Evaluation of the difference between two spatiotemporal random fields

Yun, Sooin

Permalink

https://hdl.handle.net/2142/113176 Copy

Description

Title

Evaluation of the difference between two spatiotemporal random fields

Author(s)

Yun, Sooin

Issue Date

2021-07-13

Director of Research (if dissertation) or Advisor (if thesis)

Li, Bo

Doctoral Committee Chair(s)

Li, Bo

Committee Member(s)

• Simpson, Douglas
• Shao, Xiaofeng
• Zhao, Sihai Dave

Department of Study

Statistics

Discipline

Statistics

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Spatiotemporal random fields
• Multiple tests
• FDR control
• Spatial Extremes

Abstract

Comparing the spatial characteristics of spatiotemporal random fields is often in demand in various fields of study. Especially in climatology, people are interested in learning the difference between the synthetic climate simulation model and climate field reconstructions (CFR) which are estimates of the past climate constructed based on the proxy data. The thesis focuses on testing the two spatiotemporal climate fields. First, as assessing the CFR skill is critical for improving their interpretation and ultimately for deriving better CFR estimates, we apply new methods for assessing spatiotemporal skill using formalized null hypotheses. The test provides a detailed assessment of why CFR skill varies across multiple methods, with implications for improving future CFR estimates. Also, it will be more informative, if we could point out where the difference is located by conducting the hypothesis at each location. However, comparing spatiotemporal random fields at each location requires adjusting the multiplicity due to multiple comparisons. We develop a new multiple testing approach to detect the local differences in the spatial characteristics of two spatiotemporal random fields by taking the spatial information into account. The developed procedure is robust to model misspecification and allows for weak dependency among hypotheses. Lastly, predicting the future climate extreme is critical as the consequences of the extreme temperature or precipitation include high mortality rate or impact on agriculture and infrastructure. However, to date, very little work has investigated the difference in the extreme value behavior between two climate fields. The last chapter introduces how we assess the spatial extreme difference of two spatial random fields. This developed procedure could be employed to assess whether the maximum precipitation of the climate simulation models captures real data’s extreme behavior.

Graduation Semester

2021-08

Type of Resource

Thesis

Permalink

http://hdl.handle.net/2142/113176

Copyright and License Information

Copyright 2021 Sooin Yun

Owning Collections