The Revista Cubana de Meteorología will only receive technical or scientific (non-informative), original and unpublished works that are not simultaneously postulated in other journals or editorial bodies. The text must be unobjectionable in terms of its presentation and content. In the selection of the materials proposed to be published, the degree of scientific contribution, interest of the specialized community, scientific strength and clarity, originality, timeliness of the information, seriousness in the treatment of the subject, its writing, and the compliance with the rules of editorial ethics. Articles that do not comply with the Instructions to Authors will not be accepted for publication.

Articles that do not comply with the instructions to the authors will not be accepted for publication. The Editorial Committee reserves the right to accept or not the documents presented to the journal.

Phase 1: Plagiarism detection and editorial quality assessment

Below you will find the detailed policy for the initial phase, which applies from the receipt of a manuscript to the completion of plagiarism and editorial quality checks. The procedure, responsibilities, evaluation criteria, specific actions for non-compliance, and brief communication templates with authors are described step by step.

This journal penalizes plagiarism and fraudulent data in documents, publishing the same article in more than one journal, copying and using textual information from any document, using data, images and other resources without the authorization of all natural or legal authors of the work, using information that has not been consulted from the original source by some of the authors of the document, modifying or misrepresenting the meaning and context of the information contained in other works, as well as any other inappropriate use that alters the integrity of the information.

To detect plagiarism, the editors and referees of the journal use free software Plagiarisma and Editpad. These can detect plagiarism in more than 190 languages, use different search engines and support a large number of document extensions.

For the detection of plagiarism of graphic resources in the journal we use the service TinEye. The service offers some of the fastest and most accurate image recognition APIs in the world.

To detect texts generated with Artificial Intelligence such as ChatGPT, the journal recommends using specialized tools such as GPTZero. These AI anti-plagiarism models are under constant development and are being endowed with very advanced capabilities for this function.

1. Purpose and Scope

   This policy governs the initial evaluation of papers submitted to the journal, understood as:

   • Originality verification (plagiarism detection: texts, images, and AI-generated content), and
   • Evaluation of compliance with editorial quality requirements (with the support of an AI system and human review).

   It does not cover the refereeing process (double-blind peer review) or final editorial decisions on acceptance after academic review.

2. General Principles
   • Integrity: Protect originality and intellectual honesty.
   • Transparency: Explain to authors the methodology and criteria applied.
   • Proportionality: Measures are tailored to the severity of the violation.
   • Human-AI combination: Automated results (anti-plagiarism, AI evaluator) are inputs; the final decision is made by the Editor-in-Chief.
   • Confidentiality: All data and reports are treated confidentially and in accordance with applicable data protection laws.
3. Responsibilities and Timeframes
   • Initial Responsibility: Editor-in-Chief.
   • Technical Support: Production team (for scanning and maintaining reports).
   • Initial Phase Timeframe: Up to 1 week from receipt of the manuscript to complete checks (plagiarism checks and editorial quality assessment). If an issue arises that requires interaction with the author, the deadline will be extended based on the responses received.
4. Reception and Initial Steps (Flow)
   1. Registration: The manuscript is registered in the publishing system and assigned an ID.
   2. Quick administrative checks: basic formatting, numbered files, existence of title, authors, affiliations, abstract, and ORCID, among others.
   3. Automatic anti-plagiarism analysis:
      • Text scanning against databases and repositories (anti-plagiarism software).
      • Image similarity checks (reverse search tools and metadata comparison).
      • Automatically generated fingerprint detection (AI clues) if the tool allows it.
      • A report with evidence (matching fragments, links, image results) is generated and saved in the file.
   4. Editorial Quality Evaluator with AI:
      • Execution of the evaluator developed by Cervantes-producciones Digital S.R.L.
      • The system generates a report by section (title, abstract, introduction, methods, results, conclusions, references, authors/ORCID, CRediT, keywords, etc.) with a score of 0–100 and comments.
   5. Human review:
      • The Editor-in-Chief (or designated editor) reviews both reports (anti-plagiarism and AI) and critically reads the manuscript.
   6. Initial decision (see consequence criteria below): continue with the peer review process or return/reject/request clarification.
5. Technical Criteria and Thresholds (Guidelines)

   Note: Thresholds should be used as a guide. The context of the similarity (citations, standard methods, excerpts from instruments or law) and the quality of the evidence determine the final outcome.

   5.1 Text (Similarity)

   • < 10%: Typically no incidence of textual plagiarism; reviewed by editor.
   • 10–30%: Mandatory manual review; the editor assesses the source of matches and whether they are correctly cited. Clarifications or corrections may be requested.
   • > 30%: High alarm — considered potentially significant plagiarism; requires thorough investigation and possible immediate rejection if there is text without attribution.

   5.2 Images and Figures

   • Exact match or manipulation without permission/documentation → serious issue.
   • If the image is taken from a third party: the author must provide an explicit license or permission and the corresponding citation.

   5.3 AI-Generated Content

   • If the manuscript declares use of AI tools for writing support: assess whether the use falls within the journal's policy (See Policy on the Use of Generative AI and AI-Assisted Technologies).
   • If undeclared use is detected using AI detection tools: request clarification from the author; if there is substantive undeclared use that affects originality/authorship → possible rejection.

   5.4 AI Evaluator of Editorial Quality

   • The evaluator returns scores per section (0–100).
   • Reference threshold for referral to peer review: ≥ 60–70 on average. Used in conjunction with human judgment: a manuscript with a low score may be returned to the author for improvement or rejected for insufficient editorial quality.
6. Procedure when an issue is detected
   1. 6.1 Minor issues (e.g., minor similarities due to common phrases, errors in ORCID, reference formatting)

      Action:

      • The editor prepares a brief report (attaching evidence) and requests correction from the author.
      • Author response time: 10 business days.

      Possible outcomes:

      • Correction and resubmission → process continues (re-evaluation of the corrected portion).
      • No response → archived or administratively rejected.
   2. 6.2 Medium-serious issues (e.g., specific similarity without citation, inconsistencies in CRediT, images without permission)

      Action:

      • Extended investigation: The editor requests a detailed explanation and documentation (permissions, source data, authorship file) within 10–15 business days.
      • If the explanation is plausible and the correction feasible → request formal corrections and re-evaluation.
      • If the explanation is insufficient or nonexistent → rejection and notification to the author with the reason.
   3. 6.3 Serious issues (e.g., plagiarism of substantive content, image manipulation, self-plagiarism without declaration, false attributions)

      Action:

      • Immediate rejection of the manuscript by the Editor-in-Chief.
      • Internal record: retain the anti-plagiarism report, screenshots, and correspondence.
      • Official notification to the author(s) with a copy of the evidence and explanation of the decision.
      • Escalation (when appropriate): Notify the author's employer or other journals if there is a risk of serious academic misconduct, according to journal policy and applicable regulations.
      • Temporary ban on future submissions (if applicable), according to journal regulations.
7. Communication with authors (transparency and evidence)
   • Communication: Always documented in writing (email from the editorial system).
   • Minimum notification content: clear explanation of the incident, accompanying evidence, procedure to follow, deadline, and consequences of not responding.
   • Right to reply: Authors have the opportunity to respond and provide evidence.
8. Review and Appeal
   • Appeal: The author may request reconsideration by submitting new evidence within 15 days of receiving notification of rejection due to incidents.
   • Appeal Committee: Comprised of the Editor-in-Chief and two senior editors/members not involved in the initial decision. They review evidence and issue a final decision.
   • Record of the decision: Minutes and communications are filed in the manuscript record.
9. Use of the AI Reviewer and Decisions
   • The AI reviewer's report does not replace human review; it is a systematic aid that:
   • summarizes strengths and weaknesses by section,
   • identifies issues of style, structure, and standard inconsistencies (ORCID, CRediT, abstract length, etc.),
   • offers scores and recommendations.
   • Interpretation: The editor should verify that the AI's suggestions are not false positives (e.g., repetitive technical terminology) and assess their weight in the final decision.
   • Any conflict between AI and human review should be resolved by additional review by the Editor-in-Chief.
10. Example of decision flow (summarized)
    1. Reception → anti-plagiarism + AI → report.
    2. Review by editor:
       • No issues → goes to peer review.
       • Minor issue → request correction (10 days).
       • Medium issue → investigate, request justification/documents (10–15 days).
       • Major issue → rejection and possible institutional notification.
    3. If corrected satisfactorily → re-evaluate; if not → final rejection.

Phase 2: Peer review

When the editorial process begins, the texts are incorporated into the list of publications. All contributions submitted will be subject to peer review, using the double blind system to ensure impartiality. In this process, the Senior Editor reviews the proposal and assigns an Associate Editor to control the review and editing. The Associate Editor removes the information that identifies its authors and sends the text to two reviewers who can be internal (20%) or external (80%) to the publishing institution. Based on your answers, the type of admission or refusal is decided. The answers can be of four types:

• Admission unchanged
• Admission with changes
• Non-final rejections (if the text in its initial form does not meet the requirements)
• Definitive rejection

In cases of controversy, that is, when one of the two evaluating arbitrators issues an opinion of accepted and the other of rejected or accepted with modifications, an evaluation request is sent to a third arbitrator and after its verdict, the Committee Editorial weighs the three evaluations and will issue the verdict of Admission without changes, Admission with changes, Non-definitive rejections or Definitive rejection. In all cases, the Chief Editor is responsible for the final decision on whether the document is published. In the event of suspicion or non-compliance with some of the good ethical practices established in the magazine's Code and other cases of Controversy, we will proceed as established in the Committee on Publication Ethics (COPE) Flow Charts for each case.

Final rejections should be less likely if the rules for publication have been rigorously followed. The decisions and indications will be communicated to the author by the associate editors. The authors have a period of two weeks to return the corrected publication proposals, otherwise their works will go to the end of the queue of articles to publish. Depending on the volume of changes proposed to the text, the editors reserve the power to submit them to a new cycle of revisions once they have been forwarded by the authors, prior communication to them.

After revised from a scientific point of view, the texts enter the production process, which includes a new editorial correction by the Style Editor. Then the edited work goes to the Designer, who adjusts it to the magazine layout, which can contact the authors in case of difficulties with the format of the images, annexes, among others. Once the document is assembled, it is sent again to the author and a Gallery Reviewer from the Editorial Committee to perform the gallery test. Proposed corrections must be returned within 7 days. Regarding the galley proof, the only corrections that are accepted are spelling, typographical errors or editorial elements that can lead to confusion.

The time that elapses between receipt and final publication of the articles is approximately 4 months. The author and a Gallery Reviewer from the Editorial Committee are again sent to perform the gallery test. Proposed corrections must be returned within 7 business days. Regarding the galley proof, the only corrections that are accepted are spelling, typographical errors or editorial elements that can lead to confusion. The time that elapses between receipt and final publication of the articles is approximately 4 months. The author and a Gallery Reviewer from the Editorial Committee are again sent to perform the gallery test. Proposed corrections must be returned within 7 business days. Regarding the galley proof, the only corrections that are accepted are spelling, typographical errors or editorial elements that can lead to confusion. The time that elapses between receipt and final publication of the articles is approximately 4 months.

To consult the complete arbitration sheet, you can download the file available at the address: https://ojs.edicionescervantes.com/public/site/planilla_revisores_rcm.doc

The arbitrator, from his appointment and during all arbitration proceedings, will remain independent and impartial, qualities necessary for his appointment and maintenance in the assigned role. The arbitrators must not give rise to justified doubts regarding their impartiality or independence (factual or legal), they may be challenged. The arbitration proceedings will promptly reveal such circumstances to the parties. Manuscript evaluation is a voluntary process, which is recognized by the annual publication in the journal of the names of the evaluators who have collaborated during that period.

The arbitrators must exercise the position with strict impartiality and absolute discretion, it is an anonymous process. The appointment of an arbitrator can be accepted or rejected, although it is normal to accept it which establishes the rights of the parties to fulfill the order within the established period of four weeks from the receipt of the article. In case of rejection within a maximum period of 48 hours, the editor or president of the Editorial Board must be notified, avoiding damages that may be caused by delay or breach of the respective obligations. 

It is the duty of the arbitrators, to inform the parties of any situation that jeopardize the performance of their duties. During the communication of the appointment as arbitrator, all circumstances that may give rise to a possible removal must be exposed. The arbitrator, from the moment of his appointment and during the arbitration proceedings, will promptly disclose such circumstances to the parties unless he has already informed them of them.