Unverified50% confidenceFactExact time
ECMWF's ERA5 uses a four-dimensional variational assimilation (4D-Var) method, performing one assimilation cycle every 12 hours and fusing approximately 40 million observational data points.
1
Sources
50%
Confidence
Long-term
Relevance
8/10/2026
First Seen
Sources
DeepMind开源WeatherNext：AI气象预测新标杆
Related Claims
UnverifiedECMWF从2016年开始制作ERA5，利用IFS模型CY41R2版本重新模拟，总数据量超过5PB71% similarUnverifiedNvidia 的 FourCastNet（2022）基于自适应傅里叶神经算子，在 ERA5 再分析数据集上训练，能够在毫秒级别完成全球天气预报60% similarUnverifiedOmniMerge V4 is a model formed by fusing three to four 27B models through weight merging.60% similarUnverifiedV4-Pro uses a Mixture of Experts (MoE) architecture with a total parameter count of 1.6 trillion (1.6T) and 49 billion (49B) active parameters.57% similarUnverifiedDiT（Diffusion Transformer）架构的参数量已达数十亿级别，单次推理的中间激活值在高分辨率下可占满80GB A100的全部显存56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/726764API
curl https://kongchang.com/api/v1/knowledge/claims/726764MCP
get_claim(id=726764)