東華大學圖書館 |

Understanding Drivers of Ice Mass Loss in Greenland Through Sea-Level and Climate Modeling, Remote Sensing, and Machine Learning.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Understanding Drivers of Ice Mass Loss in Greenland Through Sea-Level and Climate Modeling, Remote Sensing, and Machine Learning./
Author:	Antwerpen, Rafael Marco.
Published:	Ann Arbor : ProQuest Dissertations & Theses, : 2024,
Description:	138 p.
Notes:	Source: Dissertations Abstracts International, Volume: 86-01, Section: B.
Contained By:	Dissertations Abstracts International86-01B.
Subject:	Climate change. -
Online resource:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31484060
ISBN:	9798383564288

Understanding Drivers of Ice Mass Loss in Greenland Through Sea-Level and Climate Modeling, Remote Sensing, and Machine Learning.
Antwerpen, Rafael Marco.

Understanding Drivers of Ice Mass Loss in Greenland Through Sea-Level and Climate Modeling, Remote Sensing, and Machine Learning. - Ann Arbor : ProQuest Dissertations & Theses, 2024 - 138 p.

Source: Dissertations Abstracts International, Volume: 86-01, Section: B.

Thesis (Ph.D.)--Columbia University, 2024.

Changes in global climate conditions significantly impact ice sheet and glacier mass change leading to global mean sea level (GMSL) change. One of the largest present-day contributors to GMSL is the Greenland ice sheet (GrIS) and it will likely continue to be so in the future. To accurately predict future ice mass changes, it is crucial to understand the response of GrIS to a changing climate and to correctly represent this behavior in climate models. The GrIS' contribution to GMSL can in large part be attributed to the loss of ice and snow mass from the ice sheet surface. The surface mass loss has accelerated in the past decades due to increased surface melting and runoff in response to atmospheric warming. Surface melting is strongly controlled by ice albedo, a complex and dynamic property of ice that regulates the amount of solar radiation that is absorbed or reflected by the surface. Absorbed solar radiation leads to heating and melting of the ice surface. However, we lack a comprehensive understanding of the physical processes controlling ice mass loss, including ice albedo. These processes are, therefore, often simplified or crudely parameterized in climate models and subsequently add to large uncertainties in sea level rise predictions. This uncertainty prevents effective mitigation of and adaptation to the effects of climate change and sea level rise. It is, therefore, essential to advance our understanding of these processes and their representation in climate models. In this dissertation, I describe improvements to our understanding of the behavior of the GrIS and pose improvements to climate modeling capabilities that can lead to a reduced uncertainty of sea level rise projections.In the first chapter, I put constraints on the past response of the GrIS to a changing climate. Understanding the response of the GrIS to times in the past when temperatures were as warm or warmer than today offers insights into its current and future response to climate change. The southwestern GrIS retreated inland beyond its current margin during the (at least regionally) warmer-than-present mid-Holocene, before it readvanced. To investigate the timing and magnitude of southwest GrIS retreat and readvance in response to Holocene warmth, we model the response of the solid Earth and local relative sea level (RSL) to past ice sheet change. I compare model predictions to observations of paleo and present-day RSL and present-day vertical land motion around Nuuk, Greenland. I find that the southwest GrIS minimum extent likely occurred between 5 and 3 ka and that the historical maximum extent was likely approached between 2 and 1 ka. Comparing this timing to local and regional records of temperature and ice-sheet change suggest that the evolution of the southwestern GrIS presented here was in-phase with the likely evolution of southwestern GrIS mass balance through the Holocene.In the second chapter, I assess the performance of a regional climate model in simulating the spatiotemporal variability of GrIS ice extent and ice albedo in the period 2000-2021. A large portion of runoff from the GrIS originates from exposure of the darker ice in the ablation zone when the overlying snow melts, where surface albedo plays a critical role in modulating the energy available for melting. Ice albedo is spatially and temporally variable and contingent on non-linear feedbacks and the presence of light-absorbing constituents. An assessment of models aiming at simulating albedo variability and associated impacts on meltwater production is crucial for improving our understanding of the processes governing these feedbacks and, in turn, surface mass loss from Greenland. Our findings suggest that the regional climate model Modele Atmospherique Regional (MAR) overestimates ice albedo on average by 22.8 % compared to the ice albedo observations derived from the Moderate Resolution Imaging Spectroradiometer (MODIS). We also find that this ice albedo bias can lead to an underestimation of total meltwater production from the GrIS ice zone of 42.8 %.In the third chapter, I build upon the second chapter and present PIXAL, a physics-informed explainable machine learning architecture for Greenland ice albedo modeling. PIXAL is an Extreme Gradient Boosting (XGBoost) model and is trained on a suite of modeled topographic, atmospheric, radiative, and glaciologic variables from MAR to capture the complex and non-linear relationships with ice albedo observations from MODIS in the period 2000-2021. PIXAL outperforms MAR in modeling ice albedo on the southwestern GrIS. The performance metrics show that PIXAL achieves an R2 of 0.563, an SSIM of 0.620, an MSE of 0.005, and a MAPE of 14.699%, compared to MAR's R2 of 0.062, SSIM of 0.112, MSE of 0.032, and MAPE of 46.202%. Explainable artificial intelligence (XAI) analysis reveals that topographic features, specifically ice sheet surface height and slope, are primary drivers of ice albedo. Near-surface air temperature and runoff further impact ice albedo. These findings highlight that understanding the complex physical processes underlying ice albedo variability is essential for accurate climate modeling and sea level rise predictions. PIXAL represents a crucial advancement in ice albedo modeling and paves the way for improved climate models that can more accurately estimate GrIS ice surface melting and its contribution to sea level rise.Overall, my results have implications for future ice sheet modeling studies targeting Greenland and provide a deeper understanding of the interactions between the climate and the cryosphere and thus of future ice sheet change.

ISBN: 9798383564288Subjects--Topical Terms:

2079509
Climate change.
Subjects--Index Terms:

Albedo

Understanding Drivers of Ice Mass Loss in Greenland Through Sea-Level and Climate Modeling, Remote Sensing, and Machine Learning.
LDR:06883nmm a2200409 4500 001 2428184
005 20241210085427.5
006 m o d
007 cr#unu||||||||
008 261011s2024 ||||||||||||||||| ||eng d
020 $a 9798383564288
035 $a (MiAaPQ)AAI31484060
035 $a AAI31484060
040 $a MiAaPQ $c MiAaPQ
100 1 $a Antwerpen, Rafael Marco. $3 3815223
245 1 0 $a Understanding Drivers of Ice Mass Loss in Greenland Through Sea-Level and Climate Modeling, Remote Sensing, and Machine Learning.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2024
300 $a 138 p.
500 $a Source: Dissertations Abstracts International, Volume: 86-01, Section: B.
500 $a Advisor: Tedesco, Marco.
502 $a Thesis (Ph.D.)--Columbia University, 2024.
520 $a Changes in global climate conditions significantly impact ice sheet and glacier mass change leading to global mean sea level (GMSL) change. One of the largest present-day contributors to GMSL is the Greenland ice sheet (GrIS) and it will likely continue to be so in the future. To accurately predict future ice mass changes, it is crucial to understand the response of GrIS to a changing climate and to correctly represent this behavior in climate models. The GrIS' contribution to GMSL can in large part be attributed to the loss of ice and snow mass from the ice sheet surface. The surface mass loss has accelerated in the past decades due to increased surface melting and runoff in response to atmospheric warming. Surface melting is strongly controlled by ice albedo, a complex and dynamic property of ice that regulates the amount of solar radiation that is absorbed or reflected by the surface. Absorbed solar radiation leads to heating and melting of the ice surface. However, we lack a comprehensive understanding of the physical processes controlling ice mass loss, including ice albedo. These processes are, therefore, often simplified or crudely parameterized in climate models and subsequently add to large uncertainties in sea level rise predictions. This uncertainty prevents effective mitigation of and adaptation to the effects of climate change and sea level rise. It is, therefore, essential to advance our understanding of these processes and their representation in climate models. In this dissertation, I describe improvements to our understanding of the behavior of the GrIS and pose improvements to climate modeling capabilities that can lead to a reduced uncertainty of sea level rise projections.In the first chapter, I put constraints on the past response of the GrIS to a changing climate. Understanding the response of the GrIS to times in the past when temperatures were as warm or warmer than today offers insights into its current and future response to climate change. The southwestern GrIS retreated inland beyond its current margin during the (at least regionally) warmer-than-present mid-Holocene, before it readvanced. To investigate the timing and magnitude of southwest GrIS retreat and readvance in response to Holocene warmth, we model the response of the solid Earth and local relative sea level (RSL) to past ice sheet change. I compare model predictions to observations of paleo and present-day RSL and present-day vertical land motion around Nuuk, Greenland. I find that the southwest GrIS minimum extent likely occurred between 5 and 3 ka and that the historical maximum extent was likely approached between 2 and 1 ka. Comparing this timing to local and regional records of temperature and ice-sheet change suggest that the evolution of the southwestern GrIS presented here was in-phase with the likely evolution of southwestern GrIS mass balance through the Holocene.In the second chapter, I assess the performance of a regional climate model in simulating the spatiotemporal variability of GrIS ice extent and ice albedo in the period 2000-2021. A large portion of runoff from the GrIS originates from exposure of the darker ice in the ablation zone when the overlying snow melts, where surface albedo plays a critical role in modulating the energy available for melting. Ice albedo is spatially and temporally variable and contingent on non-linear feedbacks and the presence of light-absorbing constituents. An assessment of models aiming at simulating albedo variability and associated impacts on meltwater production is crucial for improving our understanding of the processes governing these feedbacks and, in turn, surface mass loss from Greenland. Our findings suggest that the regional climate model Modele Atmospherique Regional (MAR) overestimates ice albedo on average by 22.8 % compared to the ice albedo observations derived from the Moderate Resolution Imaging Spectroradiometer (MODIS). We also find that this ice albedo bias can lead to an underestimation of total meltwater production from the GrIS ice zone of 42.8 %.In the third chapter, I build upon the second chapter and present PIXAL, a physics-informed explainable machine learning architecture for Greenland ice albedo modeling. PIXAL is an Extreme Gradient Boosting (XGBoost) model and is trained on a suite of modeled topographic, atmospheric, radiative, and glaciologic variables from MAR to capture the complex and non-linear relationships with ice albedo observations from MODIS in the period 2000-2021. PIXAL outperforms MAR in modeling ice albedo on the southwestern GrIS. The performance metrics show that PIXAL achieves an R2 of 0.563, an SSIM of 0.620, an MSE of 0.005, and a MAPE of 14.699%, compared to MAR's R2 of 0.062, SSIM of 0.112, MSE of 0.032, and MAPE of 46.202%. Explainable artificial intelligence (XAI) analysis reveals that topographic features, specifically ice sheet surface height and slope, are primary drivers of ice albedo. Near-surface air temperature and runoff further impact ice albedo. These findings highlight that understanding the complex physical processes underlying ice albedo variability is essential for accurate climate modeling and sea level rise predictions. PIXAL represents a crucial advancement in ice albedo modeling and paves the way for improved climate models that can more accurately estimate GrIS ice surface melting and its contribution to sea level rise.Overall, my results have implications for future ice sheet modeling studies targeting Greenland and provide a deeper understanding of the interactions between the climate and the cryosphere and thus of future ice sheet change.
590 $a School code: 0054.
650 4 $a Climate change. $2 bicssc $3 2079509
650 4 $a Environmental science. $3 677245
650 4 $a Atmospheric sciences. $3 3168354
650 4 $a Computer science. $3 523869
653 $a Albedo
653 $a Cryosphere
653 $a Greenland
653 $a Ice melting
653 $a Machine learning
653 $a Remote sensing
690 $a 0404
690 $a 0768
690 $a 0725
690 $a 0984
710 2 $a Columbia University. $b Earth and Environmental Sciences. $3 2049773
773 0 $t Dissertations Abstracts International $g 86-01B.
790 $a 0054
791 $a Ph.D.
792 $a 2024
793 $a English
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31484060