Research map / Probabilistic and scientific modeling / Dynamical systems
SENDAI: A Hierarchical Sparse-measurement, EfficieNt Data AssImilation Framework
Xingyue Zhang, Yuxuan Bao, Mars Liyao Gao, J. Nathan Kutz
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add hierarchical frequency peeling (HFP) (ablation) | Cheap2Rich · Imperial Valley NDVI (64 sensors) | SSIM: 0.4041 -> 0.4411 (+9.2%) | Improved |
| add hierarchical frequency peeling (HFP) (ablation) | Cheap2Rich · Tarim Basin NDVI (64 sensors) | SSIM: 0.3505 -> 0.4257 (+21.5%) | Improved |
| add hierarchical frequency peeling (HFP) (ablation) | Cheap2Rich · Riverina NDVI (64 sensors) | SSIM: 0.2761 -> 0.3158 (+14.4%) | Improved |
| add coordinate-based INR decoder on top of HFP (ablation) | Cheap2Rich+HFP · Imperial Valley NDVI (64 sensors) | SSIM: 0.4411 -> 0.4668 (total +15.5% from Cheap2Rich) | Improved |
| add coordinate-based INR decoder on top of HFP (ablation) | Cheap2Rich+HFP · Tarim Basin NDVI (64 sensors) | SSIM: 0.4257 -> 0.4777 (total +36.3% from Cheap2Rich) | Improved |
| add coordinate-based INR decoder on top of HFP (ablation) | Cheap2Rich+HFP · Riverina NDVI (64 sensors) | SSIM: 0.3158 -> 0.3354 (total +21.5% from Cheap2Rich) | Improved |
| add HFP to Cheap2Rich (ablation) | Cheap2Rich · Riverina NDVI (64 sensors) | RMSE: 0.1537 -> 0.2526 | Hurt |
| add HFP to Cheap2Rich (ablation) | Cheap2Rich · Tarim Basin NDVI (64 sensors) | RMSE: 0.1827 -> 0.1466 | Improved |
| use SENDAI Jr. latent alignment instead of best baseline (SG+IDW) | SENDAI Jr. · Central Valley NDVI (64 sensors) | SSIM: 0.2612 -> 0.5747 (+120.0%) | Improved |
| use SENDAI Jr. instead of best baseline (SG+IDW) | SENDAI Jr. · Corn Belt NDVI (64 sensors) | SSIM: 0.1588 -> 0.4530 (+185.3%) | Improved |
| use SENDAI Jr. instead of best baseline (SG+IDW) | SENDAI Jr. · Guadalquivir NDVI (64 sensors) | SSIM: 0.1849 -> 0.3655 (+97.7%) | Improved |
| use SENDAI Jr. instead of SG+IDW | SENDAI Jr. · Guadalquivir NDVI (64 sensors) | RMSE: 0.1444 -> 0.1474 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Cheap2Rich — Cheap2Rich: A Multi-Fidelity Framework for Data Assimilation and System Identification of Multiscale Physics - Rotating Detonation Engines (ICML 2026) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Continuous Field Reconstruction from Sparse Observations with Implicit Neural Networks (ICLR 2024 · similarity 0.86)
- Sparsely-Supervised Data Assimilation via Physics-Informed Schrödinger Bridge (NeurIPS 2026 · similarity 0.86)
- DAWP: A framework for global observation forecasting via Data Assimilation and Weather Prediction in satellite observation space (NeurIPS 2025 · similarity 0.85)
- Generating Full-field Evolution of Physical Dynamics from Irregular Sparse Observations (NeurIPS 2025 · similarity 0.85)
- Deep Random Features for Scalable Interpolation of Spatiotemporal Data (ICLR 2025 · similarity 0.84)
- Latent-EnSF: A Latent Ensemble Score Filter for High-Dimensional Data Assimilation with Sparse Observation Data (ICLR 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
