The Second Competition on Spatial Statistics for Large Datasets

In the last few decades, size of spatial and spatio-temporal datasets in many research areas has rapidly increased with development data collection technologies. As a result, classical statistical methods statistics are facing computational challenges. For example, kriging predictor geostatistics becomes prohibitive on traditional hardware architectures for large as it requires high computing power memory footprint when dealing dense matrix operations. Over years, various approximation have been proposed to address such issues, however, community lacks holistic process assess their efficiency. To provide fair assessment, 2021, we organized first competition datasets, generated by our ExaGeoStat software, asked participants report results estimation prediction. Thanks its widely acknowledged success at request participants, second 2022 focusing predictions more complex processes, including univariate nonstationary stationary space-time bivariate processes. this paper, describe detail generation procedure make valuable publicly available wider adoption. Then, review submitted from fourteen teams worldwide, analyze outcomes, performance each team.