OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes
Forecasting particulate matter (PM10) requires both station-scale accuracy and continuous spatial fields, especially during severe dust storms. Chemical transport models (CTMs) provide gridded forecasts but retain local biases, whereas graph neural networks (GNNs) track monitoring sites well at shor...
Na minha lista:
| Autores principales: | , |
|---|---|
| Formato: | Artigo |
| Publicado em: |
2026
|
| Assuntos: | |
| Acesso em linha: | https://arxiv.org/abs/2607.11896 http://arxiv.org/abs/2607.11896 |
| Tags: |
Sem tags, seja o primeiro a adicionar uma tag!
|
| _version_ | 1879179669363752960 |
|---|---|
| author | He, Shuangshuang Wang, Shuo |
| author_browse | He, Shuangshuang Wang, Shuo |
| author_facet | He, Shuangshuang Wang, Shuo |
| author_sort | He, Shuangshuang |
| building | arXiv |
| collection | Acceso abierto |
| description | Forecasting particulate matter (PM10) requires both station-scale accuracy and continuous spatial fields, especially during severe dust storms. Chemical transport models (CTMs) provide gridded forecasts but retain local biases, whereas graph neural networks (GNNs) track monitoring sites well at short lead times but do not produce gridded outputs. Here we present OmniPM-Net, a Convolutional Conditional Neural Process (ConvCNP)-based fusion model that reconciles these two forecast types within a shared spatial representation. A terrain-aware Gaussian set convolution lifts irregular GNN station forecasts onto a regular grid, where a multi-scale Spatial Source Attention (SSA) module blends them with Copernicus Atmosphere Monitoring Service (CAMS) forecasts; a shared omni-query readout then decodes this representation into consistent PM10 predictions at either stations or grid cells over a 108 h horizon. Evaluated across 1,618 air-quality monitoring stations throughout China over the full year of 2024, OmniPM-Net matches the station-level accuracy of the stronger GNN baseline (mean absolute error 21.14 versus 22.00 ug/m3) and reduces the CAMS mean absolute error by 30%, while simultaneously delivering the gridded fields that the discrete GNN cannot. Its clearest gains are in the high-concentration tail, where the 90th-percentile MAE falls by 9% relative to the GNN and 25% relative to CAMS, and during dust episodes, where it improves categorical detection skill while tracking the evolving spatial trajectory. |
| first_indexed | 2026-10-16T01:36:17Z |
| format | Article |
| fullrecord | <oai_dc:dc xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ https://www.openarchives.org/OAI/2.0/oai_dc.xsd"><identifier>arxiv-2607.11896</identifier><datestamp>2026-07-15</datestamp>
<dc:title>OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes</dc:title>
<dc:creator>He, Shuangshuang</dc:creator>
<dc:creator>Wang, Shuo</dc:creator>
<dc:subject>Machine Learning</dc:subject>
<dc:subject>Artificial Intelligence</dc:subject>
<dc:subject>Atmospheric and Oceanic Physics</dc:subject>
<dc:description>Forecasting particulate matter (PM10) requires both station-scale accuracy and continuous spatial fields, especially during severe dust storms. Chemical transport models (CTMs) provide gridded forecasts but retain local biases, whereas graph neural networks (GNNs) track monitoring sites well at short lead times but do not produce gridded outputs. Here we present OmniPM-Net, a Convolutional Conditional Neural Process (ConvCNP)-based fusion model that reconciles these two forecast types within a shared spatial representation. A terrain-aware Gaussian set convolution lifts irregular GNN station forecasts onto a regular grid, where a multi-scale Spatial Source Attention (SSA) module blends them with Copernicus Atmosphere Monitoring Service (CAMS) forecasts; a shared omni-query readout then decodes this representation into consistent PM10 predictions at either stations or grid cells over a 108 h horizon. Evaluated across 1,618 air-quality monitoring stations throughout China over the full year of 2024, OmniPM-Net matches the station-level accuracy of the stronger GNN baseline (mean absolute error 21.14 versus 22.00 ug/m3) and reduces the CAMS mean absolute error by 30%, while simultaneously delivering the gridded fields that the discrete GNN cannot. Its clearest gains are in the high-concentration tail, where the 90th-percentile MAE falls by 9% relative to the GNN and 25% relative to CAMS, and during dust episodes, where it improves categorical detection skill while tracking the evolving spatial trajectory. </dc:description>
<dc:date>2026-06-12</dc:date>
<dc:type>text</dc:type>
<dc:identifier>https://arxiv.org/abs/2607.11896</dc:identifier>
<dc:identifier>http://arxiv.org/abs/2607.11896</dc:identifier>
</oai_dc:dc> |
| id | arxiv-2607.11896 |
| institution | arXiv |
| last_indexed | 2026-10-16T01:36:17Z |
| publishDate | 2026 |
| publishDateRange | 2026 |
| publishDateSort | 2026 |
| record_format | oai_dc |
| spelling | arxiv-2607.118962026-07-15 OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes He, Shuangshuang Wang, Shuo Machine Learning Artificial Intelligence Atmospheric and Oceanic Physics Forecasting particulate matter (PM10) requires both station-scale accuracy and continuous spatial fields, especially during severe dust storms. Chemical transport models (CTMs) provide gridded forecasts but retain local biases, whereas graph neural networks (GNNs) track monitoring sites well at short lead times but do not produce gridded outputs. Here we present OmniPM-Net, a Convolutional Conditional Neural Process (ConvCNP)-based fusion model that reconciles these two forecast types within a shared spatial representation. A terrain-aware Gaussian set convolution lifts irregular GNN station forecasts onto a regular grid, where a multi-scale Spatial Source Attention (SSA) module blends them with Copernicus Atmosphere Monitoring Service (CAMS) forecasts; a shared omni-query readout then decodes this representation into consistent PM10 predictions at either stations or grid cells over a 108 h horizon. Evaluated across 1,618 air-quality monitoring stations throughout China over the full year of 2024, OmniPM-Net matches the station-level accuracy of the stronger GNN baseline (mean absolute error 21.14 versus 22.00 ug/m3) and reduces the CAMS mean absolute error by 30%, while simultaneously delivering the gridded fields that the discrete GNN cannot. Its clearest gains are in the high-concentration tail, where the 90th-percentile MAE falls by 9% relative to the GNN and 25% relative to CAMS, and during dust episodes, where it improves categorical detection skill while tracking the evolving spatial trajectory. 2026-06-12 text https://arxiv.org/abs/2607.11896 http://arxiv.org/abs/2607.11896 |
| spellingShingle | Machine Learning Artificial Intelligence Atmospheric and Oceanic Physics He, Shuangshuang Wang, Shuo OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes |
| title | OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes |
| title_full | OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes |
| title_fullStr | OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes |
| title_full_unstemmed | OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes |
| title_short | OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes |
| title_sort | omnipmnet bridging discrete and gridded pm10 forecasts via omni query neural processes |
| topic | Machine Learning Artificial Intelligence Atmospheric and Oceanic Physics |
| url | https://arxiv.org/abs/2607.11896 http://arxiv.org/abs/2607.11896 |
| work_keys_str_mv | AT heshuangshuang omnipmnetbridgingdiscreteandgriddedpm10forecastsviaomniqueryneuralprocesses AT wangshuo omnipmnetbridgingdiscreteandgriddedpm10forecastsviaomniqueryneuralprocesses |