About

Isabella Velicogna is a professor whose research focuses on the study of terrestrial hydrological cycles, vegetation response to water supply changes, and the ongoing mass loss in the Polar Regions. Her work involves utilizing remote sensing data, ground observations, and climate model outputs to analyze water storage changes from GRACE, soil moisture from SMAP, MODIS, AMSR, and GOME-2, as well as vegetation indices from MODIS. She is also engaged in 3-D numerical modeling of the Glacial Isostatic Adjustment of the Earth. Her research aims to understand the dynamics of ice sheet mass balance, glacier and ice cap mass balance, and their implications for sea level rise. Velicogna's contributions include advancing the understanding of Earth's water and ice systems through innovative use of satellite data and modeling techniques, thereby providing critical insights into climate change impacts and Earth's evolving cryosphere.

Research topics

• Oceanography
• Geology
• Environmental science
• Climatology
• Geography
• Physical geography
• Earth science
• Physics
• Atmospheric sciences
• Geomorphology

Selected publications

• Uncertainty Quantification of Satellite-Based Essential Climate Variables Derived from Deep Learning

  Surveys in Geophysics · 2026-01-30 · 2 citations

  preprintOpen access

  Abstract Accurate uncertainty information associated with essential climate variables (ECVs) is crucial for reliable climate modeling and understanding the spatiotemporal evolution of the Earth system. Recent developments in deep learning have remarkably advanced the estimation of ECVs with improved accuracy. However, the quantification of uncertainties associated with outputs of such deep learning models has yet to be widely adopted. This survey explores the types of uncertainties associated with ECVs derived from deep learning methods, including aleatoric (data) and epistemic (model) uncertainty, and the techniques to quantify them. The focus is on highlighting the importance of considering uncertainty associated with inputs in the deep learning models to account for the dynamic and multifaceted nature of satellite observations. The survey starts by clarifying the definitions of aleatoric and epistemic uncertainties and their roles in a typical satellite observation processing workflow, followed by bridging the gap between conventional statistical and deep learning views on uncertainties. Then, we comprehensively review the existing uncertainty quantification methods for deep learning algorithms and discuss their strengths and limitations. A comprehensive literature review about quantifying uncertainties in the deep learning estimates of ECVs follows the theoretical survey, covering a wide range of ECVs. The specific need for modification to fit the requirements from both the Earth observation side and the deep learning side in such interdisciplinary tasks is highlighted. We further demonstrate our findings with two selected ECV examples, snow cover and terrestrial water storage, to provide clear insights into different methods by promoting quantitative comparison. In the end, we summarize our findings and provide perspectives for future research.

  PublisherOA PDFDOI

• Recent shift in California’s precipitation distribution based on observations

  Environmental Research Communications · 2026-04-01

  articleOpen accessSenior author

  Abstract Changes in precipitation in California are of great relevance to freshwater resource management, agriculture and food production, ecosystem health, wildfire risks, flood risks, and hydro-electric power generation. In this study, we use Parameter-elevation Regressions on Independent Slopes Model observational data to detect changes in California precipitation during 1981–2022. We report widespread drying after the late 1990s. The decline in mean annual precipitation between the first and last 20-year periods exceeds 100 mm yr −1 across more than 30% of California, with the largest losses concentrated in coastal regions, the Sierra Nevada, and northern California. The drying trend is obscured by large inter-annual variability and the occurrence of extreme precipitation events, especially during the wet season. The reduction in mean annual precipitation is accompanied by a decrease in precipitation variability and extreme precipitation events. The contribution of extreme precipitation events (top 8-percentile) to mean annual precipitation has decreased from 24% during 1982–2001 to 17% during 2002–2021. In contrast, extremely dry events (bottom 8-percentile) have increased modestly in several northern and central regions. A regression analysis indicates that large-scale circulation modes capture the spatial pattern of the observed drying trends during the analyzed period, even though they only explain ∼25% of the observed inter-annual variability in wet anomalies. These results indicate a gradual shift in California’s precipitation toward drier conditions and fewer extreme precipitation events in recent decades, which has implications for adaptation to drought, water resources management, and evaluation of hydroclimatic risks.

  PublisherDOI

• Decomposing land surface total water storage in the Indus, Ganges, and Brahmaputra basins

  Frontiers in Earth Science · 2025-09-05

  articleOpen accessSenior author

  Introduction The goal of this study is to decompose the influence of specific hydrologic reservoirs in the Earth’s critical zone that interact to create observed total water supply (TWS) anomalies in the highly altered and densely populated Indus, Ganges, and Brahmaputra drainage basins. Understanding the contributions to TWS anomalies can help find potential solutions for the sustainability of human water supply. Methods We compare changes in the macroscale hydrology of three important High Mountain Asian drainage basins through seasonal and long-term trends in TWS. Statistical time-series analysis of nine individual TWS components modeled by a hydrologic model are used to simulate water storage terms. Results Long-term TWS trends look similar across the study basins, we find that the drivers and causes of trends and their seasonal variability are fundamentally different in each basin. TWS declines in the Indus and Ganges watersheds are primarily driven by the depletion of aquifers (67% and 76%, respectively) due to irrigated land expansion and water overuse. The Brahmaputra lower aquifer water use stress, and its TWS drop is mostly due to the melting of glaciers, the highest rate over all three basins. The Ganges and Brahmaputra have a quasi-monotonic decline of TWS, and the Indus basin exhibits a non-monotonic trend line of TWS due to different stages of its aquifer depletion relevant to aquifer water accessibility limited by well depth thresholds. Seasonal variability is primarily controlled by soil moisture saturation, shallow groundwater levels, reservoir storage, and snow accumulation for the Ganges and Brahmaputra basins. The Indus is driven by high mountain storage of snow and glaciers. Discussion The combination of hydrologic modeling and gravity observations show the effectiveness of identifying the critical components that make up TWS. Understanding the spatially heterogeneous drivers of observed TWS decline allows us to translate satellite observations into policy-relevant information. Because this functionality is built within a process-based hydrological model, future projections can illuminate those aspects of the hydrological cycle that require additional attention by decision makers to ensure adequate water resources are available for all.

  PublisherOA PDFDOI

• Community estimate of global glacier mass changes from 2000 to 2023

  Nature · 2025-02-19 · 142 citations

  articleOpen access

  Abstract Glaciers are indicators of ongoing anthropogenic climate change 1 . Their melting leads to increased local geohazards 2 , and impacts marine 3 and terrestrial 4,5 ecosystems, regional freshwater resources 6 , and both global water and energy cycles 7,8 . Together with the Greenland and Antarctic ice sheets, glaciers are essential drivers of present 9,10 and future 11–13 sea-level rise. Previous assessments of global glacier mass changes have been hampered by spatial and temporal limitations and the heterogeneity of existing data series 14–16 . Here we show in an intercomparison exercise that glaciers worldwide lost 273 ± 16 gigatonnes in mass annually from 2000 to 2023, with an increase of 36 ± 10% from the first (2000–2011) to the second (2012–2023) half of the period. Since 2000, glaciers have lost between 2% and 39% of their ice regionally and about 5% globally. Glacier mass loss is about 18% larger than the loss from the Greenland Ice Sheet and more than twice that from the Antarctic Ice Sheet 17 . Our results arise from a scientific community effort to collect, homogenize, combine and analyse glacier mass changes from in situ and remote-sensing observations. Although our estimates are in agreement with findings from previous assessments 14–16 at a global scale, we found some large regional deviations owing to systematic differences among observation methods. Our results provide a refined baseline for better understanding observational differences and for calibrating model ensembles 12,16,18 , which will help to narrow projection uncertainty for the twenty-first century 11,12,18 .

  PublisherDOI

• Uncertainty Quantification of Satellite-based Essential Climate Variables Derived from Deep Learning – Examples of Snow Cover and Terrestrial Water Storage

  Repository for Publications and Research Data (ETH Zurich) · 2025-12-01

  otherOpen access

  Accurate uncertainty information associated with essential climate variables (ECVs) is crucial for reliable climate modeling, understanding the spatiotemporal evolution of the Earth system, and is critical for responsible decision-making. Recent developments in deep learning offer promising opportunities to provide ECV products with higher accuracy. However, the quantification of uncertainties associated with outputs of such deep learning models has yet to be widely adopted. In this contribution, we start by clarifying and bridging the gap between conventional geodetic and deep learning views on uncertainty quantification, by providing rigorous definitions of aleatoric (data) and epistemic (model) uncertainties. Then, we introduce selected approaches to quantify the two uncertainties associated with deep learning outputs. We emphasize the importance of considering the uncertainty associated with inputs in deep learning models to account for the dynamic and multifaceted nature of satellite observations. We further demonstrated our findings by applying selected uncertainty quantification approaches to Sentinel-3 and GRACE(-FO) data to derive snow cover fraction (SCF) and terrestrial water storage changes (TWSC), respectively. In the SCF application, we discuss the different sources and characteristics of aleatoric and epistemic uncertainties within a deep ensemble model. Then, various uncertainty quantification approaches, including Bayesian neural networks, Monte Carlo dropouts, deep ensembles, and deep ensembles with Monte Carlo simulations, are discussed in the TWSC application. We highlight the contribution of the deep ensembles with Monte Carlo simulations, as this method can explicitly consider the uncertainties associated with input features to provide more reliable aleatoric uncertainties.

  PublisherDOI

• Improved understanding of Okavango Delta inundation using GRACE-derived terrestrial water storage

  Environmental Research Water · 2025-11-06

  articleOpen accessCorresponding

  Hydrologic contributions from the Angolan Highlands sustain the Okavango Delta, one of the world’s largest inland wetlands. Observations show a clear link between the timing, intensity, and seasonal distribution of precipitation in the Highlands and flood extent in the Delta. However, the role of upstream water storage remains poorly understood, largely due to limited and uneven hydrological data. Here, we combine satellite-derived terrestrial water storage anomalies from the NASA Gravity Recovery and Climate Experiment (GRACE) and follow-on (GRACE-FO) missions with precipitation from the Climate Hazards Center Infrared Precipitation with Station dataset to assess interannual drivers of maximum flood extent in the Okavango Delta from 2003–2024. We show that GRACE-based terrestrial water storage explains 62% of the variance in annual inundation extent (90% CI: 32%–83%), outperforming precipitation alone (52%, 90% CI: 18%–80%), and that combining both explanatory variables increases explained variance to 73% (90% CI: 52%–89%). Spatial analyses of terrestrial water storage identify the Angolan Highlands as the key source region to the Delta, consistent with prior hydrological and modeling studies. These findings provide the first demonstration, to our knowledge, of GRACE-based terrestrial water storage improving flood modeling in the Okavango Delta and motivate future work to better monitor water towers driven by both precipitation and subsurface aquifer contributions.

  PublisherDOI

• Global Warming Has Accelerated: Are the United Nations and the Public Well-Informed?

  Environment Science and Policy for Sustainable Development · 2025-01-02 · 142 citations

  articleOpen access

  Global temperature leaped more than 0.4°C (0.7°F) during the past two years, the 12-month average peaking in August 2024 at +1.6°C relative to the temperature at the beginning of last century (the 1880-1920 average). This temperature jump was spurred by one of the periodic tropical El Niño warming events, but many Earth scientists were baffled by the magnitude of the global warming, which was twice as large as expected for the weak 2023-2024 El Niño. We find that most of the other half of the warming was caused by a restriction on aerosol emissions by ships, which was imposed in 2020 by the International Maritime Organization to combat the effect of aerosol pollutants on human health. Aerosols are small particles that serve as cloud formation nuclei. Their most important effect is to increase the extent and brightness of clouds, which reflect sunlight and have a cooling effect on Earth. When aerosols – and thus clouds – are reduced, Earth is darker and absorbs more sunlight, thus enhancing global warming. Ships are the main aerosol source in the North Pacific and North Atlantic Oceans. We quantify the aerosol effect from the geographical distribution of sunlight reflected by Earth as measured by satellites, with the largest expected and observed effects in the North Pacific and North Atlantic Oceans. We find that aerosol cooling, and thus climate sensitivity, are understated in the best estimate of the United Nations Intergovernmental Panel on Climate Change (IPCC). Global warming caused by reduced ship aerosols will not go away as tropical climate moves into its cool La Niña phase. Therefore, we expect that global temperature will not fall much below +1.5°C level, instead oscillating near or above that level for the next few years, which will help confirm our interpretation of the sudden global warming. High sea surface temperatures and increasing ocean hotspots will continue, with harmful effects on coral reefs and other ocean life. The largest practical effect on humans today is increase of the frequency and severity of climate extremes. More powerful tropical storms, tornadoes, and thunderstorms, and thus more extreme floods, are driven by high sea surface temperature and a warmer atmosphere that holds more water vapor. Higher global temperature also increases the intensity of heat waves and – at the times and places of dry weather – high temperature increases drought intensity, including “flash droughts” that develop rapidly, even in regions with adequate average rainfall. Polar climate change has the greatest long-term effect on humanity, with impacts accelerated by the jump in global temperature. We find that polar ice melt and freshwater injection onto the North Atlantic Ocean exceed prior estimates and, because of accelerated global warming, the melt will increase. As a result, shutdown of the Atlantic Meridional Overturning Circulation (AMOC) is likely within the next 20-30 years, unless actions are taken to reduce global warming – in contradiction to conclusions of IPCC. If AMOC is allowed to shut down, it will lock in major problems including sea level rise of several meters – thus, we describe AMOC shutdown as the “point of no return.” We suggest that an alternative perspective – a complement to the IPCC approach – is needed to assess these issues and actions that are needed to avoid handing young people a dire situation that is out of their control. This alternative approach will make more use of ongoing observations to drive modeling and more use of paleoclimate to test modeling and test our understanding. As of today, the threats of AMOC shutdown and sea level rise are poorly understood, but better observations of polar ocean and ice changes in response to the present accelerated global warming have the potential to greatly improve our understanding. Read the transcript Watch the video on Vimeo © 2025 The Author(s). Published with license by Taylor & Francis Group, LLC

  PublisherDOI

• An iterative method combining GRACE/GRACE- FO, ICESat/ICESat-2, GPS, and SMB from climate model outputs to improve Antarctic mass balance and GIA

  2024-03-08

  preprintOpen accessCorresponding

  Consensus from mass balance studies published in the past decade revealed an increase in mass loss from the Antarctic Ice Sheet. However, independent methods (e.g., gravity, altimetry, and mass budget) for determining the rate of mass loss disagree about the magnitude and uncertainty of estimates for East Antarctica. This discrepancy is largely due to the size of East Antarctica, the low signal-to-noise ratios of existing sensors, and the confluence of uncertainties for each method. As a result, ice sheet models may not be well-constrained to make future projections of ice mass change and sea level change. Here, we combine GRACE/GRACE-FO, ICESat/ICESat-2, and GPS observations and extend Velicogna et al. (2002) iterative algorithm to improve the estimate of the Antarctic mass balance. We add outputs from surface mass balance (SMB) and firn models to satellite altimetry, time-variable gravity, and GPS measurements. We evaluate the sensitivity of the approach to realistic ice mass spatial and time distributions using an effective density map derived from ICESat-2 data, and reconstructions of SMB and firn depth from climate model outputs. Our result provides information on how the GPS network could be completed to resolve major uncertainties, obtain a better glacial isostatic adjustment (GIA), and a better estimate of ice mass balance.

  PublisherDOI

• Improving ice sheet and glaciers mass balance estimates using Microwave Instrument (MWI) GRACE-FO and GRACE data 

  2024-09-16

  preprintOpen access1st authorCorresponding

  The ice mass balance of the Antarctic and Greenland Ice Sheets and the World's Glaciers and Ice Caps (GIC) has a major impact on sea level rise. Here we present an update on monthly estimates of glaciers and ice sheet mass balance from the GRACE/GRACE-FO mission up to 2024 using the Microwave Instrument (MWI). We report a slow-down in mass loss in Greenland after 2012 and a mass gain in Antarctica in 2021-2023. In recent years, we note a pause in the mass loss of Antarctica, a mass loss in Greenland that is no longer accelerating, and a steady loss from the GIC. The ice sheet mass loss has slowed down to average 255 Gt/yr for Greenland and 108 Gt/yr for Antarctica during 2002-2024. Combining the GRACE/GRACE-FO data with a longer-time record for 1979-2024, however, we find that the ice sheet mass loss has been accelerating for both Greenland and Antarctica, i.e., the recent slowdown is a fluctuation on top of a long-term decrease.

  PublisherDOI

• Mass balance of the Greenland and Antarctic ice sheets from 1992 to 2020

  Earth system science data · 2023-04-20 · 285 citations

  articleOpen accessCorresponding

  Abstract. Ice losses from the Greenland and Antarctic ice sheets have accelerated since the 1990s, accounting for a significant increase in the global mean sea level. Here, we present a new 29-year record of ice sheet mass balance from 1992 to 2020 from the Ice Sheet Mass Balance Inter-comparison Exercise (IMBIE). We compare and combine 50 independent estimates of ice sheet mass balance derived from satellite observations of temporal changes in ice sheet flow, in ice sheet volume, and in Earth's gravity field. Between 1992 and 2020, the ice sheets contributed 21.0±1.9 mm to global mean sea level, with the rate of mass loss rising from 105 Gt yr−1 between 1992 and 1996 to 372 Gt yr−1 between 2016 and 2020. In Greenland, the rate of mass loss is 169±9 Gt yr−1 between 1992 and 2020, but there are large inter-annual variations in mass balance, with mass loss ranging from 86 Gt yr−1 in 2017 to 444 Gt yr−1 in 2019 due to large variability in surface mass balance. In Antarctica, ice losses continue to be dominated by mass loss from West Antarctica (82±9 Gt yr−1) and, to a lesser extent, from the Antarctic Peninsula (13±5 Gt yr−1). East Antarctica remains close to a state of balance, with a small gain of 3±15 Gt yr−1, but is the most uncertain component of Antarctica's mass balance. The dataset is publicly available at https://doi.org/10.5285/77B64C55-7166-4A06-9DEF-2E400398E452 (IMBIE Team, 2021).

  PublisherOA PDFDOI

Recent grants

• Collaborative Research: Calculation of Antarctic Gravity Field from GRACE Satellite Data and Comparison with Independent Measurements

  NSF · $205k · 2004–2008

• Collaborative Research: Calculation of Antarctic Gravity Field from GRACE Satellite Data and Comparison with Independent Measurements

  NSF · $57k · 2008–2009

Frequent coauthors

• Eric Rignot

  University of California, Irvine

  161 shared

• John Wahr

  94 shared

• Tyler Sutterley

  Seattle University

  93 shared

• M. R. van den Broeke

  Utrecht University

  73 shared

• Yara Mohajerani

  Irvine University

  61 shared

• J. Mouginot

  Institut polytechnique de Grenoble

  55 shared

• P. Kishore

  45 shared

• John S. Kimball

  39 shared

Labs

• Isabella Velicogna's LabPI

  Research on terrestrial hydrological cycles, vegetation response to water supply changes, mass loss in Polar Regions, and glacial isostatic adjustment of the Earth.