Contribution

This study develops and tests a reproducible blockchain-based framework for transparent verification, analysis, and governance of climate data. The architecture integrates three layers: a data ingestion layer that standardizes and validates observations, a blockchain layer that ensures immutability and provenance through proof-of-stake consensus, and a statistical analysis layer that applies deterministic methods for anomaly detection and trend evaluation (Fig 1). Unlike prior blockchain applications in climate domains, which have primarily addressed carbon credits, energy certificates, or sensor metadata, this framework integrates statistical analysis into the verification pipeline, creating an end-to-end auditable system from raw observation to analytical interpretation. Technical feasibility is evaluated using 8,403 hours of real-world temperature observations from Manila, Philippines, during 2024.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Three-layer architecture of the proposed blockchain-based climate data verification framework.

Layer 1 ingests and validates raw observations, generating cryptographic hashes for each quality-controlled record. Layer 2 stores hashes on-chain via proof-of-stake consensus while maintaining complete records in off-chain distributed storage. Layer 3 applies deterministic statistical methods to detect anomalies. The dashed arrow indicates the correction-by-amendment protocol, in which error corrections are appended as new transactions without altering original records.

https://doi.org/10.1371/journal.pclm.0000764.g001

Organization of the paper

The remainder of this paper is organized as follows. The Materials and Methods section describes the data source, preprocessing and quality-control procedures, the statistical anomaly-detection method, and the conceptual blockchain verification model. The Results section presents dataset characteristics, detected anomalies, and estimated blockchain performance parameters. The Discussion examines proof-of-stake sustainability, comparison with existing verification systems, implications for climate policy, limitations, and future implementation pathways.

Data source and preprocessing

Hourly air temperature observations were obtained from the NOAA Integrated Surface Database for Station 984290 (Manila, Philippines; 14.52°N, 121.02°E) (Global Hourly - Integrated Surface Database (ISD), 2021). The raw file contained 8,408 hourly records spanning January through December 2024. Before analysis, the following quality-control and preprocessing steps were applied in sequence:

1. Records flagged by NOAA as erroneous (coded −9999) were removed, leaving 8,403 valid observations.
2. Temperature values were converted from the source format (tenths of degrees Celsius) to degrees Celsius.
3. Timestamps were parsed and aligned to hourly intervals in Coordinated Universal Time (UTC).
4. Data completeness was assessed against the 8,784 expected hourly observations for the full year (2024 is a leap year), yielding 95.7% completeness.

These steps constitute the data ingestion layer of the framework. In an operational blockchain deployment, each validated observation would generate a SHA-256 cryptographic hash at this stage, prior to commitment to the immutable ledger, ensuring that only quality-controlled records enter the verification pipeline.

Statistical anomaly detection

Data cleaning, aggregation, and statistical analyses were performed using Python 3.11 (Pandas 2.2.2, NumPy 1.26) within a Jupyter Notebook environment. A deterministic global-baseline z-score method was selected for anomaly detection. This approach was chosen over machine-learning alternatives for three reasons: (a) it requires no training data, eliminating dependence on historical datasets that may themselves contain unverified observations; (b) the algorithm is fully deterministic, meaning any user with the same input data and threshold will obtain identical results; and (c) the method is transparent to non-specialists, supporting the framework's goal of end-to-end auditability by stakeholders without advanced statistical expertise.

The global mean and standard deviation were computed across the full dataset:

(1)

(2)

where x_i denotes the observed temperature at hour i and N is the total number of valid observations. The standardized anomaly score for each observation was then computed as:

(3)

Observations with |z_i| > 3.0 were classified as statistically extreme. The threshold of 3.0 standard deviations was selected because it corresponds to the conventional criterion for identifying outliers in approximately normal distributions, flagging values expected to occur less than 0.3% of the time under stationary conditions. To assess sensitivity, the analysis was repeated at thresholds of |z| > 2.5 and |z| > 3.5. At |z| > 2.5, the number of flagged observations increased to 122, introducing detections from 10 additional months—including both high- and low-temperature extremes—that did not correspond exclusively to documented heat events. At |z| > 3.5, no observations were flagged. The threshold of 3.0 isolated 33 anomalies exclusively within the April–May heat wave period, providing the best balance between sensitivity and specificity for this dataset. The modular design permits users to substitute alternative thresholds or detection methods without altering the surrounding verification architecture.

Detected anomalies were cross-checked against contemporaneous meteorological reports, confirming the April–May 2024 regional heat events in the Philippines [28]. Output tables include timestamps, z-scores, and absolute temperature values.

Conceptual blockchain verification model

A proof-of-stake architecture was modeled conceptually to demonstrate how verified climate observations could be cryptographically secured. The framework is parameterized using Ethereum PoS as a reference model, which provides a well-documented reference implementation with established performance characteristics [23]. Key design parameters are as follows. Each validated observation generates a SHA-256 cryptographic hash, which is stored on-chain, while the complete observation record is stored off-chain in distributed storage such as the InterPlanetary File System (IPFS), a peer-to-peer network that enables content-addressed, verifiable data storage. Under the Ethereum PoS model, block time averages approximately 12 seconds, with transaction finality achieved within approximately 13 minutes (two epochs of 32 slots), meaning that once committed, observation records cannot be retroactively altered without controlling a supermajority of staked validators [29]. For a pilot climate-data network, a small set of geographically distributed validator nodes operated by independent institutions—such as meteorological agencies, universities, or civil society organizations—could provide an initial level of decentralization, with the network expandable to a larger consortium as participation grows.

Energy efficiency estimates were derived from published Ethereum PoS data. Following the September 2022 transition from proof of work, Ethereum's annual energy consumption fell from approximately 78 terawatt-hours to approximately 0.0026 terawatt-hours, a reduction of 99.95% [23]. For the present framework, estimated energy consumption for a 90-day deployment processing 2,160 observations is approximately 0.06 kilowatt-hours, based on published Ethereum proof-of-stake benchmarks.

Because blockchain ledgers are append-only, data points committed to the chain cannot be deleted or overwritten. If a sensor malfunction or quality-control failure is identified after commitment, the framework addresses this through a correction-by-amendment protocol rather than data deletion. A new transaction is appended to the chain containing: (a) a reference to the original observation hash, (b) the corrected value or a retraction flag, and (c) a cryptographic signature from the correcting authority. The original record remains on-chain as a permanent audit trail, while the framework would be designed so that downstream queries and analyses resolve to the most recent valid version. This approach preserves the immutability guarantee — no historical record is altered — while providing a transparent, auditable mechanism for error correction that mirrors the retraction and correction practices standard in scientific publishing.

Performance assessment

Analytical performance was measured by the number and temporal clustering of detected anomalies, as well as by computational efficiency. For a representative 90-day dataset, total storage requirements were approximately 2.1 MB, and analyses were executed in under one minute on a standard desktop computer.

Ethics statement

This research used publicly available environmental data and involved no human participants, animals, or identifiable information. Large language models (Claude, Anthropic) were used for language editing, formatting, and coding assistance; the author reviewed, verified, and is fully responsible for all content.

Dataset characteristics (NOAA Station 984290, Manila)

From 1 January to 31 December 2024, the file contained 8,408 hourly records, of which 8,403 included valid temperature values (5 entries were flagged −9999). Relative to the 8,784 expected hours (2024 is a leap year), completeness was 95.7%. Temperatures (converted from tenths of °C) ranged from 22.0 to 38.0 °C (mean 29.17 °C, SD 2.56 °C).

Anomaly detection

A deterministic global baseline z-score method was used to flag statistical extremes (|z| > 3). The results provided in the project output file (anomalies.csv) list 33 hourly anomalies between 36.9 and 38.0 °C (Fig 2). These were temporally clustered in April (n = 21) and May (n = 12) 2024, consistent with regional heat episodes [28]. Independent recomputation using the archived dataset confirmed all dates, temperatures, and z-scores listed in the project output file (anomalies.csv). The blockchain verification layer was modeled but not executed in code; the repository reflects the statistical analysis only.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Hourly temperature and flagged anomalies, Manila Station 984290 (Jan–Dec 2024).

Hourly temperatures (°C) are shown as a thin line; the global mean is shown as a thick line. Points indicate hours flagged as statistical extremes using a global-baseline z-score threshold of |z| > 3. Anomalies cluster in April–May. Data: NOAA Integrated Surface Database, Station 984290 (Manila, Philippines).

https://doi.org/10.1371/journal.pclm.0000764.g002

Conceptual performance estimates (architecture feasibility)

To assess the feasibility of a verification layer, we modeled a proof-of-stake design using on-chain content hashes with off-chain storage. Under this design, a representative 90-day deployment would require on the order of ~138 kB on-chain metadata and ~2.1 MB off-chain data, with second-scale ingestion latency and substantially lower energy use than proof-of-work systems. These are model-based estimates, not measurements from a live chain (Table 2).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Estimated storage requirements for a 90-day climate dataset (2,160 observations).

https://doi.org/10.1371/journal.pclm.0000764.t002

Proof-of-Stake consensus and environmental sustainability

The energy efficiency estimates presented here rely fundamentally on proof-of-stake (PoS) consensus, which differs categorically from proof-of-work (PoW) mechanisms. PoW systems like Bitcoin require massive computational competition to validate transactions, consuming ~120 terawatt-hours annually—equivalent to the electricity use of entire nations [31]. In contrast, PoS validators are selected based on staked assets rather than computational power, eliminating energy-intensive mining. Ethereum's transition from PoW to PoS in September 2022 reduced network energy consumption by 99.95%, demonstrating that blockchain verification need not conflict with climate mitigation goals [23].

For climate data applications, this distinction is non-negotiable. A PoW-based climate verification system would consume more energy than the monitoring infrastructure it seeks to secure, creating an unacceptable environmental cost. PoS architectures avoid this paradox entirely while maintaining cryptographic security [29,32]. Moreover, modern PoS platforms such as Ethereum support smart contracts—self-executing code that can automate data validation, dispute resolution, and governance mechanisms directly on-chain [33]. This programmability enables the framework to enforce quality control protocols, trigger alerts for anomalies, and manage multi-institutional data contributions without centralized administration [34].

Comparison with Centralized and Alternative Verification Systems

Current climate data management relies primarily on institutional repositories such as the National Oceanic and Atmospheric Administration's (NOAA) Integrated Surface Database, the European Centre for Medium-Range Weather Forecasts' (ECMWF) ERA5 reanalysis, and national meteorological archives [35,36]. These systems employ rigorous internal quality control but lack mechanisms for external stakeholders to verify data integrity or reproduce analytical workflows independently [37]. Recent initiatives promoting FAIR principles (Findable, Accessible, Interoperable, Reusable) have improved metadata standards and data accessibility, but do not address immutability or tamper-evidence [20].

Alternative approaches to building trust include data citation standards, digital object identifiers (DOIs), and version-control systems such as Git [22]. While valuable for attribution and reproducibility, these tools do not prevent retrospective alteration of archived data. Cryptographic checksums (e.g., SHA-256 hashes) can detect tampering but require trusted third parties to maintain hash registries [21]. Distributed ledger technology eliminates this single point of failure by distributing verification across independent validators.

Prior blockchain applications in climate domains have focused primarily on carbon credit tracking [17], renewable energy certificates [38], and supply chain transparency [39]. Several projects have explored blockchain for environmental sensor networks [24], but these typically verify sensor metadata rather than integrating statistical analysis into the verification layer. The framework presented here uniquely combines immutable data storage with on-chain or verifiable off-chain computation of anomaly detection, creating an end-to-end auditable pipeline from observation to interpretation. Table 3 compares these approaches across eight verification features relevant to climate data systems.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Comparison of verification approaches for climate data systems.

https://doi.org/10.1371/journal.pclm.0000764.t003

Interpretation in context

The results show that simple, well-documented statistics (global mean, standard deviation, and z-scores) can yield reproducible flags that non-specialists can audit. This transparency is critical: if only experts can verify analytical methods, the trust gap persists. By using deterministic algorithms with published code [30], any stakeholder with basic programming skills can independently reproduce the anomaly flags and confirm that the results were not selectively reported or post-hoc manipulated. A blockchain-based verification layer—if implemented as modeled—could add tamper-evident provenance to these same observations without altering the analytical logic, thereby addressing the trust gap between collection, curation, and interpretation. The approach is designed to complement, not replace, conventional archives by adding independent verifiability and public audit trails.

Effective implementation requires governance frameworks that define data admission criteria, validator responsibilities, and dispute resolution processes [40]. Multi-stakeholder consortia—including meteorological agencies, academic institutions, and civil society organizations—could operate validator nodes, distributing authority and preventing capture by any single entity. Smart contracts could encode quality standards, automatically flagging observations that deviate from sensor specifications or fail cross-validation checks [41]. Such mechanisms would operationalize the technical verification layer demonstrated in this study.

Implications for climate policy and communication

The framework addresses a specific dimension of the climate trust deficit: concerns about data manipulation and institutional opacity. While transparent verification alone cannot overcome ideological resistance or misinformation campaigns [42], it removes a technical barrier that legitimizes skepticism. When data integrity is independently auditable, debates can shift from questioning whether data were altered to interpreting what verified observations mean [43].

For policymakers, verifiable climate data strengthens the evidentiary basis for mitigation and adaptation decisions [44]. Regulatory frameworks increasingly demand traceable, auditable environmental data—particularly for carbon markets, emissions reporting, and climate finance mechanisms [45]. A distributed verification infrastructure could serve as trusted input for these systems, reducing compliance costs and fraud risks [46].

Implementation Pathways and Future Work

Pilot deployments at multiple stations and variables (e.g., humidity, wind, precipitation) with live, low-energy chains would allow empirical measurement of throughput, costs, and validator behavior. Methodologically, daily or seasonal baselines and multi-metric anomaly scoring (e.g., heat index) could be compared with the present global baseline method. Governance specifications (data admission rules, dispute resolution, versioning) should be operationalized and tested. The primary intended users of this framework fall into three categories: meteorological agencies and data producers, who would operate validator nodes and contribute quality-controlled observations; researchers and policymakers, who would query verified datasets and reproduce analyses; and civil society organizations and the public, who would audit data provenance and analytical outputs without requiring specialized expertise.

Successful adoption requires addressing technical, institutional, and social barriers [52]. Meteorological agencies may resist ceding control over data validation to distributed networks, requiring a phased integration in which blockchain verification complements rather than replaces existing workflows. Validator node operation must be economically sustainable, either through public funding models or tokenized incentive mechanisms that reward data contributions [53].

Socio-economic barriers also warrant consideration. Deploying and maintaining validator nodes requires computational infrastructure, stable internet connectivity, and technical personnel — resources that are unevenly distributed across institutions and regions. Low-income countries and smaller meteorological agencies may lack the capacity to participate without external support, risking a digital divide in which verification infrastructure mirrors existing inequalities in data access [8]. Addressing this requires tiered participation models in which institutions contribute according to capacity — for example, as data providers, light-client verifiers, or full validator nodes — alongside targeted funding from international climate finance mechanisms or multilateral organizations. The framework's low energy requirements (approximately 0.06 kilowatt-hours per 90-day deployment) substantially reduce the infrastructure threshold compared with proof-of-work systems, but connectivity and training costs remain nontrivial for resource-constrained settings.

Interoperability with legacy systems is essential. Application programming interfaces (APIs) can enable traditional databases to cross-reference blockchain-verified records, allowing gradual migration without disrupting operational forecasting and research pipelines [54]. Educational initiatives must target both data producers (meteorological technicians, sensor operators) and data users (researchers, policymakers) to build capacity for distributed verification literacy.

International coordination presents both opportunities and challenges. A global climate data blockchain could transcend national boundaries, enabling cross-border verification and reducing duplication of monitoring infrastructure. However, geopolitical tensions and differing regulatory frameworks complicate consensus on governance structures [55]. Regional pilots—such as networks covering specific climate zones or transboundary watersheds—may provide tractable proving grounds before scaling globally [56].