BACI: 2015-2019

TOWARDS A BIOSPHERE ATMOSPHERE CHANGE INDEX

WP5 Synthetic Index and Attribution Scheme: the BACIndex

Lead: FSU Jena
Main Contact: Joachim Denzler
Team: Maha Shadaydeh, Yanira Guanche, Joachim Denzler (WP5 lead)

OBJECTIVES

The goal of WP5 is the development of a synthetic change or novelty index (BACIndex) that allows for detecting sudden events and abnormal changes in the multivariate EO data streams. The index will be generic, but in the context of the project, specific emphasis will be on detecting abrupt changes that are relevant to the essential ecosystem variables, i.e. properties relevant to the functioning of terrestrial ecosystems, biosphere-atmosphere exchanges of matter and energy, and biodiversity related properties. The problem of detecting change in given data is a challenging machine learning problem and is often referred to as “novelty detection” or “discovery”. WP5 also entails an attribution scheme, i.e. machine learning techniques that allow for measuring the influence of input variables on the novelty detection score.


Hotspots of Anomalies: combining different machine learning methods we try to detect extreme events in historical data and define those areas with higher amount of abnormal records. Here is shown the average of anomalies over the year 2010, with the 2010 Russian heat wave.

Papers that were supported by this WP

1Flach, M., Gans, F., Brenning, A., Denzler, J., Reichstein, M., Rodner, E., Bathiany, S., Bodesheim, P., Guanche, Y., Sippel, S., Mahecha, M. D. (2017). Multivariate anomaly detection for Earth observations: a comparison of algorithms and feature extraction techniques. Earth System Dynamics, 8(3), 677-696. doi:10.5194/esd-8-677-2017.
2Guanche Garcia, Y., Rodner, E., Flach, M., Sippel, S., Mahecha, M. D., Denzler, J. (2016). Detecting multivariate biosphere extremes. In A. Banerjee, W. Ding, V. Dy (Eds.), Proceedings of the 6th International Workshop on Climate Informatics: CI2016: NCAR Technical Note NCAR/TN-529+PROC (pp. 9-12). Boulder: National Center for Atmospheric Research.
3Rodner, E., Barz, B., Guanche, Y., Flach, M., Mahecha, M. D., Bodesheim, P., Reichstein, M., Denzler, J. (2016). Maximally divergent intervals for anomaly detection. In ICML 2016 Anomaly Detection Workshop. doi:10.17871/BACI_ICML2016_Rodner.
4Sippel, S., Zscheischler, J., Heimann, M., Lange, H., Mahecha, M. D., van Oldenborgh, G. J., Otto, F. E. L., Reichstein, M. (2017). Have precipitation extremes and annual totals been increasing in the world’s dry regions over the last 60 years? Hydrology and Earth System Sciences, 21(1), 441-458. doi:10.5194/hess-21-441-2017.
5Sippel, S., Zscheischler, J., Mahecha, M. D., Orth, R., Reichstein, M., Vogel, M., Seneviratne, S. I. (2017). Refining multi-model projections of temperature extremes by evaluation against land–atmosphere coupling diagnostics. Earth System Dynamics, 8(2), 387-403. doi:10.5194/esd-8-387-2017.