AI-Powered FRA Atlas: An AI- and WebGIS-Based Decision Support Framework for Forest Rights Act Implementation Monitoring

Abstract

The Forest Rights Act (FRA), 2006 recognizes the land and resource rights of forest-dwelling communities in India, but its implementation is hindered by paper-based records, fragmented databases, delayed claim verification and weak monitoring. This paper presents an AI-powered FRA Atlas integrated with a WebGIS-based Decision Support System (DSS) that combines digitized FRA records, Sentinel-2 satellite imagery, machine-learning land-cover classification, a retrieval-augmented policy assistant, a statutory scheme-eligibility rule engine and a tamper-evident audit trail in one platform. A Random Forest classifier operating on five Sentinel-2 bands and three spectral indices assigns pixels to seven land-cover classes. On a 4,200-sample holdout set it reaches 87.24% overall accuracy, Cohen’s ? of 0.8511 and a macro F1-score of 0.8726, and 5-fold stratified cross-validation gives 87.35% ± 0.24%. The hybrid retrieval module achieves 85.71% top-1 accuracy and a mean reciprocal rank of 0.9286 on seven policy queries. The classifier is trained on calibrated synthetic spectral samples rather than field-labelled FRA land, so the reported figures measure separability under those distributions and are not field accuracy; validation on real labelled imagery remains the main open item.

Introduction

1. Introduction

The Forest Rights Act (FRA), 2006 was introduced to recognize and protect the rights of Scheduled Tribes and other traditional forest-dwelling communities over forest land and resources.

The implementation process involves submitting claims through the Gram Sabha, verifying the claims, approving them, and issuing land-rights titles under three categories:

• Individual Forest Rights (IFR): Rights of individuals over eligible forest land.

• Community Rights (CR): Rights of communities over forest resources.

• Community Forest Resource (CFR): Rights of communities to protect, conserve, and manage traditionally used forest resources.

However, traditional FRA implementation depends heavily on paper records and disconnected databases, resulting in delays, inconsistent information, and limited transparency.

To address these challenges, the proposed project develops an AI-powered FRA Atlas and WebGIS-based Decision Support System (DSS) that integrates digital records, satellite imagery, artificial intelligence, and interactive dashboards.

2. Problem Statement and Motivation

The existing FRA implementation process faces several challenges:

• Manual documentation and lengthy claim-processing procedures.

• Inconsistent and incomplete land-rights records.

• Difficulty in verifying actual land boundaries.

• Lack of real-time claim-status monitoring.

• Limited transparency and coordination between departments.

• Difficulty accessing and searching scanned documents.

• High claim-rejection rates and limited awareness among eligible communities.

The proposed system uses AI, GIS, remote sensing, and web technologies to improve claim monitoring, land verification, data management, and decision-making.

3. Objectives of the Project

The main objectives of the proposed system are:

1. To design and develop an AI-powered FRA Atlas for monitoring forest land claims.

2. To integrate WebGIS for visualizing forest-rights information geographically.

3. To digitize scanned FRA documents and extract important information using OCR.

4. To automate claim-status tracking and land-distribution monitoring.

5. To improve land verification using satellite imagery and AI-based classification.

6. To provide interactive dashboards, government-scheme eligibility guidance, and an AI question-answering assistant.

7. To improve transparency, accountability, and data management while protecting community rights.

4. Literature Review

Previous research related to the proposed system can be divided into five main areas:

• FRA Implementation Studies: Research has identified delays, unclear procedures, inconsistent records, and governance-related problems in implementing the Forest Rights Act.

• Existing Digital Initiatives: Projects such as the Maharashtra FRA Portal and Tripura Digital FRA Atlas demonstrate the potential of digital technologies in FRA management.

• Remote Sensing and Land-Cover Mapping: Satellite imagery, particularly Sentinel-2, helps identify vegetation, water bodies, built-up areas, and changes in land use.

• Machine Learning for Classification: Algorithms such as Random Forest and deep-learning models like U-Net are used to classify satellite images and identify different land-cover types.

• Document Digitization and AI: OCR converts scanned documents into searchable text, while Retrieval-Augmented Generation (RAG) helps answer questions using relevant legal documents and stored information.

Existing studies demonstrate the usefulness of these individual technologies, but their integration into a complete FRA monitoring and decision-support system remains an important research requirement.

5. Research Gap

The existing FRA management systems have several limitations:

1. Lack of centralized digital records connecting land claims and geographical boundaries.

2. Limited use of AI-based satellite imagery analysis for land verification.

3. Difficulty digitizing and searching historical claim documents.

4. Lack of automated claim tracking and inconsistency detection.

5. Limited integration of scheme-eligibility guidance and AI-based question answering.

6. Insufficient transparency and monitoring across different administrative levels.

The proposed system addresses these gaps by integrating document digitization, geospatial analysis, machine learning, AI assistance, and decision-support dashboards into one platform.

6. Proposed System Framework

The proposed FRA Atlas consists of nine major components:

1. Data Ingestion and Digitization: Collects and converts FRA records and scanned documents into digital information.

2. Geospatial Database: Stores claim details and connects them with geographical land boundaries.

3. Satellite Image Processing: Acquires and processes Sentinel-2 satellite imagery.

4. AI-Based Land Classification: Uses machine-learning algorithms to identify different land-cover categories.

5. Inconsistency Detection: Identifies differences between recorded claims and actual geographical information.

6. Claim Tracking and Analytics: Monitors claim progress, approval status, and land distribution.

7. AI Assistant and Eligibility Engine: Provides answers to FRA-related questions and guidance on government-scheme eligibility.

8. Tamper-Evident Audit Trail: Maintains traceable records of system activities and changes.

9. WebGIS and Decision Dashboards: Provides interactive maps, reports, visualizations, and decision-support information.

7. Technology Stack

The proposed system uses modern web-development and geospatial technologies.

A. Frontend Technologies

• Next.js 14: Used to develop the web application.

• React 18: Creates interactive user-interface components.

• TypeScript: Provides structured and type-checked programming.

• Tailwind CSS: Used for designing and styling the application.

• Leaflet.js: Displays interactive geographical maps.

• Recharts: Creates analytical charts and dashboards.

• Lucide: Provides interface icons.

B. Backend and AI Technologies

The proposed architecture also incorporates:

• Python and FastAPI: For backend services and API development.

• OCR (Tesseract): For extracting text from scanned FRA documents.

• Retrieval-Augmented Generation (RAG): For answering questions using relevant legal and government documents.

• Random Forest: For satellite-based land-cover classification.

• Sentinel-2 imagery: For geographical analysis and land monitoring.

• NDVI, NDWI, and NDBI: For identifying vegetation, water bodies, and built-up areas.

8. Working Principle

The system operates through the following process:

1. FRA claim documents and land records are collected and digitized.

2. OCR technology extracts relevant information from scanned documents.

3. The extracted information is stored in a structured geospatial database.

4. Sentinel-2 satellite images are acquired and processed.

5. Machine-learning algorithms classify land-cover types.

6. The system compares recorded claim information with geographical data to identify possible inconsistencies.

7. Claim progress and land distribution are monitored through analytical tools.

8. The AI assistant provides information using relevant FRA documents and legal knowledge.

9. Authorities access the results through interactive WebGIS maps and decision-support dashboards.

The system is designed to assist statutory authorities, including Gram Sabhas and Sub-Divisional and District Level Committees. It does not replace their legal authority to approve or reject FRA claims.

9. Research Contribution and Significance

The main contribution of the project is the integration of established technologies into a single platform for forest-rights monitoring and management.

Its potential benefits include:

• Improved digitization and management of FRA records.

• Faster access to claim-related information.

• Better geographical visualization of forest land.

• Improved land-cover monitoring through satellite imagery.

• Identification of discrepancies in land records.

• More transparent claim-status monitoring.

• Improved access to FRA-related information and government schemes.

• Better administrative decision-making and accountability.

The system also recognizes the importance of community consent, privacy, and responsible data governance when handling sensitive forest-rights information.

Conclusion

This paper presented an AI-powered FRA Atlas with a WebGIS-based Decision Support System, covering its design, implementation and first quantitative evaluation. The platform combines a Next.js frontend, a FastAPI backend, geodesic GIS validation, OCR and LLM-based document extraction, Sentinel-2 land analysis, RAG-based policy retrieval, a statutory scheme-eligibility rule engine and a SHA-256 audit chain. The Random Forest land-cover classifier reached 87.24% holdout accuracy (? = 0.8511, macro F1 = 0.8726) and 87.35% ± 0.24% in 5-fold cross-validation, with errors concentrated between crop and grassland and between road and bare land, and the retriever reached 85.71% top-1 accuracy (MRR 0.9286). Because these results rest on calibrated synthetic training data and a small retrieval test set, the system should be regarded as a research and engineering prototype; validation on field-labelled imagery, measured OCR and usability results, and community-centred governance are required before real-world deployment.

References

[1] M. Manimegala, T. Gayathri, M. M. Deva Harsar, and Y. Hans Rohit, “AI-Powered FRA Atlas and WebGIS-Based Decision Support System for Forest Rights Act Implementation Monitoring,” 2026. [2] K. Nikawat and P. Tak, “A Critical Study of Forest Rights Act, 2006,” 2024. [3] S. B. S., K. G. Dileep Kumar, and B. Ramesh, “An Analytical Study of Forest Rights Act Implementation in Karnataka,” 2020. [4] J. Rogan and D. Chen, “Remote sensing technology for mapping and monitoring land-cover and land-use change,” Progress in Planning, vol. 61, no. 4, pp. 301–325, 2004. [5] L. Zhang, L. Zhang, and B. Du, “Deep learning for remote sensing data: A technical tutorial on the state of the art,” IEEE Geosci. Remote Sens. Mag., vol. 4, no. 2, pp. 22–40, 2016. [6] L. Breiman, “Random forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, 2001. [7] N. Gorelick, M. Hancher, M. Dixon, S. Ilyushchenko, D. Thau, and R. Moore, “Google Earth Engine: Planetary-scale geospatial analysis for everyone,” Remote Sens. Environ., vol. 202, pp. 18–27, 2017. [8] J. P. Shim, M. Warkentin, J. F. Courtney, D. J. Power, R. Sharda, and C. Carlsson, “Past, present, and future of decision support technology,” Decision Support Systems, vol. 33, no. 2, pp. 111–126, 2002. [9] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of deep bidirectional transformers for language understanding,” in Proc. NAACL-HLT, 2019, pp. 4171–4186. [10] S. Singh, “The implications of the Scheduled Tribes and Other Traditional Forest Dwellers (Recognition of Forest Rights) Act, 2006,” Indian Forester, vol. 140, no. 11, pp. 1081–1084, 2014, doi: 10.36808/if/2014/v140i11/54443. [11] S. Narayan, “Legal framework for forest rights in India: A critical analysis,” Univ. of North Bengal, 2024. [Online]. Available: https://ir.nbu.ac.in/handle/123456789/5580 [12] O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional networks for biomedical image segmentation,” in Proc. MICCAI, LNCS vol. 9351, 2015, pp. 234–241. [13] M. Drusch et al., “Sentinel-2: ESA’s optical high-resolution mission for GMES operational services,” Remote Sens. Environ., vol. 120, pp. 25–36, 2012. [14] C. F. Brown et al., “Dynamic World, near real-time global 10 m land use land cover mapping,” Scientific Data, vol. 9, art. 251, 2022. [15] Z.-R. Peng and M.-H. Tsou, Internet GIS: Distributed Geographic Information Services for the Internet and Wireless Networks. Hoboken, NJ, USA: Wiley, 2003. [16] Ministry of Tribal Affairs, Govt. of India, “Forest Rights Act (FRA) digital platform,” Lok Sabha unstarred question AU2117. [Online]. Available: https://eparlib.sansad.in/bitstream/123456789/3022375/1/AU 2117_r9gKnx.pdf [17] “Tripura Governor launches Northeast’s first digital FRA atlas for tribal rights,” NewKerala.com, Jul. 1, 2026. [Online]. Available: https://www. newkerala.com/news/a/committed-tribal-dignitytripura-governor-launch es-northeasts-first-426.htm [18] “Two faces of the same algorithm: Algorithmic disparate impact on India’s Adivasi communities,” LiveLaw. [Online]. Available: https://w ww.livelaw.in/lawschool/articles/algorithmic-disparateimpact-india-adi vasi-communities-544518 [19] J. W. Rouse, R. H. Haas, J. A. Schell, and D. W. Deering, “Monitoring vegetation systems in the Great Plains with ERTS,” in Proc. 3rd ERTS Symp., NASA SP-351, 1974, pp. 309–317. [20] S. K. McFeeters, “The use of the normalized difference water index (NDWI) in the delineation of open water features,” Int. J. Remote Sens., vol. 17, no. 7, pp. 1425–1432, 1996. [21] Y. Zha, J. Gao, and S. Ni, “Use of normalized difference built-up index in automatically mapping urban areas from TM imagery,” Int. J. Remote Sens., vol. 24, no. 3, pp. 583–594, 2003. [22] R. Smith, “An overview of the Tesseract OCR engine,” in Proc. 9th Int. Conf. Document Analysis and Recognition (ICDAR), 2007, pp. 629–633. [23] P. Lewis et al., “Retrieval-augmented generation for knowledge-intensive NLP tasks,” in Advances in Neural Information Processing Systems, vol. 33, 2020, pp. 9459–9474. [24] Groq Inc., “GroqCloud API documentation.” [Online]. Available: https://console.groq.com/docs [25] S. Ramírez, “FastAPI documentation.” [Online]. Available: https://fastapi.tiangolo.com [26] V. Agafonkin, “Leaflet: An open-source JavaScript library for interactive maps.” [Online]. Available: https://leafletjs.com [27] PostGIS Project, “PostGIS documentation.” [Online]. Available: https://postgis.net [28] M. Jones, J. Bradley, and N. Sakimura, “JSON Web Token (JWT),” IETF RFC 7519, May 2015. [29] H. Butler, M. Daly, A. Doyle, S. Gillies, S. Hagen, and T. Schaub, “The GeoJSON format,” IETF RFC 7946, Aug. 2016. [30] F. Pedregosa et al., “Scikit-learn: Machine learning in Python,” J. Mach. Learn. Res., vol. 12, pp. 2825–2830, 2011. [31] P. Helber, B. Bischke, A. Dengel, and D. Borth, “EuroSAT: A novel dataset and deep learning benchmark for land use and land cover classification,” IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 12, no. 7, pp. 2217–2226, 2019. [32] D. Zanaga et al., “ESA WorldCover 10 m 2020 v100,” Zenodo, 2021, doi: 10.5281/zenodo.5571936. [33] J. Cohen, “A coefficient of agreement for nominal scales,” Educ. Psychol. Meas., vol. 20, no. 1, pp. 37–46, 1960. [34] NIST, “Secure Hash Standard (SHS),” FIPS PUB 180-4, Aug. 2015. [35] N. Reimers and I. Gurevych, “Sentence-BERT: Sentence embeddings using Siamese BERT-networks,” in Proc. EMNLP-IJCNLP, 2019, pp. 3982–3992.

Copyright

Copyright © 2026 Mrs. Chandana K S, Manish RZ, Prajwal H P, Sagar G Kunte, P B Darshan Prasad. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.