# Luitel and coauthors assess GraphCast’s Indian Summer Monsoon forecasts [Canonical reading](https://escapemeridian.com/readings/luitel-and-coauthors-assess-graphcast-s-indian-summer-monsoon-71138e21) Somnath Luitel, Parthasarathi Mukhopadhyay, Manmeet Singh, Sandeep Juneja and Saptarishi Dhanuka evaluate GraphCast forecasts for the Indian Summer Monsoon against ERA-5 reanalysis and IMERG satellite precipitation. They examine June–September seasons from 2021–2024, at forecast leads of 24, 48 and 72 hours. The authors report that GraphCast reproduces monsoon rainfall’s spatial pattern and annual cycle with high fidelity at short leads, and that daily temporal correlation with IMERG is high. But they also report moderately suppressed temporal variance and underrepresentation of extreme rainfall and the 95th percentile. The results distinguish rainfall’s average spatial pattern from its variability and intensity: stronger performance on the former does not mean the model represents intense events equally well. The study also reports that GraphCast closely captures the vertical structure and seasonal evolution of Q1 and Q2, but degrades in northward-propagating intraseasonal variability. These findings describe the tested forecasts and period; the abstract does not establish whether they extend to other seasons, regions or forecast settings. Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source arXiv physics.ao-ph · Article Original source date: Jun 30, 2026 (2026-06-30). Date basis: Date basis not provided. Added to Escape Meridian: 2026-11-10T08:02:02.863Z. First observed: 2026-11-03T08:51:12.903Z. Retrieved: 2026-11-03T18:01:06.747Z. - [GraphCast Skill and Systematic Biases in Indian Summer Monsoon Forecasts: Evaluation Against ERA5 and IMERG](https://arxiv.org/abs/2607.11905) ## Limits The abstract describes deterministic forecasts initialized at 00 UTC, for June–September 2021–2024 and leads of 24, 48 and 72 hours; it does not establish whether the results generalize beyond these settings. The supplied evidence is the paper’s abstract and does not provide further methodological detail or results beyond its summary. ## Open question Do the reported biases persist for other forecast leads, seasons or evaluation periods? ## Source-stated The authors evaluated deterministic GraphCast forecasts for the Indian Summer Monsoon against ERA-5 reanalysis and IMERG satellite precipitation for June–September 2021–2024, at 24-, 48- and 72-hour leads. > against ERA-5 reanalysis and IMERG satellite precipitation for the boreal summer season (June-September, JJAS) over 2021-2024 - [Supporting source](https://arxiv.org/abs/2607.11905) ## Source-stated The authors report that GraphCast reproduced the spatial pattern and annual cycle of monsoon rainfall with high fidelity at short leads, with high daily temporal correlation with IMERG. > GraphCast reproduces the spatial pattern and annual cycle of monsoon rainfall with high fidelity at short leads - [Supporting source](https://arxiv.org/abs/2607.11905) ## Source-stated The authors report suppressed temporal variance and underrepresentation of extreme rainfall and the 95th percentile. > temporal variance is moderately suppressed across timescales (power-spectrum variance ratio ~0.63), and the intensity distribution systematically underrepresents extreme events and the 95th percentile. - [Supporting source](https://arxiv.org/abs/2607.11905) ## Source-stated The authors report that GraphCast captures the vertical structure and seasonal evolution of Q1 and Q2, but degrades in northward-propagating intraseasonal variability. > GraphCast closely captures the vertical structure and seasonal evolution of Q1 and Q2, but degrades in northward-propagating intraseasonal variability. - [Supporting source](https://arxiv.org/abs/2607.11905) ## Source and contributor context Luitel, Somnath, Mukhopadhyay, Parthasarathi, Singh, Manmeet, Juneja, Sandeep, Dhanuka, Saptarishi · Article author These are credits stated by the source; they do not establish a profile match or authorship of every passage. ## Source media No suitable media was found by the supported page extraction. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## arXiv arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed. - [Explore this context](/entities/arxiv) - [Reference](https://arxiv.org/) - [Reference](https://info.arxiv.org/help/api/index.html) ## Editorial provenance Version 1. Publication ID: e9038532-0eab-4ee2-8668-13a5a82f2c31. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-10T08:02:02.863Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: material\_development. Observation: 0fa68c90a79262946d8bcba1ec8d72ed6377b27fae6e97c073e9e07cc17d46a0. Source text hash: fc59da7efca29f67f3372de717840619db6d6b29e86fc998b768f1a76b6707ba. ## Revision and source-check history ### Version 1 — 2026-11-10T08:02:02.863Z Somnath Luitel, Parthasarathi Mukhopadhyay, Manmeet Singh, Sandeep Juneja and Saptarishi Dhanuka evaluate GraphCast forecasts for the Indian Summer Monsoon against ERA-5 reanalysis and IMERG satellite precipitation. They examine June–September seasons from 2021–2024, at forecast leads of 24, 48 and 72 hours. The authors report that GraphCast reproduces monsoon rainfall’s spatial pattern and annual cycle with high fidelity at short leads, and that daily temporal correlation with IMERG is high. But they also report moderately suppressed temporal variance and underrepresentation of extreme rainfall and the 95th percentile. The results distinguish rainfall’s average spatial pattern from its variability and intensity: stronger performance on the former does not mean the model represents intense events equally well. The study also reports that GraphCast closely captures the vertical structure and seasonal evolution of Q1 and Q2, but degrades in northward-propagating intraseasonal variability. These findings describe the tested forecasts and period; the abstract does not establish whether they extend to other seasons, regions or forecast settings. ## Original source arXiv physics.ao-ph · Article Original source date: Jun 30, 2026 (2026-06-30). Date basis: Date basis not provided. Added to Escape Meridian: 2026-11-10T08:02:02.863Z. First observed: 2026-11-03T08:51:12.903Z. Retrieved: 2026-11-03T18:01:06.747Z. - [GraphCast Skill and Systematic Biases in Indian Summer Monsoon Forecasts: Evaluation Against ERA5 and IMERG](https://arxiv.org/abs/2607.11905) ## Limits The abstract describes deterministic forecasts initialized at 00 UTC, for June–September 2021–2024 and leads of 24, 48 and 72 hours; it does not establish whether the results generalize beyond these settings. The supplied evidence is the paper’s abstract and does not provide further methodological detail or results beyond its summary. ## Open question Do the reported biases persist for other forecast leads, seasons or evaluation periods? ## Source-stated The authors evaluated deterministic GraphCast forecasts for the Indian Summer Monsoon against ERA-5 reanalysis and IMERG satellite precipitation for June–September 2021–2024, at 24-, 48- and 72-hour leads. > against ERA-5 reanalysis and IMERG satellite precipitation for the boreal summer season (June-September, JJAS) over 2021-2024 - [Supporting source](https://arxiv.org/abs/2607.11905) ## Source-stated The authors report that GraphCast reproduced the spatial pattern and annual cycle of monsoon rainfall with high fidelity at short leads, with high daily temporal correlation with IMERG. > GraphCast reproduces the spatial pattern and annual cycle of monsoon rainfall with high fidelity at short leads - [Supporting source](https://arxiv.org/abs/2607.11905) ## Source-stated The authors report suppressed temporal variance and underrepresentation of extreme rainfall and the 95th percentile. > temporal variance is moderately suppressed across timescales (power-spectrum variance ratio ~0.63), and the intensity distribution systematically underrepresents extreme events and the 95th percentile. - [Supporting source](https://arxiv.org/abs/2607.11905) ## Source-stated The authors report that GraphCast captures the vertical structure and seasonal evolution of Q1 and Q2, but degrades in northward-propagating intraseasonal variability. > GraphCast closely captures the vertical structure and seasonal evolution of Q1 and Q2, but degrades in northward-propagating intraseasonal variability. - [Supporting source](https://arxiv.org/abs/2607.11905) ## Source and contributor context Luitel, Somnath, Mukhopadhyay, Parthasarathi, Singh, Manmeet, Juneja, Sandeep, Dhanuka, Saptarishi · Article author These are credits stated by the source; they do not establish a profile match or authorship of every passage. ## Source media No suitable media was found by the supported page extraction. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## arXiv arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed. - [Explore this context](/entities/arxiv) - [Reference](https://arxiv.org/) - [Reference](https://info.arxiv.org/help/api/index.html) ## Editorial provenance Version 1. Publication ID: e9038532-0eab-4ee2-8668-13a5a82f2c31. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-10T08:02:02.863Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: material\_development. Observation: 0fa68c90a79262946d8bcba1ec8d72ed6377b27fae6e97c073e9e07cc17d46a0. Source text hash: fc59da7efca29f67f3372de717840619db6d6b29e86fc998b768f1a76b6707ba. ### uncertain — 2026-11-10T08:02:02.863Z — anthropic/claude-opus-5.5 Publication: e9038532-0eab-4ee2-8668-13a5a82f2c31 attribution — pass: The draft names the five listed authors and consistently frames the findings as what the authors report or evaluate. It does not present them as Escape Meridian conclusions. > Authors:Somnath Luitel, Parthasarathi Mukhopadhyay, Manmeet Singh, Sandeep Juneja, Saptarishi Dhanuka summary — uncertain: The facts are faithful, but a general reader is not told that GraphCast is Google's machine-learning global weather prediction system, which is central to the AI relevance. Q1 and Q2 appear without explanation; the source defines them as the apparent heat source and moisture sink. ERA-5 reanalysis and IMERG are also unexplained. The draft omits the authors' overall framing of moderate underrepresentation rather than a wet bias or variance collapse, and their benchmark purpose for AI weather models. Smallest repair: identify GraphCast, gloss Q1/Q2, and add the authors' concluding characterization. > Google's GraphCast, a machine-learning global weather prediction system date — pass: originalPublished is 2026-06-30, which matches the v1 submission date. It is not the date of the later v3 revision or of the feed entry. > \[Submitted on 30 Jun 2026 (v1), last revised 30 Oct 2026 (this version, v3)\] support — pass: Every claim matches the abstract. The interpretive sentence about pattern versus intensity follows from the reported results. The limitations and the question about generalization are appropriately conditional, and the draft does not turn the evaluation into a real-world reliability claim. > the intensity distribution systematically underrepresents extreme events and the 95th percentile media — pass: The selected media is the verified, direct arXiv abstract page for this paper, which is the actual source. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. No public references recorded here yet. [Reference data (JSON)](/readings/71138e217d1032d21c90fc17f1dffcc4756f1f0356db0332f643400919d06fe7/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/luitel-and-coauthors-assess-graphcast-s-indian-summer-monsoon-71138e21.json)