An Online Paleoclimate Data Assimilation With a Deep Learning‐Based Network

Abstract An online paleoclimate data assimilation (PDA) that utilizes climate forecasts from a deep learning‐based network (NET) along with assimilation of proxies to reconstruct surface air temperature, is investigated here. The NET is trained on ensemble simulations from the Community Earth System...

Full description

Saved in:
Bibliographic Details
Main Authors: Haohao Sun, Lili Lei, Zhengyu Liu, Liang Ning, Zhe‐Min Tan
Format: Article
Language:English
Published: American Geophysical Union (AGU) 2025-06-01
Series:Journal of Advances in Modeling Earth Systems
Subjects:
Online Access:https://doi.org/10.1029/2024MS004675
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849683514711080960
author Haohao Sun
Lili Lei
Zhengyu Liu
Liang Ning
Zhe‐Min Tan
author_facet Haohao Sun
Lili Lei
Zhengyu Liu
Liang Ning
Zhe‐Min Tan
author_sort Haohao Sun
collection DOAJ
description Abstract An online paleoclimate data assimilation (PDA) that utilizes climate forecasts from a deep learning‐based network (NET) along with assimilation of proxies to reconstruct surface air temperature, is investigated here. The NET is trained on ensemble simulations from the Community Earth System Model‐Last Millennium Ensemble. Due to the nonlinear features with high‐dimensional input, NET gains better predictive skills compared to the linear inverse model (LIM) in a reduced empirical orthogonal function (EOF) space. Thus, an alternative for online PDA is to couple the NET with the integrated hybrid ensemble Kalman filter (IHEnKF). Moreover, an analog blending strategy is proposed to increase ensemble spread and mitigate filter divergence, which blends the analog ensembles selected from climatological samples based on proxies and cycling ensembles advanced by NET. To account for the underestimated uncertainties of real proxy data, an observation error inflation method is applied, which inflates the proxy error variance based on the comparison between the estimated proxy error variance and its climatological innovation. Consistent results are obtained from the pseudoproxy experiments and the real proxy experiments. The more informative ensemble priors from the online PDA using NET enhance the reconstructions than the online PDA using LIM, and both outperform the offline PDA with randomly sampled climatological ensemble priors. The advantages of online PDA with NET over the online PDA with LIM and offline PDA become more pronounced, as the proxy data become sparser.
format Article
id doaj-art-30dceebb620a4dd285f6c324c2ed9f98
institution DOAJ
issn 1942-2466
language English
publishDate 2025-06-01
publisher American Geophysical Union (AGU)
record_format Article
series Journal of Advances in Modeling Earth Systems
spelling doaj-art-30dceebb620a4dd285f6c324c2ed9f982025-08-20T03:23:51ZengAmerican Geophysical Union (AGU)Journal of Advances in Modeling Earth Systems1942-24662025-06-01176n/an/a10.1029/2024MS004675An Online Paleoclimate Data Assimilation With a Deep Learning‐Based NetworkHaohao Sun0Lili Lei1Zhengyu Liu2Liang Ning3Zhe‐Min Tan4Key Laboratory of Mesoscale Severe Weather Ministry of Education, and School of Atmospheric Sciences Nanjing University Nanjing ChinaKey Laboratory of Mesoscale Severe Weather Ministry of Education, and School of Atmospheric Sciences Nanjing University Nanjing ChinaDepartment of Geography The Ohio State University Columbus OH USANational Key Laboratory for Virtual Geographic Environment Ministry of Education, and School of Geography Nanjing Normal University Nanjing ChinaKey Laboratory of Mesoscale Severe Weather Ministry of Education, and School of Atmospheric Sciences Nanjing University Nanjing ChinaAbstract An online paleoclimate data assimilation (PDA) that utilizes climate forecasts from a deep learning‐based network (NET) along with assimilation of proxies to reconstruct surface air temperature, is investigated here. The NET is trained on ensemble simulations from the Community Earth System Model‐Last Millennium Ensemble. Due to the nonlinear features with high‐dimensional input, NET gains better predictive skills compared to the linear inverse model (LIM) in a reduced empirical orthogonal function (EOF) space. Thus, an alternative for online PDA is to couple the NET with the integrated hybrid ensemble Kalman filter (IHEnKF). Moreover, an analog blending strategy is proposed to increase ensemble spread and mitigate filter divergence, which blends the analog ensembles selected from climatological samples based on proxies and cycling ensembles advanced by NET. To account for the underestimated uncertainties of real proxy data, an observation error inflation method is applied, which inflates the proxy error variance based on the comparison between the estimated proxy error variance and its climatological innovation. Consistent results are obtained from the pseudoproxy experiments and the real proxy experiments. The more informative ensemble priors from the online PDA using NET enhance the reconstructions than the online PDA using LIM, and both outperform the offline PDA with randomly sampled climatological ensemble priors. The advantages of online PDA with NET over the online PDA with LIM and offline PDA become more pronounced, as the proxy data become sparser.https://doi.org/10.1029/2024MS004675data assimilationdeep learningpaleoclimate reconstruction
spellingShingle Haohao Sun
Lili Lei
Zhengyu Liu
Liang Ning
Zhe‐Min Tan
An Online Paleoclimate Data Assimilation With a Deep Learning‐Based Network
Journal of Advances in Modeling Earth Systems
data assimilation
deep learning
paleoclimate reconstruction
title An Online Paleoclimate Data Assimilation With a Deep Learning‐Based Network
title_full An Online Paleoclimate Data Assimilation With a Deep Learning‐Based Network
title_fullStr An Online Paleoclimate Data Assimilation With a Deep Learning‐Based Network
title_full_unstemmed An Online Paleoclimate Data Assimilation With a Deep Learning‐Based Network
title_short An Online Paleoclimate Data Assimilation With a Deep Learning‐Based Network
title_sort online paleoclimate data assimilation with a deep learning based network
topic data assimilation
deep learning
paleoclimate reconstruction
url https://doi.org/10.1029/2024MS004675
work_keys_str_mv AT haohaosun anonlinepaleoclimatedataassimilationwithadeeplearningbasednetwork
AT lililei anonlinepaleoclimatedataassimilationwithadeeplearningbasednetwork
AT zhengyuliu anonlinepaleoclimatedataassimilationwithadeeplearningbasednetwork
AT liangning anonlinepaleoclimatedataassimilationwithadeeplearningbasednetwork
AT zhemintan anonlinepaleoclimatedataassimilationwithadeeplearningbasednetwork
AT haohaosun onlinepaleoclimatedataassimilationwithadeeplearningbasednetwork
AT lililei onlinepaleoclimatedataassimilationwithadeeplearningbasednetwork
AT zhengyuliu onlinepaleoclimatedataassimilationwithadeeplearningbasednetwork
AT liangning onlinepaleoclimatedataassimilationwithadeeplearningbasednetwork
AT zhemintan onlinepaleoclimatedataassimilationwithadeeplearningbasednetwork