This project aims to create a map of pollution in Nan region, in the north of Thailand, where the agricultural fires during the burning season (January-April) generate high PM2.5 concentrations.

The first step is to calibrate the sensors and set them up all throughout the study zone. We have a total of 9 AirGradient sensors : 6 OpenAir which require electricity and wifi and 3 OpenAir Max which are autonomous and work on a solar panel. To get a high-resolution map, 9 sensors are not sufficient so we train a random forest algorithm to predict PM2.5 concentrations at a more precise scale, using fire hotspots, weather and topographic data that we get respectively from NASA, local weather station and Copernicus APIs.

A map created by Agathe Chauvel and Elisa Skowronek is updated every hour on a dashboard.

a presentation by Elisa Skowronek

Observatories, such as Saen Thonh social-ecological observatory, play a critical role in advancing ecosystem-based innovations and locally adapted solutions, which are necessary to improve the prevention of disease transmission at the interface between human, animal, and ecosystem health.

see the publication

Since 2008, a series of collaborative projects with local communities and administrations

A dataset  (dataverse IRD) holds the results of different Random Forest classifications using the combination of Sentinel-2 Optical and Sentinel-1 Radar images. The different images were acquired in 2019. The dataset covers Saen Thong sub-district, Nan province in northern Thailand, a mountainous area with a monsoon climate. The classifications cover three categories of land: 1. uncultivated steep mountain slopes with forest (Park, protected), 2. cultivated mostly steep slopes with annual crops (e.g. upland rice, maize), or tree plantations (e.g rubber, teak, bamboo), or community forest, 3. cultivated flatland with paddy fields, besides most urbanization and modern infrastructure is also located.

Mahuzier, C.; Morand, S.; Chaisiri, K.; De Rouw, A.; Soulileuth, B.; Thinphovong, C.; Tran, A.; Valentin, C., 2022, "Random Forest land cover classifications of Sentinel satellite images in 2019, Saen Thong, Thailand", https://doi.org/10.23708/GENR6J, DataSuds, V2 

A dataset contains three runoff and erosion vector maps classified by value ranges: i) a map of runoff coefficient (Krc %) predicted from local surface conditions and runoff data from 535 1- m² plots in Southeast Asia; (ii) a map of mean soil loss (kgm-2) predicted from the same database; and (iii) a map of gully index, which is based on catchment area, runoff coefficient and LS, a topographic factor. These runoff and erosion maps were produced as part of the ANR FutureHealthSEA project: predictive scenarios of health in Southeast Asia, linking land use and climate change to infectious diseases.

Mahuzier, C.; De Rouw, A.; Morand, S.; Valentin, C., 2023, "Runoff and erosion maps in Saen Thong sub-district (Tha Wang Pha district, Nan province, Thailand)", https://doi.org/10.23708/2RJYVV, DataSuds, V1