Publications·November 30, 2024
This UNESCAP working paper takes a hard look at how Cambodia can use satellite and ground-based remote sensing to strengthen air-quality management. It compiles the current policy and institutional landscape (who does what), maps the monitoring network, summarizes what EO data are already used, documents the status of Pandora and GEMS, and lays out a practical set of recommendations to move from research pilots to operational, decision-ready products. The thread running through the paper is simple: Cambodia’s air-quality challenges are real and growing, but the country can close key knowledge and capacity gaps quickly by standardising data pipelines, building people and platforms, and embedding EO outputs in policy workflows.
Cambodia’s air-quality picture & institutions
Burden & trends. Annual mean population exposure to PM₂.₅ ≈ 22 µg/m³ (≈4.4× WHO guideline). In 2019, fine-particle exposure was linked to ~3,499 deaths (≈21 per 100,000), with stroke, ischaemic heart disease and COPD as leading causes. In Phnom Penh, the 2018 annual mean was ≈19.3 µg/m³, while the Jan–Aug 2024 mean rose to ≈25.1 µg/m³. Monthly averages for 2017–2024 show clear seasonality.
Who’s in charge. The Ministry of Environment (MoE) leads, via the Department of Air Quality and Noise Management (DAQNM); provincial departments implement and report up to MoE; local governments handle on-the-ground enforcement and public engagement.
Ground monitoring footprint. MoE reports 13 units in Phnom Penh, 43 units at 24 provincial offices, plus 3 mobile units—a useful base that still leaves large spatial gaps that satellites can help fill.
Policy architecture. Foundational laws (1996 Environmental Protection Law, Sub-decrees on air/noise (1999/2000), ODS management (2005), MSW management (2015), a 2020 circular on ambient-air measures, the 2023 Code on Environment & Natural Resources) culminate in the Clean Air Plan for Cambodia (CAPC, 2021). CAPC sets a 35% PM₂.₅ reduction by 2030, Euro IV vehicle norms (from 2022) and Euro V by 2027, a diesel-sulphur roadmap aligned with Euro V, and sectoral measures (industry, transport, residential, construction, open burning), alongside calls for inventories, source apportionment, modelling and stronger monitoring (including SLCPs). Cambodia also participates in EANET, the ASEAN Haze Agreement, and UNESCAP’s regional air-pollution programme.
The remote-sensing landscape: what’s in use
Satellites & platforms. Practitioners rely on MODIS/VIIRS (fires, AOD), TROPOMI (NO₂, SO₂, HCHO), OMI/GOME-2, and reanalyses like MERRA-2; geostationary assets—Himawari-8/9 (AHI) and GK2A—supply high-frequency imagery. Cambodia taps Global Forest Watch and NASA FIRMS for real-time fire alerts.
Ground-based remote sensing. AERONET/MPLNET in the region add vertical/aerosol-column context; Pandora brings trace-gas columns for validation of satellites (notably GEMS).
Use cases so far. Academic work (a small but growing corpus) covers burning trends, AOD-based PM₂.₅ estimates, and combined modelling–satellite analyses (e.g., WRF-Chem for the March-2019 burning episode that affected MSEA and beyond). EO is also used in SERVIR Southeast Asia tools and Cambodia’s Protected Area Alert System.
International support. KMA helped deploy 27 AWS and a GK2A receiving/analysis system; JMA’s Himawari program supports open data; SERVIR SEA (USAID/NASA/ADPC) is scaling regional AQ services and capacity.
Pandora & GEMS status in Cambodia
Pandora (PGN / Pandora Asia Network). A Phnom Penh Pandora station (PAN-215) became operational on 7 Aug 2024; early NO₂ column and surface-proxy time series (Aug–Oct 2024) show distinct peaks and seasonal structure. Current pain points are limited trained staff, lack of detailed SOPs/technical guides, hardware interruptions (mini-computer/ports), and insufficient analytic capacity to fully exploit the data—underscoring the need for QA/QC routines, spares, and training.
GEMS (geostationary). The paper situates GEMS as a high-cadence complement to LEO sensors and Pandora, with applications in hotspot detection, trend monitoring, local vs. transported attribution, and health-impact assessment. Near-term challenges are data integration (with Pandora and ground AQMN), platforming (storage, access, APIs), human capacity, and inter-ministerial coordination to ensure EO outputs feed planning, enforcement and public advisories.
Key knowledge gaps EO can close (and how)
Unmonitored areas. Construct monthly/annual PM₂.₅ maps using AOD (MODIS/MISR/SeaWiFS) + CTMs (e.g., GEOS-Chem) + bias-correction (e.g., GWR), validated with AQMN and Pandora.
Short-lived climate pollutants (SLCPs). Use TROPOMI/GEMS/Pandora to track NO₂, SO₂, HCHO, CO, O₃ and aerosol indices, linking to sectoral sources and chemistry for targeted mitigation.
Long-term trends & policy evaluation. Blend EO and ground data to judge CAPC progress; parse seasonal burning vs urban sources; produce province-level indicators for routine policy reviews.
Transboundary transport & airsheds. Pair GEMS/TROPOMI with trajectories (HYSPLIT/WRF) to map airsheds, quantify cross-border episodes, and inform ASEAN cooperation.
Health & equity. Fuse high-resolution PM/trace-gas fields with population and vulnerability indices to prioritise interventions for at-risk communities. MAIA-style composition insights will sharpen exposure–response work.
Agricultural fires. Move beyond simple fire counts to burn-area + FRP-based emissions, stratified by vegetation/fuel type, to improve inventories and evaluate burn-management measures.
What’s holding Cambodia back (today)
People & SOPs: Too few trained operators/analysts; missing standard operating procedures for instruments and data QA/QC.
Platforms: No unified data management & access layer (for GEMS, Pandora, AQMN, fires, AOD, models) with reproducible pipelines and open APIs.
Productisation: Analyses haven’t been consistently turned into operational decision tools (alerting, compliance support, episode forensics).
Coordination: Fragmented responsibilities limit inter-ministerial data flows and hamper regional sharing.
These themes recur across the Pandora/GEMS sections and the capacity-assessment framework.
Recommendations (actionable takeaways)
Update CAPC to explicitly mainstream EO: name GEMS/Pandora/AOD products in monitoring, compliance, and evaluation protocols; align indicators and review cycles.
Build a national EO-AQ platform: automate ingest, bias-correction, and fusion; publish near-real-time maps, archives, and APIs; include uncertainty metadata and bilingual documentation.
Harden operations: expand Pandora nodes; fund spares/O&M; co-locate Pandora with reference PM₂.₅/NO₂/SO₂/O₃; implement cal/val routines; standardise QA/QC and data flags.
Capacity building at scale: recurring training for operators/analysts (EO ingest, ML bias-correction, WRF/CTM, nowcasting/forecasting); create a helpdesk and shared code repositories.
Expand pollutant scope: go beyond PM₂.₅ & fires to operationalise NO₂, SO₂, O₃, HCHO mapping and source diagnostics.
Regional network: strengthen ASEAN/EANET/SERVIR collaboration for standardised data sharing, joint episodes analyses, and co-development of tools; use airshed evidence to target cross-border action.
Key words
Remote sensing; PM₂.₅; GEMS; Pandora (PGN); Himawari-8/9 (AHI); GK2A; MODIS/VIIRS/TROPOMI; fire radiative power (FRP); airshed & transboundary transport; SLCPs; capacity building; data platform; Clean Air Plan for Cambodia.