
;
Alisa Kim
2
;
Jeremy Walton
3
;
Douglas I. Kelley
4
;
Robert J. Parker
5
and
Ranjini Swaminathan
6
Affiliations:
1
Amazon Web Services, U.K.
;
2
Amazon Web Services, Germany
;
3
Met Office Hadley Centre for Climate Science and Services, Exeter, U.K.
;
4
U.K. Centre for Ecology and Hydrology, Wallingford, U.K.
;
5
National Centre for Earth Observation, Space Park Leicester, University of Leicester, U.K.
;
6
National Centre for Earth Observation, Department of Meteorology, University of Reading, U.K.
Keyword(s):
Explainable AI, Explainability, Biomes, Clustering, Segmentation.
Abstract:
We present an explainable clustering approach for use with 3D tensor data and use it to define terrestrial biomes from observations in an automatic, data-driven fashion. Our approach allows us to use a larger number of features than is feasible for current empirical methods for defining biomes, which typically rely on expert knowledge and are inherently more subjective than our approach. The data consists of 2D maps of geophysical observation variables, which are rescaled and stacked to form a 3D tensor. We adapt an image segmentation algorithm to divide the tensor into homogeneous regions before partitioning the data using the k-means algorithm. We add explainability to the classification by approximating the clusters with a compact decision tree whose size is limited. Preliminary results show that, with a few exceptions, each cluster represents a biome which can be defined with a single decision rule.