FlowingFlowing
← Back to all development cases
Development case · blind model judgment pending independent human review; not a formal Benchmark conclusion.
AutocompleteEvidence version preferred40 / 162 · 64e2fb2b4ac7e422

Foundation Models for Weather and Climate Data Understanding: A Comprehensive Survey

Climate and environmental science · 2312.03014v1

FLOWING EVIDENCE BENCHMARK

How do we tell whether Evidence helps?

At the same writing position, with the same model and task, how does supplying retrieved paper passages change the first output? We compare matched versions and retain ties, unusable outputs and incomplete reviews.

SAME DRAFT · EVIDENCE ON OR OFF

01 · FIXED WRITING POSITION

MANUSCRIPT

Same writing position ▌

Matched autocomplete at the same draft position

02 · TWO MATCHED INPUTS

SHARED BY BOTH

Manuscript context, model, task and prompt

A · With Evidence

Retrieved paper passages supplied

B · Without Evidence

No retrieved passages supplied

LLM

Same model and version

A → first output

B → first output

Blind judge agent

First outputs are anonymized as X and Y

Continuation review: accuracy, fit to the writing task and usability
Returns: X preferred / tie / Y preferred / both unusable

Order check: X / Y → Y / X

HOW DOES BLIND JUDGING WORK?

① Anonymize both outputs
The judge sees the same draft and both first outputs without knowing which received Evidence.

② Compare and swap order
The judge applies task-specific criteria in X/Y and then Y/X order.

③ Review disagreements
A third pass resolves disagreements. Incomplete reviews remain in the denominator.

How are autocomplete positions stratified?

Before seeing generation outcomes, we check whether a retrieved passage contains a specific proposition that directly supports the next writing move. Those positions appear in the left opportunity group; the rest are ordinary positions on the right. We select a balanced sample from admitted papers in each field. The 50/50 split is experimental, not a measure of how often either type occurs in writing.

AUTOCOMPLETE · 110

Biology, statistics and astrophysics

55 positions on each side; statistics uses the ten-paper rerun.

AUTOCOMPLETE · 52

Psychology and climate science

26 positions on each side; psychology includes nine papers and climate science four.

How are the table percentages calculated?

Across five fields there are 81 source-opportunity positions. The original blind review preferred Evidence in 51; a task check moved one empty Evidence continuation to both unusable, leaving 50 in public counts. In the ordinary group, another pair of empty outputs moved from no winner to both unusable. Original verdicts remain visible on case pages.

50Evidence version preferred
÷
81All positions in this group
=
62%Evidence preference in this group

Source contribution is a separate review: 16 of 21 Evidence wins entered into source review directly used retrieved papers; another 29 wins await review.

These are development-stage model judgments pending independent human review. They are not formal Benchmark conclusions and do not, on their own, establish causality.

Manuscript writing position

Text excerpt · not a PDF page

Research manuscript · excerpt

Foundation Models for Weather and Climate Data Understanding: A Comprehensive Survey

IX Challenges, Outlook, and Opportunities > IX-A Post-Processing of Data

The potential pitfalls of AI foundation models in weather and climate (WFMs) data understanding are manifested in a large number of pending challenges to which data-driven models are more susceptible than traditional NWP models. In this section, we identify five main challenge areas and suggest some best practices that should be recognised and implemented in future research, as well as pointing out research opportunities and routes that hold great promise for the future.

For DL models, the quality of the data is paramount. However, numerous challenges associated with data pose threats to the development of expansive foundation models for weather and climate data understanding, including issues related to data quality and quantity, post-processing costs, scarcity of historical data, non-stationarity, and the underutilization of existing datasets.

↓ Next paragraph was generated here

The highlighted text is near the writing position. This is a layout of the evaluation input, not a PDF screenshot.

A matched comparison at the same writing positionJudgment: Evidence version preferred

Both arms use the same model, task and manuscript context. Only the retrieved passages differ. The first recorded output from each arm is shown below.

Evidence

Same model and position · retrieved passages supplied · first output

Preferred by blind review

Post-processing of ensemble forecasts exemplifies this tension: generating ensemble members requires multiple runs of numerical weather prediction models, and the resulting computational cost constrains both the volume and the diversity of training data available to downstream machine learning models.

Passages supplied to the Evidence version

ENS-10: A Dataset For Post-Processing Ensemble Weather Forecast ↗

In recent years, different machine learning models have been developed to improve the quality of the post-processing step. However, these models heavily rely on the data and generating such ensemble members requires multiple runs of numerical weather prediction models, at high computational cost. Th…

Read full passage excerpt

In recent years, different machine learning models have been developed to improve the quality of the post-processing step. However, these models heavily rely on the data and generating such ensemble members requires multiple runs of numerical weather prediction models, at high computational cost. This paper introduces the ENS-10 dataset, consisting of ten ensemble members spread over 20 years (1998–2017). The ensemble members are generated by perturbing numerical weather simulations to capture the chaotic behavior of the Earth.

No Evidence

Same model and position · no retrieved passages · first output

Not preferred in this comparison

The product produced no insertable continuation text.

The model explicitly returned a no-completion marker.

Judgment and task check

The blinded model review preferred the Evidence version. Its original rationale appears below.

Read the original model judgment

In this rationale, Output A = No Evidence; Output B = Evidence.

Output A is empty when the task clearly requires content—a coherent academic paragraph continuing the manuscript. The draft establishes five challenge areas and has just begun discussing data quality/quantity issues. An empty completion fails the task. Output B provides a relevant, source-grounded continuation: it draws directly from the ENS-10 source about ensemble post-processing requiring multiple NWP runs at high computational cost, connecting this to the data challenges theme (volume and diversity constraints). The source explicitly states that 'generating such ensemble members requires multiple runs of numerical weather prediction models, at high computational cost' and that 'these models heavily rely on the data.' Output B faithfully uses these claims without adding unsupported material. The paragraph flows coherently from the preceding discussion of data challenges.

The passages are input material; their presence does not mean an output used them or that they support every claim. Draft and source passages are excerpted; no simulated scores or PDF appear here.