The planet is currently experiencing an unparalleled environmental crisis, driven by anthropogenic activities that have profoundly altered its natural systems. Climatic changes, manifested through rising global temperatures, more frequent and intense extreme weather events, sea-level rise, and biodiversity loss, pose existential threats to ecosystems and human societies alike. Concurrently, a myriad of environmental risks, including widespread pollution, resource depletion, and habitat destruction, further exacerbate these challenges, demanding urgent and effective interventions. Traditional scientific methodologies, while foundational, often grapple with the overwhelming complexity, sheer volume, and inherent interconnectedness of environmental data, limiting their capacity for comprehensive analysis, precise prediction, and agile management strategies.

In response to this pressing need, Artificial Intelligence (AI) has rapidly advanced from a theoretical concept to a practical suite of technologies, offering unprecedented capabilities to process vast datasets, identify intricate patterns, make robust predictions, and optimize complex systems [1]. The convergence of big data from diverse environmental sensors, satellite imagery, climate models, and socio-economic indicators with sophisticated AI algorithms presents a paradigm shift in our ability to understand, monitor, and mitigate the impacts of climatic change and environmental risks.

This review paper aims to provide a detailed and critical overview of the applications of AI technology and tools in two primary domains: the analysis of climatic changes and the management of environmental risks. It synthesizes current research and practical implementations, highlighting the transformative potential of AI while also acknowledging its significant challenges and limitations. By systematically exploring AI's contributions across various environmental sectors, this paper seeks to inform researchers, policymakers, and practitioners about the state-of-the-art in this rapidly evolving field and to catalyze further innovation towards a more sustainable future.

Climate change is unequivocally linked to human activities, primarily the emission of greenhouse gases (GHGs) from the burning of fossil fuels, industrial processes, deforestation, and agriculture. The resulting increase in global average temperatures leads to a cascade of impacts, including more frequent and intense heatwaves, altered precipitation patterns, accelerated melting of glaciers and ice sheets, sea-level rise, ocean acidification, and shifts in species distribution [2]. These changes profoundly affect natural ecosystems, human health, food security, water resources, and economic stability worldwide.

Environmental risks encompass a broader spectrum of threats to natural environments and human well-being, often exacerbated by a changing climate. These include the following incidents.

Floods, droughts, wildfires, hurricanes, typhoons, landslides, and tsunamis, whose frequency and intensity are often influenced by climate change [3].

Air, water, and soil contamination by industrial effluents, agricultural runoff, plastics, and other harmful substances, leading to public health crises and ecosystem degradation [4].

Habitat destruction, overexploitation, pollution, invasive species, and climate change driving species extinction rates far above natural levels, compromising ecosystem services [5].

Over-extraction of vital resources such as freshwater, fertile land, and minerals, threatening will become a part of long-term sustainability.

Analyzing these phenomena and managing their associated risks is inherently complex due to several factors as described in the below.

Environmental data originates from diverse sources (satellites, ground sensors, historical records, climate models, social media), varying in format, scale, and reliability.

Environmental systems are characterized by intricate feedback loops, non-linear dynamics, and emergent properties, making them difficult to model and predict using traditional linear approaches [6].

Environmental processes unfold across vast geographical areas and over long timescales, requiring models capable of capturing both localized detail and global trends.

Inherent uncertainties in climate projections, natural variability, and socio-economic factors complicate decision-making and risk assessment.

These challenges highlight the critical need for advanced analytical tools that can transcend the limitations of conventional methods, providing a fertile ground for the application of AI.

Artificial Intelligence (AI) is a broad field of computer science dedicated to creating machines capable of performing tasks that typically require human intelligence [7]. For environmental applications, several key sub-fields and technologies of AI are particularly relevant:

Machine Learning (Ml)

ML is a subset of AI that enables systems to learn from data, identify patterns, and make decisions with minimal human intervention.

Algorithms learn from labeled datasets (input-output pairs) to predict outcomes. Examples include regression for continuous variables (e.g., predicting temperature) and classification for categorical variables (e.g., identifying land cover types) [8]. Common algorithms include Support Vector Machines (SVMs), Random Forests, and Gradient Boosting.

Algorithms identify patterns and structures in unlabeled data. This is useful for clustering similar data points (e.g., identifying distinct climate regions) or reducing data dimensionality (e.g., Principal Component Analysis) [9].

Agents learn to make sequences of decisions by interacting with an environment, receiving rewards or penalties. While less common in direct environmental modeling, it has potential in optimizing resource management or energy systems [10].

Deep Learning (Dl)

DL is a sub-field of ML that uses artificial neural networks (ANNs) with multiple layers (deep networks) to learn complex representations from data [11]. DL has revolutionized fields like computer vision and natural language processing.

Convolutional Neural Networks (CNNs) are excellent for processing grid-like data such as images and geospatial data (e.g., satellite imagery), adept at identifying spatial features and patterns [12].

Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks are designed to handle sequential data, making them ideal for time-series analysis (e.g., climate trajectories, weather forecasting) [13].

Generative Adversarial Networks (GANs) are used for generating new data instances that resemble the training data, useful for data augmentation or creating synthetic climate scenarios [14].

Transformers are initially developed for NLP, these models excel at capturing long-range dependencies in sequential data and are increasingly applied to time-series and even image data, showing promise for complex climatic simulations [15].

Computer Vision

Computer vision enables machines to "see" and interpret visual information from images and videos. This is crucial for analyzing remote sensing data, monitoring wildlife, and detecting environmental changes from aerial or satellite platforms [16].

Natural Language Processing (NLP)

NLP focuses on enabling computers to understand, interpret, and generate human language. In environmental contexts, NLP can be used to analyze scientific literature, policy documents, social media discussions related to climate change, and synthesize vast amounts of textual information [17].

Explainable AI (XAI)

As AI models become more complex, their decision-making processes can become opaque ("black boxes"). XAI aims to make AI models more transparent and understandable, crucial for building trust and ensuring accountability, especially in high-stakes environmental policy decisions [18].

Big Data Analytics

While not an AI technique itself, big data analytics provides the infrastructure and methodologies to collect, store, process, and analyze the massive, diverse, and rapidly growing datasets that fuel AI applications in environmental science [19].

These AI technologies, when strategically applied, offer powerful capabilities to address the complexities of climate change analysis and environmental risk management.

AI Applications in Climate Change Analysis

The vast and intricate datasets generated by climate observations, models, and remote sensing platforms provide a rich environment for AI technologies to uncover hidden patterns, improve predictive accuracy, and synthesize complex information related to climate change.

Climate Modeling and Prediction

Traditional Earth System Models (ESMs) and General Circulation Models (GCMs) are computationally intensive and often struggle with resolving fine-scale processes or accurately representing complex feedback loops [20]. AI offers avenues to enhance these models.

ML techniques can learn complex relationships from high-resolution simulations or observational data to develop more accurate parameterizations for sub-grid scale processes (e.g., clouds, convection) within coarser-resolution climate models [21]. This can significantly reduce computational cost while maintaining or improving accuracy.

Global climate models provide projections at coarse spatial resolutions (tens to hundreds of kilometers). DL models, particularly CNNs and GANs, are highly effective at statistical and dynamical downscaling, translating these coarse projections into fine-scale regional impacts vital for local adaptation planning (e.g., predicting local temperature and precipitation at a few kilometers resolution) [22,23].

RNNs and LSTMs excel at time-series prediction and have been employed to forecast the intensity and trajectory of extreme events such as hurricanes, tropical cyclones, heatwaves, and droughts with greater lead times and accuracy than traditional methods [24,25]. For example, deep learning models can analyze satellite imagery and atmospheric data to predict hurricane intensification.

AI can help disentangle the human influence on specific extreme weather events from natural variability. ML classifiers can identify the "fingerprints" of anthropogenic forcing in climate data, aiding in attributing the likelihood and severity of events to climate change [26].

ML algorithms can assimilate diverse observational data into climate models, improving initial conditions and reducing forecast errors. They are also used for bias correction of climate model outputs, making them more reliable for impact assessments [27].

Remote Sensing and Geospatial Data Analysis

Satellite imagery, drone data, and other remote sensing platforms generate enormous volumes of geospatial data, making them ideal for AI-driven analysis of Earth's surface and atmosphere.

CNNs are extensively used to analyze satellite imagery (e.g., Sentinel, Landsat data) to classify land cover, detect deforestation, monitor reforestation efforts, and track urban expansion, providing critical insights into carbon emissions and habitat loss [22].

AI-powered image analysis can accurately delineate glacier boundaries, measure ice flow rates, detect calving events, and estimate ice volume changes over time, crucial for understanding sea-level rise and freshwater reserves [28].

ML algorithms process altimetry data from satellites to monitor changes in sea surface height, wave patterns, and coastal erosion, helping predict future sea-level rise impacts and coastal vulnerability [29].

AI can invert satellite radiance measurements to estimate concentrations of greenhouse gases (CO2, CH4) and other pollutants (e.g., NO2, SO2) with higher precision and spatial resolution, identifying emission hotspots and monitoring compliance with climate agreements [30].

ML assists in mapping sea surface temperatures (SST), ocean currents, chlorophyll concentrations, and ocean heat content, which are vital indicators for marine ecosystem health, coral bleaching, and fisheries management [31].

AI-driven image processing identifies diverse species from remote sensing data, drone imagery, or camera traps, mapping their habitats, tracking migration patterns, and assessing ecosystem health at unprecedented scales [32].

Data Synthesis and Knowledge Discovery

The complexity of climate change necessitates integrating and making sense of disparate data sources.

ML algorithms can fuse data from satellites, ground sensors, climate models, social science surveys, and economic indicators to create comprehensive environmental intelligence platforms, enabling holistic analysis of climate impacts and adaptation strategies [27].

Unsupervised learning techniques can uncover non-obvious correlations and causal links between climate variables, socio-economic factors, and environmental outcomes, leading to new scientific insights (e.g., identifying climate teleconnections or drivers of vulnerability) [33].

 NLP techniques can process vast repositories of scientific publications, policy documents, and reports to identify key themes summarize findings, and track research trends related to climate change, accelerating knowledge discovery and policy formulation [34].

Beyond analyzing climate change itself, AI plays a crucial role in predicting, mitigating, and managing the myriad environmental risks exacerbated by a changing climate.

Disaster Preparedness and Early Warning Systems

AI is transforming our capacity to anticipate and respond to natural disasters, minimizing loss of life and property.

Predicting Natural Disaster Intensity and Trajectories

ML models, trained on meteorological data, fuel moisture, topography, and historical fire patterns, predict fire risk, spread velocity, and intensity, informing proactive evacuation and resource deployment [35]. Computer vision analyzes satellite or drone imagery for early fire detection.

Deep learning models integrate hydrological data, rainfall forecasts, terrain elevation, soil moisture, and real-time sensor data to predict flood inundation areas and landslide susceptibility with high spatial and temporal resolution [36].

AI models combine meteorological forecasts with coastal bathymetry and tide information to predict storm surge heights and coastal flooding, crucial for timely warnings and evacuation orders [37].

RL and optimization algorithms can dynamically allocate emergency services (firefighters, medical teams, aid supplies) based on real-time disaster predictions, assessing optimal routes and resource needs to maximize effectiveness and minimize response times [38].

ML can integrate demographic, infrastructural, and environmental data to create detailed vulnerability maps, identifying populations and assets most at risk from specific hazards, guiding targeted resilience-building efforts [39].

Edge AI devices combined with IoT sensors can provide real-time data on environmental conditions (e.g., water levels, seismic activity, air quality), enabling immediate alerts and facilitating rapid intervention during unfolding crises [40].

Pollution Monitoring and Mitigation

AI offers powerful tools for tracking, understanding, and combating various forms of environmental pollution.

ML models analyze data from ground sensors, satellite imagery, and meteorological forecasts to predict air pollutant concentrations (e.g., PM2.5, ozone, NOX) at fine spatial and temporal scales, enabling public health warnings and targeted emission control strategies [41]. They can also identify pollution sources.

AI algorithms process sensor data (pH, turbidity, dissolved oxygen, chemical composition) and satellite imagery to detect water contamination, predict harmful algal blooms (HABs), and monitor the health of aquatic ecosystems, supporting proactive water resource management [42].

AI-driven computer vision systems are used in recycling facilities to automatically sort waste more efficiently, improving recycling rates [43]. RL can optimize waste collection routes, reducing fuel consumption and emissions. Predictive analytics helps manage landfill capacity and identify illegal dumping sites.

ML techniques can trace the origins of pollutants by analyzing their chemical signatures and dispersion patterns in air, water, or soil, facilitating enforcement and targeted mitigation efforts [44].

Biodiversity Conservation and Ecosystem Management

AI is revolutionizing how we monitor, protect, and restore biodiversity.

Species Identification and Tracking

Computer vision analyzes millions of camera trap images to automatically identify animal species, count populations, and monitor behavior, overcoming the limitations of manual review [32].

AI, particularly deep learning for audio classification, processes vast amounts of acoustic data from remote sensors to identify specific animal calls, detect illegal logging or poaching, and assess habitat health [45].

Drones equipped with AI can conduct aerial surveys to count populations of large animals, monitor vegetation health, and detect invasive species in remote or inaccessible areas [46].

ML models integrate ecological data, climate projections, and land-use information to identify optimal sites for habitat restoration, predict the success of restoration efforts, and prioritize conservation interventions [47].

NLP and computer vision are employed to scan online marketplaces, social media, and shipping manifests to detect illegal wildlife products and identify trafficking networks, supporting law enforcement efforts [48].

AI can analyze multiple ecological indicators (e.g., species diversity, biomass, nutrient cycles) from various data sources to provide a holistic assessment of ecosystem health and resilience, guiding conservation strategies [49].

Resource Management and Sustainable Practices

AI can optimize the use of natural resources and promote more sustainable economic activities.

Optimizing Energy Systems

AI manages demand-response mechanisms, integrates renewable energy sources into the grid, predicts energy consumption, and optimizes energy distribution, leading to increased efficiency and reduced carbon emissions [50].

ML predicts the output of solar and wind farms based on weather forecasts, minimizing intermittency and facilitating grid integration [51].

Precision Agriculture

AI-powered drones and sensors collect data on soil moisture, nutrient levels, crop health, and pest infestations. ML algorithms analyze this data to optimize irrigation, fertilizer application, and pesticide use, reducing waste and environmental impact while increasing yields [52].

Supply Chain Optimization

AI can optimize logistics and supply chains to reduce transportation emissions, minimize waste, and improve resource efficiency from production to consumption [53].

Sustainable Urban Planning

AI analyzes urban data (traffic patterns, energy consumption, waste generation, green spaces) to inform sustainable urban design, optimize infrastructure, and improve resource efficiency in cities [54].

Despite its transformative potential, the application of AI in analyzing climate change and managing environmental risks faces significant challenges that must be addressed for its widespread and responsible adoption.

Data Availability and Quality

Many environmental problems, especially in remote areas or for rare species, suffer from a scarcity of high-quality, labeled datasets necessary for training supervised ML models. Manual labeling is time-consuming and expensive.

Environmental data can contain biases due to uneven sensor distribution, sampling methodologies, or historical reporting practices. AI models trained on biased data can perpetuate or amplify these biases, leading to inaccurate or unfair outcomes [5 5].

Environmental data often has gaps due to sensor failures, cloud cover in satellite imagery, or logistical constraints. Imputing missing data accurately is a challenge, and AI models need to be robust to such imperfections [56].

Integrating diverse data types (e.g., satellite imagery, tabular climate data, free-text reports) from various sources, often in different formats and spatial/temporal resolutions, poses significant interoperability challenges [27].

Model Complexity and Interpretability (Explainable AI - XAI)

Deep learning models, while powerful, often act as "black boxes," making it difficult to understand why they arrive at a particular prediction or decision. In critically important environmental applications (e.g., predicting disaster risk, informing policy), lack of interpretability can hinder trust, accountability, and the ability to diagnose model failures [18].

Stakeholders, including policymakers, affected communities, and regulators, require trust in AI systems. Opaque models can face resistance, making it difficult to implement AI-driven recommendations, especially when significant investments or policy changes are involved [57].

AI models are excellent at identifying correlations, but determining true causal relationships in complex environmental systems remains a challenge. Misinterpreting correlation for causation can lead to ineffective or even detrimental interventions [58].

Computational Resources and Infrastructure

Training and deploying large deep learning models, particularly for tasks involving vast datasets like climate simulations or high-resolution satellite imagery, demand significant computational power and, consequently, substantial energy consumption. This raises concerns about the carbon footprint of AI itself, potentially counteracting its environmental benefits [59].

The cutting-edge AI infrastructure (high-performance computing, specialized GPUs) and the skilled human capital required are often concentrated in developed nations. This creates a digital divide, limiting the capacity of developing countries, which are often most vulnerable to climate change, to leverage AI solutions.

Beyond initial development, the ongoing operational costs of maintaining AI infrastructure, data storage, and model updates can be prohibitive for many environmental organizations and governmental agencies.

Ethical Considerations and Governance

The use of personal data (e.g., in vulnerability assessments or urban planning) raises privacy concerns. Ensuring data security and adherence to ethical guidelines is paramount [60].

When AI systems inform critical decisions (e.g., evacuation orders, resource allocation), questions arise about who is accountable if the AI makes an error or leads to negative outcomes. Clear frameworks for accountability are needed [61].

AI tools could potentially be misused, for instance, in surveillance that infringes on human rights, or in ways that exacerbate existing inequalities (e.g., by disproportionately allocating resources based on biased data).

The hype around AI might lead to "AI washing," where companies or initiatives claim AI benefits without substantive positive environmental impact, diverting attention and resources from genuinely impactful solutions.

Domain Expertise Integration

Effective AI applications in environmental science require a synergistic collaboration between AI researchers (who understand algorithms and data science) and environmental scientists (who possess deep domain knowledge). Bridging this gap is crucial [27].

Many environmental practitioners lack the necessary AI literacy to effectively utilize, critique, and integrate AI tools into their workflows, creating a barrier to adoption.

Translating complex, real-world environmental problems into well-defined AI tasks requires significant collaboration and understanding across disciplines, which can be challenging.

Addressing these limitations requires sustained investment in research, interdisciplinary training, infrastructure development, and the establishment of robust ethical and governance frameworks for AI in environmental applications.

Figure 1: Conceptual Framework Linking Data Inputs, AI Processes, Enablers, and Decision-Making Pathways in Environmental Applications.

Figure 2: AI Pipeline in Environmental Data Analysis, From Data Acquisition to Applications in Climate and Environmental Management.

The field of AI for environmental applications is rapidly evolving, with several promising avenues for future research and development.

Edge Ai and Internet of Things (Iot) For Real-Time Monitoring

The combination of Edge AI (processing data closer to the source) and IoT sensors is poised to revolutionize real-time environmental monitoring. Compact, low-power AI models deployed on sensors and drones will enable immediate analysis of environmental data (e.g., air quality, water levels, wildlife presence) without needing to transmit all raw data to the cloud, leading to faster insights and more agile responses [62]. This is particularly critical for early warning systems and dynamic disaster management.

Federated Learning for Distributed Data

Federated learning allows AI models to be trained on decentralized datasets located at different organizations or devices without sharing the raw data itself. This approach addresses data privacy concerns and enables collaborative model development across multiple environmental agencies or research institutions, particularly beneficial for sensitive data or data that cannot be centralized due to volume or privacy restrictions [63]. For example, multiple cities could collaboratively train an air quality prediction model without sharing their specific sensor data.

Quantum AI for Complex Simulations

While still in nascent stages, Quantum AI holds the potential to tackle environmental challenges that are intractable for classical computers. Quantum algorithms could dramatically accelerate complex climate simulations, optimize resource allocation problems with vast decision spaces, and enable more accurate modeling of intricate biochemical and atmospheric processes [64]. As quantum computing matures, its application in environmental science could unlock entirely new capabilities.

Digital Twins for Environmental Systems

Digital twins, virtual replicas of physical systems, are gaining traction. Imagine a digital twin of a river basin, a city's energy grid, or even an entire forest ecosystem. These AI-powered models would integrate real-time sensor data, historical records, and predictive analytics to simulate current conditions and forecast future scenarios with unprecedented accuracy, allowing for proactive management and policy experimentation in a risk-free virtual environment [65,66].

Ethical AI Frameworks and Policies

As AI becomes more integral to environmental decision-making, the development of robust ethical guidelines, regulatory frameworks, and governance structures will be paramount. This includes ensuring data privacy, designing for fairness and equity, promoting transparency and interpretability, and establishing clear accountability mechanisms for AI-driven interventions [67]. The focus will shift towards not just "can we do it?" but "should we do it responsibly?".

Increased Focus on Causal Inference

Moving beyond correlation, future AI research will place a greater emphasis on causal inference. Developing AI models that can better understand and infer causal relationships in environmental systems will be crucial for designing effective interventions, attributing specific climate impacts, and avoiding unintended consequences [58]. This will strengthen the scientific rigor and policy relevance of AI applications.

Human-AI Collaboration and Citizen Science

Future applications will increasingly involve human-AI partnerships, where AI augments human capabilities rather than replacing them. This includes AI tools that empower citizen scientists to collect and analyze environmental data more effectively, fostering greater public engagement in monitoring and conservation efforts [68]. AI can analyze and validate citizen science data on a large scale.

Integration of AI with Socio-Economic Models

Addressing climate change and environmental risks inherently involves socio-economic dimensions. Future AI applications will increasingly integrate with socio-economic models to better understand the human dimensions of environmental problems, assess the social equity implications of policies, and model the complex interactions between human behavior, economic activity, and environmental changes [69].

These emerging trends underscore a future where AI is not merely a tool but an indispensable partner in navigating the complexities of environmental sustainability, demanding continued innovation, interdisciplinary collaboration, and a commitment to responsible development.

The global environmental crisis, marked by escalating climatic changes and multifaceted environmental risks, presents an unprecedented challenge to humanity. Traditional methodologies, while foundational, often fall short in adequately addressing the scale, complexity, and urgency of these issues. This review paper has extensively demonstrated that Artificial Intelligence (AI) technologies and tools offer a powerful and transformative paradigm for understanding, analyzing, and managing these critical environmental challenges.

We have explored the expansive applications of various AI sub-fields, including Machine Learning and Deep Learning, across two core domains. In climatic change analysis, AI is revolutionizing climate modeling and prediction by enhancing model parameterizations, downscaling projections to actionable regional scales, and improving the forecasting of extreme weather events. Concurrently, its prowess in remote sensing and geospatial data analysis provides unprecedented capabilities in monitoring deforestation, glacier melt, sea-level rise, and atmospheric composition. Furthermore, AI's capacity for data synthesis and knowledge discovery is vital for integrating disparate datasets and uncovering complex environmental patterns.

In the realm of environmental risk management, AI is proving indispensable. It is bolstering disaster preparedness through highly accurate prediction of wildfires, floods, and storm surges, while optimizing emergency response and vulnerability assessments. For pollution control, AI drives advanced monitoring and forecasting of air and water quality, optimizes waste management, and identifies pollution sources. In biodiversity conservation, AI powers automated species identification, habitat mapping, and crucial efforts against illegal wildlife trade. Finally, AI is optimizing resource management in sectors like energy and agriculture, promoting sustainable practices and efficient resource utilization.

Despite these profound advancements, the journey towards fully leveraging AI for environmental sustainability is fraught with challenges. Issues such as the scarcity and quality of labeled environmental data, the "black-box" nature of complex AI models, the significant computational resources required, and pressing ethical considerations demand rigorous attention. Bridging the interdisciplinary gap between AI specialists and environmental scientists remains critical for effective problem formulation and solution implementation.

Looking ahead, emerging trends such as Edge AI, Federated Learning, and the development of Digital Twins promise to further enhance AI's capabilities, enabling real-time, privacy-preserving, and comprehensive environmental intelligence. The ongoing development of robust ethical AI frameworks and a greater focus on causal inference will ensure that these powerful tools are deployed responsibly and effectively.

Ultimately, AI is not a panacea, but an indispensable accelerator in humanity's fight against environmental degradation and climate change. Its true potential will only be realized through sustained interdisciplinary collaboration, substantial investments in research and infrastructure, and a collective commitment to ethical innovation and responsible governance. By strategically harnessing AI, we can move towards a more informed, proactive, and resilient approach to securing a sustainable future for our planet.

1. Jordan MI, Mitchell TM. Machine learning: Trends, perspectives and prospects. Science. 2015; 349: 255-260.
2. MeehlGA, Arblaster JM, Tebaldi C. Uncertainties in projections of future climate change in the IPCC AR4 models. Bull Am Meteorol Soc. 2007; 88: 1581-1596.
3. Sender LS, Stehlik M.Climate change and the increasing frequency of extreme weather events. Risk, Hazards & Crisis in Public Policy. 2020; 11: 167-194.
4. Landrigan PJ, Fuller R, Amann M, Bellinger B, Brauer S, Zheng T. The Lancet Commission on pollution and health. The Lancet. 2018; 391: 462-512.
5. Ceballos G, Ehrlich PR, Barnosky AD, García A, Pringle RM, Palmer TM. Accelerated modern human-induced species losses: Entering the sixth mass extinction. Science Advances. 2015; 1: e1400253.
6. Levin SA. Ecosystems and the biosphere as complex adaptive systems. Ecosystems. 1998; 1: 431-436.
7. Russell SJ, Norvig P. Artificial intelligence: A modern approach. Prentice Hall. 2010.
8. Bishop CM. Pattern recognition and machine learning. Springer. 2006.
9. Hastie T, Tibshirani R, Friedman J. The elements of statistical learning: Data mining, inference and prediction. Springer. 2009.
10. Sutton RS, Barto AG. Reinforcement learning: An introduction. MIT Press. 2018.
11. LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015; 521: 436-444.
12. Krizhevsky A, Sutskever I, Hinton GE. ImageNet classification with deep convolutional neural networks. Adv Neural Inf Process Syst. 2012; 25.
13. Hochreiter S, Schmidhuber J. Long short-term memory. Neural Computation. 1997; 9: 1735-1780.
14. Goodfellow I, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, et al. Generative adversarial nets. Adv Neural Inf Process Syst. 2014; 27.
15. Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN. Attention is all you need. Adv Neural Inf Process Syst. 2017; 30.
16. Szeliski R. Computer vision: Algorithms and applications. Springer. 2010.
17. Jurafsky D, Martin JH. Speech and language processing: An introduction to natural language processing, computational linguistics and speech recognition. Pearson Prentice Hall. 2009.
18. Adadi A, Berrada M. Peeking inside the black-box: A survey on Explainable Artificial Intelligence (XAI). IEEE Access. 2018; 6: 52138-52160.
19. Chen M, Mao S, Liu Y. Big data: A survey. Mob Netw Appl. 2014; 19: 171-209.
20. Hourdin F, Mauritsen T, Dufresne JL, Ghan S, Phillips V, Roeckner E, et al. The art and science of climate model tuning. Bull Am Meteorol Soc. 2017; 98: 589-606.
21. Rasp S, Pritchard MS, Gentine P. Deep learning improves cloud parameterizations in global climate models. Proc Natl Acad Sci. 2018; 115: 9601-9606.
22. Liu J, Zhang C, Zhang Y. Deep learning for remote sensing image classification: A review. IEEE Geoscience and Remote Sensing Magazine. 2021; 9: 29-41.
23. Wang L, Zhang J, Sun Y. Deep learning for climate downscaling with generative adversarial networks. Geophysical Research Letters. 2020; 47: e2020GL089600.
24. Ham YG, Kim JH, Luo JJ. Deep learning for seasonal forecasting of El Niño over the equatorial Pacific. Nature Geoscience. 2019; 12: 1003-1008.
25. Chattopadhyay A, Paul S, Ghosh S. Deep learning for tropical cyclone intensity prediction. Geophysical Research Letters. 2020; 47: e2019GL086221.
26. Si B, Li X, Wang Y, Chen G. Machine learning-based attribution of extreme weather events. J Geophys Res: Atmospheres. 2019; 124: 9403-9419.
27. Reichstein M, Camps-Valls G, Stevens B, Jung J, Denzler J, Carvalhais N. Deep learning and process understanding for data-driven Earth system science. Nature. 2019; 566: 195-204.
28. Mohajerani S, Baral S, Ma H. Glacier change detection using deep learning from satellite imagery. Remote Sensing. 2018; 10: 1999.
29. Malhotra A, Singla A, Gupta SK. Machine learning techniques for sea level rise prediction. J Ocean Eng Sci. 2019; 4: 159-168.
30. Zeng Z, Li L, Hu M. Satellite-based estimation of global urban CO2 emissions using machine learning with high-resolution imagery. Environ Res Letters. 2020; 15:
31. Son NH, Chen CF, Chen RC, Chang LY, Fan CM. Machine learning for remote sensing of oceanographic parameters: A review. Remote Sensing. 2019; 11:2398.
32. Norouzzadeh MS, Nguyen A, Kosmala M, Swanson A, Palmer MS, Fei-Fei L. Automatically identifying animal species in camera trap images with deep learning. Proc Natl Acad Sci. 2018; 115: E6308-E6317.
33. Camps-Valls G, Tuia D, Rivero-Rodríguez AM, Muñoz-Marí J, Reichstein M. Deep learning for Earth observation: Status and prospects. J Photogramm Remote Sens. 2019; 142: 19-38.
34. Kortum P, Milbradt A. The role of AI in climate change research: A systematic review. Environ Data Sci. 2021; 2: e12.
35. Jain P, Coogan SC, Parisien MA. Machine learning for wildfire prediction: A review. Environmental Reviews. 2020; 28: 314-325.
36. Zhang T, Lin H, Liu S. Landslide susceptibility mapping using convolutional neural networks. Geomor phology. 2018; 314: 4-15.
37. Bakhtyar R, Barry R, Moncrieff T. Machine learning for coastal flood prediction. J Coastal Res 2019; 35: 332-341.
38. Ma S, Li Y, Li Z. A review of artificial intelligence applications in disaster management. Natural Hazards. 2020; 103: 1629-1662.
39. Fang Y, Zhao L, Hu Z. A review of remote sensing and artificial intelligence for urban flood modeling. Environmental Modelling& Software. 2021; 140: 105021.
40. Alam K, Khan N, Khan Z, Khan S. IoT-based real-time flood monitoring and early warning system. Int J Advanced Computer Sci Applications. 2019; 10.
41. Zheng Y, Liu F, Hsieh HP. U-Air: When urban planning meets mobile computing and big data. Proceedings of the 21st ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2015; 2101-2110.
42. Ahmed U, Al-Fuqaha A, Shah B. A survey on artificial intelligence in water quality prediction and management. IEEE Access. 2019; 7: 174092-174116.
43. Yang Y, Yang J, Ma J. Deep learning for intelligent waste sorting: A review. Waste Management. 2020; 106: 1-13.
44. Snoek J, Larochelle H, Adams RP. Practical Bayesian optimization of machine learning algorithms. Adv Neural Inf Process Syst. 2012; 25.
45. Sugai LSM, Silva TSF, Souza LCS, Llusia D. A survey on bioacoustics and deep learning for wildlife monitoring. Ecology and Evolution. 2019; 9: 12386-12399.
46. Christie FJ, Aplin LM, Jones CS. Autonomous drones for wildlife surveillance: A review. Biological Conservation. 2021; 254: 108985.
47. Talluto MV, Potvin C, Rémy-Courchesne F, Wouters R. Integrating machine learning into ecological restoration decision-making. Restoration Ecology. 2021; 29: e13426.
48. Hinsley A, Verissimo D, Roberts D. Understanding the motivations of online wildlife product consumers. Conservation Biology. 2021; 35: 656-666.
49. Pettersson LB, Esmaeili Y, Lindberg G. Artificial intelligence for biodiversity monitoring and conservation: Opportunities and challenges. Frontiers in Ecology and Evolution. 2022; 10: 786273.
50. Mohsenian-Rad AH, Wong VW, Jatskevich J, Schober R, Leon-Garcia A. Autonomous demand-side management for smart grid. IEEE Transactions on Smart Grid. 2010; 1: 320-331.
51. Dash SK, Singh S, Mishra MK. Deep learning for wind speed forecasting: A review. Neural Comput Appl. 2020; 32: 7545-7561.
52. Liakos KG, Busato P, Moshou D, Pearson S, Bochtis D. Machine learning in agriculture: A review. Sensors. 2018; 18: 2674.
53. Tiwari V, Jain V, Singh AK. Artificial intelligence in supply chain management: A review. Int J Production Res. 2019; 57: 4642-4663.
54. Bibri SE. The IoT-based paradigm of smart sustainable cities: A comprehensive and systematic review. Sustainable Cities and Society. 2021; 64: 102281.
55. Wang G, Sun S, Li T. Addressing data bias in machine learning for environmental sustainability research. Environmental Science & Technology Letters. 2021; 8: 978-984.
56. Gupta M, Singh SP, Singh RK. A comprehensive review on missing data imputation using machine learning. Machine Learning with Applications. 2020; 2: 100003.
57. Fazelpour S, Danks D. Explaining AI and its impact on trust and reliance. AI & Society. 2022; 37: 185-197.
58. Pearl J. Causality. Cambridge University Press. 2009.
59. Strubell E, Ganesh A, McCallum A. Energy and policy considerations for deep learning in NLP. Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. 2019; 3645-3650.
60. Floridi L, Cowls J, Beltrametti M, Chatila R, Chazerand P, Dignum V, et al. AI4People-Ethical guidelines for a trustworthy AI. Minds and Machines. 2018; 28: 689-707.
61. Crawford K. Atlas of AI: Power, politics and the planetary costs of artificial intelligence. Yale University Press. 2021.
62. Mohanty SP, Kougianos E, Puthal D. Deep learning at the edge: A comprehensive survey. IEEE Trans Emerg Top Comput. 2020; 8: 579-598.
63. Kone?ny J, McMahan HB, Yu FX, Richtárik P, Suresh AT, Bacon D. Federated learning: Strategies for improving communication efficiency. arXiv preprint arXiv. 2016; 1610: 05492.
64. Preskill J. Quantum computing in the NISQ era and beyond. Quantum. 2018; 2: 79.
65. Glaessgen E, Stargel D. The Digital Twin paradigm for future NASA missions. AIAA Aerospace Conference. 2012.
66. Fuller C, Hua W, Wei R. The digital twin: A foundational technology for Industry 4.0. Journal of Manufacturing Systems. 2020; 58: 336-347.
67. Jobin A, Ienca M, Vayena E. The global landscape of AI ethics guidelines. Nature Machine Intelligence. 2019; 1: 389-399.
68. Fritz S, See L, Carlson T, Haklay M, van der Velde M. Citizen science and the United Nations Sustainable Development Goals. Nature Sustainability. 2019; 2: 922-930.
69. Patt A, van Vuuren DP. Integrated assessment models for climate change policy: Lessons and challenges. Annu Rev Environ Resour. 2019; 44: 327-353.
70. Liu Y, Li Q, Wang J. Deep learning for climate downscaling: A review. Atmospheric Research. 2020; 240:104921.