Jump to Content

Climate and sustainability

Leveraging machine learning (ML) and artificial intelligence (AI) to address climate change and help build a sustainable future for all.

Landscape with clouds in the background

Climate and sustainability

About the team

Climate change is an urgent threat to humanity with far-reaching societal and economic consequences. Our climate and sustainability teams harness the power of AI to address this challenge, as part of Google's long-standing commitment to climate action.

AI has expanded the type of information we can apply to deep computing power and unlocked an increasing number of methods for interpreting information and creating groundbreaking innovations. It enables us to identify previously impossible solutions to climate challenges — from reducing greenhouse gas emissions in cities to building models that improve our ability to predict and respond to climate-driven natural disasters.

Google Research is leading multiple climate and sustainability efforts in various stages of development, and we continue to explore innovations to accelerate progress in this space. As part of our mission to build a safer, more sustainable future for all, we collaborate closely with cities, governments, startups and aid organizations.

Our Climate and Sustainability team is global, with researchers, partners and projects spanning the US, Europe, Africa, Latin America and Asia. Learn more below about how we are using AI to improve the lives of billions of people worldwide.

Team focus summaries


AI innovations that help partners and individuals reduce anthropogenic greenhouse gas in the atmosphere, including AI applications in transportation, energy, and carbon removal.

Adaptation & resilience

Helping communities, non-governmental organizations (NGOs) and governments predict and address extreme weather events, food insecurity, and other effects of climate change.


Advanced statistical and machine learning-based atmospheric models to improve weather and climate predictions.

Featured publications

A Neural Encoder for Earthquake Rate Forecasting
Oleg Zlydenko
Brendan Meade
Alexandra Sharon Molchanov
Sella Nevo
Yohai bar Sinai
Scientific Reports (2023)
Preview abstract Forecasting the timing of earthquakes is a long-standing challenge. Moreover, it is still debated how to formulate this problem in a useful manner, or to compare the predictive power of different models. Here, we develop a versatile neural encoder of earthquake catalogs, and apply it to the fundamental problem of earthquake rate prediction, in the spatio-temporal point process framework. The epidemic type aftershock sequence model (ETAS) effectively learns a small number of parameters to constrain assumed functional forms for the space and time relationships of earthquake sequences (e.g., Omori-Utsu law). Here we introduce learned spatial and temporal embeddings for point process earthquake forecast models that capture complex correlation structures. We demonstrate the generality of this neural representation as compared with ETAS model using train-test data splits and how it enables the incorporation of additional geophysical information. In rate prediction tasks, the generalized model shows > 4% improvement in information gain per earthquake and the simultaneous learning of anisotropic spatial structures analogous to fault traces. The trained network can be also used to perform short-term prediction tasks, showing similar improvement while providing a 1,000-fold reduction in run-time. View details
Flood forecasting with machine learning models in an operational framework
Asher Metzger
Chen Barshai
Dana Weitzner
Frederik Kratzert
Gregory Begelman
Guy Shalev
Hila Noga
Moriah Royz
Niv Giladi
Ronnie Maor
Sella Nevo
Yotam Gigi
HESS (2022)
Preview abstract Google’s operational flood forecasting system was developed to provide accurate real-time flood warnings to agencies and the public, with a focus on riverine floods in large, gauged rivers. It became operational in 2018 and has since expanded geographically. This forecasting system consists of four subsystems: data validation, stage forecasting, inundation modeling, and alert distribution. Machine learning is used for two of the subsystems. Stage forecasting is modeled with the Long Short-Term Memory (LSTM) networks and the Linear models. Flood inundation is computed with the Thresholding and the Manifold models, where the former computes inundation extent and the latter computes both inundation extent and depth. The Manifold model, presented here for the first time, provides a machine-learning alternative to hydraulic modeling of flood inundation. When evaluated on historical data, all models achieve sufficiently high-performance metrics for operational use. The LSTM showed higher skills than the Linear model, while the Thresholding and Manifold models achieved similar performance metrics for modeling inundation extent. During the 2021 monsoon season, the flood warning system was operational in India and Bangladesh, covering flood-prone regions around rivers with a total area of 287,000 km2, home to more than 350M people. More than 100M flood alerts were sent to affected populations, to relevant authorities, and to emergency organizations. Current and future work on the system includes extending coverage to additional flood-prone locations, as well as improving modeling capabilities and accuracy. View details
AI Increases Global Access to Reliable Flood Forecasts
Asher Metzger
Dana Weitzner
Frederik Kratzert
Guy Shalev
Martin Gauch
Sella Nevo
Shlomo Shenzis
Tadele Yednkachw Tekalign
Vusumuzi Dube
arXiv (2023)
Preview abstract Floods are one of the most common natural disasters, with a disproportionate impact in developing countries that often lack dense streamflow gauge networks. Accurate and timely warnings are critical for mitigating flood risks, but hydrological simulation models typically must be calibrated to long data records in each watershed. Here we show that AI-based forecasting achieves reliability in predicting extreme riverine events in ungauged watersheds at up to a 5-day lead time that is similar to or better than the reliability of nowcasts (0-day lead time) from a current state of the art global modeling system (the Copernicus Emergency Management Service Global Flood Awareness System). Additionally, we achieve accuracies over 5-year return period events that are similar to or better than current accuracies over 1-year return period events. This means that AI can provide flood warnings earlier and over larger and more impactful events in ungauged basins. The model developed in this paper was incorporated into an operational early warning system that produces publicly available (free and open) forecasts in real time in over 80 countries. This work highlights a need for increasing the availability of hydrological data to continue to improve global access to reliable flood warnings. View details
Preview abstract Contrails (condensation trails) are the ice clouds that trail behind aircraft as they fly through cold and moist regions of the atmosphere. Avoiding these regions could potentially be an inexpensive way to reduce over half of aviation's impact on global warming. Development and evaluation of these avoidance strategies greatly benefits from the ability to detect contrails on satellite imagery. Since little to no public data is available to develop such contrail detectors, we construct and release a dataset of several thousand Landsat-8 scenes with pixel-level annotations of contrails. The dataset will continue to grow, but currently contains 3431 scenes (of which 47\% have at least one contrail) representing 800+ person-hours of labeling time. View details
Preview abstract The majority of IPCC scenarios call for active CO2 removal (CDR) to remain below 2ºC of warming. On geological timescales, ocean uptake regulates atmospheric CO2 concentration, with two homeostats driving sequestration: dissolution of deep ocean calcite deposits and terrestrial weathering of silicate rocks, acting on 1ka to 100ka timescales. Many current ocean-based CDR proposals effectively act to accelerate the latter. Here we present a method which relies purely on the redistribution and dilution of acidity from a thin layer of the surface ocean to a thicker layer of deep ocean, with the aim of accelerating the former carbonate homeostasis. This downward transport could be seen analogous to the action of the natural biological carbon pump. The method offers advantages over other ocean CDR methods and direct air capture approaches (DAC): the conveyance of mass is minimized (acidity is pumped in situ to depth), and expensive mining, grinding and distribution of alkaline material is eliminated. No dilute substance needs to be concentrated, avoiding the Sherwood’s Rule costs typically encountered in DAC. Finally, no terrestrial material is added to the ocean, avoiding significant alteration of seawater ion concentrations and issues with heavy metal toxicity encountered in mineral-based alkalinity schemes. The artificial transport of acidity accelerates the natural deep ocean invasion and subsequent compensation by calcium carbonate. It is estimated that the total compensation capacity of the ocean is on the order of 1500GtC. We show through simulation that pumping of ocean acidity could remove up to 150GtC from the atmosphere by 2100 without excessive increase of local ocean pH. For an acidity release below 2000m, the relaxation half time of CO2 return to the atmosphere was found to be ~2500 years (~1000yr without accounting for carbonate dissolution), with ~85% retained for at least 300 years. The uptake efficiency and residence time were found to vary with the location of acidity pumping, and optimal areas were calculated. Requiring only local resources (ocean water and energy), this method could be uniquely suited to utilize otherwise-stranded open ocean energy sources at scale. We examine technological pathways that could be used to implement it and present a brief techno-economic estimate of 130-250$/tCO2 at current prices and as low as 86$/tCO2 under modest learning-curve assumptions. View details
Preview abstract TAE Technologies, Inc. (TAE) is pursuing an alternative approach to magnetically confined fusion, which relies on field-reversed configuration (FRC) plasmas composed of mostly energetic and well-confined particles by means of a state-of-the-art tunable energy neutral-beam (NB) injector system. TAE’s current experimental device, C-2W (also called “Norman”), is the world’s largest compact-toroid device and has made significant progress in FRC performance, producing record breaking, high temperature (electron temperature, Te >500 eV; total electron and ion temperature, Ttot >3 keV) advanced beam-driven FRC plasmas, dominated by injected fast particles and sustained in steady-state for up to 30 ms, which is limited by NB pulse duration. C-2W produces significantly better FRC performance than the preceding C-2U experiment, in part due to Google’s machine-learning framework for experimental optimization, which has contributed to the discovery of a new operational regime where novel settings for the formation sections yield consistently reproducible, hot, and stable plasmas. Active plasma control system has been developed and utilized in C-2W to produce consistent FRC performance as well as for reliable machine operations using magnets, electrodes, gas injection, and tunable NBs. The active control system has demonstrated a stabilization of FRC axial instability. Overall FRC performance is well correlated with NBs and edge-biasing system, where higher total plasma energy is obtained with increasing both NB injection power and applied-voltage on biasing electrodes. C-2W divertors have demonstrated a good electron heat confinement on open-field-lines using strong magnetic mirror fields as well as expanding the magnetic field in the divertors (expansion ratio >30); the electron energy lost per ion, ~6–8, is achieved, which is close to the ideal theoretical minimum. View details
Preview abstract This work presents a high-fidelity simulation framework for modeling large-scale wildfire scenarios that take into consideration realistic topographies, atmospheric conditions, turbulence/fire interaction, and flow dynamics. With the overall goal of enabling large-scale ensemble simulations and the integration of the simulation results into machine-learning applications, this modeling framework has been implemented into the TensorFlow programming environment. To demonstrate the capability of this simulation framework in predicting large-scale fires, we performed high-resolution simulations of a realistic wildfire scenario that is representative of the 2017 Tubbs fire. View details
Estimates of broadband upwelling irradiance from GOES-16 ABI
Sixing Chen
Vincent Rudolf Meijer
Joe Ng
Geoff Davis
Carl Elkin
Remote Sensing of Environment, vol. 285 (2023)
Preview abstract Satellite-derived estimates of the Earth’s radiation budget are crucial for understanding and predicting the weather and climate. However, existing satellite products measuring broadband outgoing longwave radiation (OLR) and reflected shortwave radiation (RSR) have spatio-temporal resolutions that are too coarse to evaluate important radiative forcers like aircraft condensation trails. We present a neural network which estimates OLR and RSR based on narrowband radiances, using collocated Cloud and Earth’s Radiant Energy System (CERES) and GOES-16 Advanced Baseline Imager (ABI) data. The resulting estimates feature strong agreement with the CERES data products (R^2 = 0.977 for OLR and 0.974 for RSR on CERES Level 2 footprints), and we provide open access to the collocated satellite data and model outputs on all available GOES-16 ABI data for the 4 years from 2018–2021. View details
Multimodal contrastive learning for remote sensing tasks
Umangi Jain
Alex Wilson
Self-Supervised Learning - Theory and Practice, NeurIPS 2022 Workshop
Preview abstract Self-supervised methods have shown tremendous success in the field of computer vision, including subfields like remote sensing and medical imaging. Most popular contrastive-loss based methods like SimCLR, MoCo, MoCo-v2 use multiple views of the same image by applying contrived augmentations on the image to create positive pairs and contrast them with negative examples. Although these techniques work well, most of these techniques have been tuned on ImageNet (and similar computer vision datasets). While there have been some attempts to capture a richer set of deformations in the positive samples, in this work, we explore a promising alternative to generating positive examples for remote sensing data within the contrastive learning framework. Images captured from different sensors at the same location and nearby timestamps can be thought of as strongly augmented instances of the same scene, thus removing the need to explore and tune a set of hand crafted strong augmentations. In this paper, we propose a simple dual-encoder framework, which is pre-trained on a large unlabeled dataset (~1M) of Sentinel-1 and Sentinel-2 image pairs. We test the embeddings on two remote sensing downstream tasks: flood segmentation and land cover mapping, and empirically show that embeddings learnt from this technique outperforms the conventional technique of collecting positive examples via aggressive data augmentations. View details
Next Day Wildfire Spread: A Machine Learning Dataset to Predict Wildfire Spreading From Remote-Sensing Data
Fantine Huot
Lily Hu
Tharun Pratap Sankar
Matthias Ihme
Yi-fan Chen
IEEE Transactions on Geoscience and Remote Sensing, vol. 60 (2022), pp. 1-13
Preview abstract Predicting wildfire spread is critical for land management and disaster preparedness. To this end, we present “Next Day Wildfire Spread,” a curated, large-scale, multivariate dataset of historical wildfires aggregating nearly a decade of remote-sensing data across the United States. In contrast to existing fire datasets based on Earth observation satellites, our dataset combines 2-D fire data with multiple explanatory variables (e.g., topography, vegetation, weather, drought index, and population density) aligned over 2-D regions, providing a feature-rich dataset for machine learning. To demonstrate the usefulness of this dataset, we implement a neural network that takes advantage of the spatial information of these data to predict wildfire spread. We compare the performance of the neural network with other machine learning models: logistic regression and random forest. This dataset can be used as a benchmark for developing wildfire propagation models based on remote-sensing data for a lead time of one day. View details