Land cover serial mapping

Land cover serial mapping

Jakarta, August 2019 – Following the Google Earth Engine training in July, the MapBiomas Indonesia network participated in an advanced training and workshop on land cover mapping.

Three instructors from the MapBiomas Brasil network conducted the three day training in Jakarta: Tasso Azevedo (General Coordinator of MapBiomas Global Network), Marcos Rosa (Technical Coordinator of MapBiomas Brasil), and Felipe Lenti (researcher and mapping expert from Amazon Environmental Research Institute, IPAM).

Like the previous training, participants included spatial analysts from HAKA Aceh, Hutan Kita Institute in South Sumatra, GENESIS Bengkulu, SAMPAN in West Kalimantan, Save our Borneo in Central Kalimantan, Green of Borneo in North Kalimantan, KOMIU in Central Sulawesi, Mnukwar in West Papua, JERAT Papua, WWI, and Yayasan Auriga Nusantara.

In the first stage of work, Indonesia was divided up into grids so that areas could be distributed to MapBiomas team members for collecting imagery to form mosaics. This mosaicking process is carried out by selecting Landsat satellite imagery taken by NASA satellites with the least cloud cover. Landsat imagery was chosen because it is publicly accessible and has been continuously available for the last few decades. 

In the second stage, participants sample the mosaics and determine land cover classification. The sample collection will be carried out for every year of in the range of 2000 to 2019, a twenty-year time span. The next step is to ensure the percentage and distribution of samples are adequate for each landcover class, as well as for each year of observation. 

During the workshop, participants collaboratively identified the 10 land cover classes for Collection 1: natural forest, mangroves, plantation forest, non-forest vegetation, palm oil, other agriculture, mining, non-vegetated area, aquaculture, other water bodies. 

MapBiomas Indonesia team members will now continue the mosaicking and sampling processes until they are complete.

An Initiative of:


Supported by: