Artificial intelligence in climate science and Long-Term Environmental Health and Green Future -Prediction Monitoring and Green technology

229 International Journal of Advance and Applied Research www.ijaar.co.in ISSN – 2347-7075 Impact Factor – 8.141 Peer Reviewed Bi-Monthly Vol. 6 No. 38 September - October - 2025 Artificial intelligence in climate science and Long-Term Environmental Health and Green Future -Prediction Monitoring and Green technology Sonali Purushottam Kumawat Assistant Professor, Department of Computer Science Dr. D. Y. Patil Arts, Commerce & Science College, Akurdi, Pune Corresponding Author – Sonali Purushottam Kumawat DOI - 10.5281/zenodo.17315552 Abstract: This paper discusses and examines ways to enhance environmental sustainability through the use of Artificial Intelligence (AI). The integration of Artificial Intelligence contributes to environmental sustainability by enabling the detection and monitoring of green land, supporting efforts in climate stabilisation, sustainable agriculture and ecosystem conservation. The research focuses on identifying regions with existing vegetation and detecting vacant or unused land appropriate for plantation activities. It enables users to easily identify regions rich in plant life as well as those lacking vegetation, aiding in strategic environmental planning. The app was developed to display the satellite images and the corresponding output. By analysing satellite imagery, the app provides a visual overview of plant distribution and vacant land. The proposed application uses AI and remote sensing data to provide accurate vegetation mapping and identify empty spaces. By identifying both planted and barren zones, the app contributes to better land management and ecosystem planning. Introduction: With fast urban growth and climate change, making good land use plans is more important now than ever. Traditional ways of checking land use often depend on manual surveys and old data, which can be slow and not very accurate. Traditional methods of checking plant cover often take a long time because they rely on manual surveys. AI is the main technology that allows the app to spot and tell the difference between areas with plants and those that are empty or not used. By using AI, the app can quickly look through big and complex data sets, like satellite images, drone photos, or photos taken from the ground, to give useful information for managing how land is used. The ability to find empty spaces that are good for planting can help with reforestation, making cities greener, and farming in a way that is better for the environment. This app uses satellite images along with machine learning to show a clear picture of land cover, helping to distinguish between areas with plants and areas without. The app is useful for different people, like farmers trying to grow more crops, city planners looking to add more green areas, and environmentalists working to restore habitats. This app helps by connecting available data with useful actions, giving users a tool to detect and watch for plants in different kinds of land. It offers a solution that is scalable and automatic, making it better and faster. IJAAR Vol. 6 No. 38 ISSN – 2347-7075 Sonali Purushottam Kumawat 230 Research looks at how AI can help make environmental sustainability better in many areas, such as protecting biodiversity, maintaining ecosystems, etc. A big part of this is using AI to monitor the environment, which helps make better decisions and take effective actions. Because of this, adding AI technologies can greatly improve how we work on environmental sustainability. As AI becomes more powerful and widely used, its role in helping industries focused on sustainability is growing. Especially in areas with a big impact on the environment, AI can bring big positive changes. Right now, we are at a key point in time: “advances in big data”, “computer hardware”, and “AI algorithms” have made it possible to solve environmental problems that were once thought impossible. With AI, taking care of our planet – including the oceans – “feels more achievable than ever.” In this study, we look at five different ways AI can help with environmental sustainability, giving a full overview of its potential effect in major areas. Remote Sensing Foundations for Vegetation Mapping: Remote sensing technologies, including satellite images and drone data, are central to modern land assessment. Optical remote sensing uses visible and near-infrared light to capture information that helps measure vegetation health and structure through tools like NDVI and EVI. Radar-based methods, such as Synthetic Aperture Radar (SAR), add to this by offering images in all weather conditions and detailed insights into forest structures. Traditional Classification Methods: Earlier, mapping efforts used rulebased or threshold-based methods, like using decision trees based on NDVI or object height. These were fast and easy to understand for the broad categorisation of vegetation, though they were not very accurate. Better methods, like supervised machine learning classifiers such as Support Vector Machines (SVM), Random Forest (RF), decision trees, and k-Nearest Neighbours, have improved accuracy and adaptability, especially when there is a lot of data. Techniques like ensemble learning, especially Random Forest, have increased performance and reliability across different environments. Deep Learning and Advanced AI Techniques: The use of deep learning has greatly improved remote sensing analysis. Models like Convolutional Neural Networks (CNNs), 1D/2D/3D variants, autoencoders, and GANs are now common in processing hyperspectral images and classifying land cover. Models such as DeepLabV3+ use advanced convolution techniques in an encoder-decoder setup to precisely identify different vegetation areas. Meta-Analytical Reviews and Methodological Insights: Studies have reviewed the latest developments in remote sensing and AI detection. These highlight key challenges, such as detecting objects at different scales, in rotated positions, or that are small and faint. They also suggest ways to improve detection accuracy and model reliability in the future. Another review focuses on using deep learning for change detection, classifying methods into supervised, unsupervised, or transfer learning approaches. Sensor Fusion and Explainable AI (XAI) in Mapping: Using both optical and radar data improves classification and allows for continuous monitoring, even when there are clouds. IJAAR Vol. 6 No. 38 ISSN – 2347-7075 Sonali Purushottam Kumawat 231 Explainable AI techniques, like SHAP values, are being used more to make AI decisions easier to understand – this is especially important for important tasks like identifying different tree species or tracking vegetation. Real-World Applications and New Tools: AI has real-world uses in many areas:  Agricultural mapping: AI helps in precision farming by showing land cover, which helps assess crop health and manage natural resources better.  Soil and water management: Machine learning models like Random Forest, SVM, and CNNs are used for monitoring large areas for soil salinity and water resources with geospatial data. arccjournals.com.  Climate-risk and financial modelling: AI-based remote sensing links environmental changes, such as patterns of urban heat or areas prone to flooding, with social and economic risks in housing and city planning. MDPI.  Regenerative agriculture and soil carbon mapping: Organisations are using ML to create detailed maps of soil carbon and reduce the need for traditional sampling, making these processes more scalable and accurate, Reuters. Methodology: Machine Learning: Satellite imagery 1) Image input taken from a satellite. 2) The application developed uses technology. A) Backend: Python, Django Rest Framework, NumPy, Pandas, OpenCV, SciKit Learn, DeepLabV3+ with ResNet-101 Backbone. B) Frontend: ReactJS, Material UI, JavaScript , Axios ( API Calls ) 3) Input is applied to an AI application. 4) Implementing the CNN, which uses the DeepLabV3+ model. 5) Implementing the LULC method to categorise and map different types of land surface features. 6) Display the output from the proposed implementation. Study Area: Location: Flame University Area. Model: DeepLabV3+ with ResNet-101 Backbone Library: segmentation_models_pytorch(SMP) Dataset: ADE20K ( The MIT Scene Parsing Benchmark) IJAAR Vol. 6 No. 38 ISSN – 2347-7075 Sonali Purushottam Kumawat 232 Results: Result Analysis: Input Image Output Image Parameter Value Value Width x Height 1349 x 708 1349 x 708 Total Pixels 9,55,092 9,55,092 Format JPEG / JPG JPEG / JPG Pixel Density Variable Variable Palette Natural Colour Spectrum ADE20K Colour Scheme PSNR SVM Prediction label SVM (Class) Probability SVM Decision Value 11.315 dB 1 (0.44 0.64) 0.99999968 IJAAR Vol. 6 No. 38 ISSN – 2347-7075 Sonali Purushottam Kumawat 233 Conclusion: This work presents an AI-based approach for identifying and monitoring planting and non-planting zones by utilising spatial data extracted through dimension prediction techniques from Google Maps. The methodology combines satellite imagery analysis with machine learning to automatically detect land usability based on surface patterns, vegetation coverage, and spatial attributes. By applying AI models to geospatial inputs, the system efficiently classifies large areas without relying on manual fieldwork. The integration of Google Maps enhances the framework by providing consistent and widely accessible mapping data, making the approach scalable and adaptable across different regions. The proposed system demonstrates strong potential for applications in precision agriculture, land resource planning, and environmental monitoring. Future enhancements may involve integrating deep learning models, incorporating time-series data, or refining the system for region-specific agricultural practices. References: 1. Manish Yadav1, Gurjeet Singh2: 1Student of MCA 3rd Sem., Lords University, Chikani Alwar, Rajasthan-301028 2Dean, Lords School of Computer Applications & IT, Lords University, Chikani, Alwar, Rajasthan-301028 2. EPRS |European Parliamentary Research Service Author philip boucher Scientific foresight unit (STOA)PE 641547 June 2020 3. Sarker, I. H. (2021). Machine learning: Algorithms, real-world applications and research directions. SN computer science, 2(3), 160 4. Abdullah, A. Y. M., Masrur, A., Adnan, M. S. G., Baky, M., Al, A., Hassan, Q. K., et al. (2019). Spatio-temporal patterns of land use/land cover change in the heterogeneous coastal region of Bangladesh between 1990 and 2017. Remote Sensing 11, 790. doi: 10.3390/rs11070790 5. David Popp, The Role of Green Technology Transfer in Climate Policy, RESOURCESMAG.ORG (2010), https://www.resourcesmag.org/commonreso urces/the-role-of-green-technologytransfer-i n-climate-policy/ (last visited Jun 28, 2010). 6. Robert Richardson, DEPLETING EARTH‟S RESOURCES, MSUTODAY.MSU.EDU (2018), 7. https://msutoday.msu.edu/news/2018/depl eti ng-earths-resources/ (last visited Sep 1, 2018). 8. Robert Richardson, Yes, humans are depleting Earth‟s resources, but IJAAR Vol. 6 No. 38 ISSN – 2347-7075 Sonali Purushottam Kumawat 234 „footprint‟ estimates don‟t tell the full story, THECONVERSATION.COM (2018), https://theconversation.com/yeshumans-are-depleting-earths-resourcesbut-footprint-estimates-dont-tell-the-fullstory-100705 (last visited Jul 31, 2018). 9. A Adnan A. Y. Mustafa, Linda G. Shapiro and Mark A. Ganter,"3D Object Recognition from Color Intensity Images", 13th Int. Conf. on Pattern Recognition, Vienna, Austria, pp. 25-30, /August, 1996. 10. Wang Xiang-yang, Sun Wei-wei a, Wu Zhi-fang, Yang Hong-ying, Wang Qinyan “Color image segmentation using PDTDFB domain hidden Markov tree model” Applied Soft Computing 29 (2015) 138– 152. 11. Yang Haibo, Wang Zongmin, Zhao Hongling, Guo Yu “Water body Extraction Methods Study Based on RS and GIS ” 2011 3rd International Conference on Environmental Science and Information Application Technology (ESIAT 2011). 12. Yuqiang Wang, Renzong Ruan, Yuanjian SHE, Meichun YAN “Extraction of Water Information based on Radarsat Sar and Landsat ETM+ ” 2011 3rd International Conference on Environmental Science and Information Application Technology (ESIAT 2011). 13. U. S. Geological Survey (USGS)” Landsat Orthorectified ETM+ Pan Sharpened” Sioux Falls, SD USA, USGS and Science Center (EROS), https://lta.cr.usgs.gov/Tri_Dec_GLOO. 14. Xiaoxiao Lia, B., Soe W. Myintb, Yujia Zhangb, Christopher Gallettib, Xiaoxiang Zhangc, Billie L. Turner II “Object-based land-cover classification for metropolitan Phoenix, Arizona, using aerial photography” International Journal of Applied Earth Observation and Geoinformation 33 (2014) 321–330. 15. Usama Fayyad, Gregory PiatetskyShapiro, and Padhraic Smyth “From Data Mining to Knowledge Discovery in Databases “American Association for Artificial Intelligence. All rights reserved. 0738-46021996. 16. I. Witten, E. Frank “Data Mining: practical Machine Learning Tools and Techniques”. 17. J. Zahang, W. Hsu and M. L. Lee. “An information-Driven Framework for image Mining”, in proceedings of 12th International Conference on Database and Expert Systems Applications.