Publications·November 30, 2024
This UNESCAP working paper takes stock of how Thailand is using satellite and ground-based remote sensing to understand and manage air pollution. It catalogues the current toolkit (from country EO satellites to geostationary sensors and sun-photometer networks), documents the roll-out and status of Pandora spectrometers and the GEMS (Geostationary Environment Monitoring Spectrometer) data stream, and sets out a pragmatic action plan to move from research to operational, policy-relevant products. The core message: Thailand has strong building blocks (institutional capacity at GISTDA, growing academic output, and multiple regional partnerships), but needs more consistent operations, data integration, open access, and targeted capacity building to translate measurements into real-time decisions.
Backdrop: Air quality, institutions and the role of EO
Thailand’s air-quality problems are multi-source—transport and residential emissions in Bangkok, biomass/open burning in the North, and transboundary plumes affecting the South. The report argues that remote sensing can: (i) fill spatial gaps in the ground network, (ii) track long-range transport, and (iii) support warning/forecasting during severe episodes. It maps the institutional landscape: GISTDA is the national space agency; NARIT supports satellite data processing; line agencies (notably PCD) operate regulatory monitoring; and municipal authorities are pivotal “last-mile” users who can blend satellite and ground data for public advisories and enforcement.
Thailand’s current EO/use cases (quick scan)
National EO missions. From TMSat/Thai-Paht to THEOS/Thaichote, Thailand has homegrown capacity for optical imagery supporting cartography, land use and environmental applications.
PM mapping & AQI. GISTDA generates daily PM₂.₅ maps and AQI, fusing Himawari-8/9 and other satellites, identifies hotspots, and publishes wind/smog forecasts.
Ground-based remote sensing. As of 2023 Thailand hosts ~21 AERONET sites and 3 MPLNET lidars, adding vertical/aerosol-column context to PCD’s reference monitors.
Fire detection. The Thailand Fire Monitoring System (MODIS, Thaichote; with VIIRS fire alerts via Global Forest Watch) provides daily heat-point layers and weekly wildfire shapefiles to guide fuel management and fire-breaks.
Regional tools. Mekong Air Quality Explorer (SERVIR–NASA/USAID) blends satellite and ground data with ML bias-correction for 3-day forecasts; Thailand is a development partner.
What the research ecosystem is producing
An informal ESCAP meta-analysis tallies 44 Thai/Thailand-focused publications using remote sensing for air pollution since 2012, with a sharp uptick during 2020–2023. Topics span NO₂, O₃, SO₂, CO, PM₂.₅/PM₁₀, AOD, burning, and long-range transport, using sensors such as MODIS, VIIRS, TROPOMI, and platforms like Google Earth Engine and NASA Giovanni. This evidences a strong and expanding academic base to support policy.
Pandora in Thailand — where it stands
Three Pandora spectrometers (part of the Pandonia Global Network / Pandora Asia Network) have been installed to provide high-precision column measurements of NO₂, HCHO, O₃ and SO₂ and to validate GEMS:
PAN-190 (Bangkok) — Operational and public after resolving humidity/UPS/tracker issues; clear seasonal patterns are visible (higher NO₂/HCHO in hot season; O₃ peaks in winter & summer; episodic SO₂ spikes). Data-quality flags remain a concern and call for sustained QA/QC.
PAN-213 (Chiang Mai) — Focused on agricultural-burning haze; currently non-operational due to head-sensor connectivity; hardware returned for repair.
PAN-214 (Songkhla) — Operational under validation, tracking transboundary influences; still facing intermittent head-sensor comms issues.
These sites are complemented by planned expansion (e.g., Nakhon Pathom).
GEMS — present use and near-term potential
Thailand is retrieving GEMS L2 in near-real time and building a visualization interface while cross-validating with Pandora. Proposed applications include:
Aerosol/source identification & transport tracking, coupling Visible/UV Aerosol Index, NO₂/HCHO columns, and PM₂.₅/PM₁₀ products with trajectory and chemical transport models;
Near-real-time monitoring for episode response;
Spatiotemporal O₃ analysis in green belts; and
Distinguishing local vs. transported contributions to inform regional cooperation.
An example machine-learning RTL model for PM₂.₅ estimation shows R² ≈ 0.93–0.95 and RMSE ~4–7 µg m⁻³ across 2020–2024—strong performance that degrades modestly at very high concentrations and under changing meteorology, reinforcing the need for continuous validation.
Cooperation across borders and institutions
Thailand participates in the ASEAN Agreement on Transboundary Haze Pollution, has tabled a “Clear Sky” seasonal-haze strategy, and engages in trilateral collaboration with Cambodia and Lao PDR. In January 2024, USAID/TICA/Laos MONRE launched a SERVIR partnership to build a shared, near-real-time air-quality network, aligning with the report’s call for standardised regional data sharing and joint analysis. Municipalities are cast as crucial integrators—merging GEMS, Pandora and ground stations for local alerts, compliance, and public guidance.
What’s holding things back
The report is frank about operational pain points:
Data engineering: satellite resampling/regionalisation, ground data cleansing/weighting, and harmonisation of spatial/temporal resolutions are non-trivial; oversampling and ML models need compute and skills.
Productisation: turning analyses into decision-support tools (e.g., map services, dashboards) requires reliability, accessibility, and clear uncertainty communication.
Access & documentation: several products remain partially open with limited pre/post-processing support and English-language documentation; this slows uptake beyond expert circles.
Recommendations — a practical action agenda
Standardise a national–regional monitoring fabric: reference-grade ground stations, low-cost sensor meshes, plus Pandora–GEMS nodes at strategic sites; publish a real-time portal with open APIs.
Expand Pandora coverage with priority to rural/underserved areas; institutionalise QA/QC (humidity control, tracker maintenance, flag handling) and routine co-location with regulatory PM₂.₅/NO₂/SO₂/O₃ monitors.
Operationalise GEMS workflows: automate ingestion, bias-correct with Pandora/AERONET and PCD networks, and stand up episodic tools (e.g., straw-burn plume tracking, local vs. transported diagnostics).
Strengthen capacity: regular training programmes for analysts and operators (ingestion, ML, WRF-coupled dispersion, nowcasting/forecasting); create a help-desk and shared code repositories.
Make products truly open & usable: English/Thai documentation, reproducible pipelines, and web map services; publish sensor specs/levels, and uncertainty metadata.
Policy integration: align outputs with PCD advisories, seasonal burn management, and enforcement; evaluate “25 Clean Air Measures” using EO indicators; prefer burn-area–based emission estimates to simple fire counts.
Cooperation & finance: keep ASEAN/SERVIR momentum; pursue blended support (technical assistance + equipment O&M budgets) to ensure continuous operations, not one-off pilots.
Key words
Remote sensing; GEMS; Pandora; AERONET/MPLNET; MODIS/VIIRS/TROPOMI; Himawari-8/9; PM₂.₅ estimation; biomass burning; long-range transport; SERVIR/GISTDA/PCD; open data; decision support.