Improving Forest Classification of the Brazilian National Forest Inventory using the CEO Platform

Co-authors: Thiago Spagnolo, Valeria Contessa, and Pedro Vivar

The Brazilian Forest Service (Serviço Florestal Brasileiro-SFB) in collaboration with the Food and Agriculture Organization of the United Nations (FAO) recently held a hybrid workshop to support the Brazilian National Forest Inventory (IFN-BR) in its effort of forest classification. IFN-BR is collecting field data across all of Brazil. The workshop collected data in six biomes across Brazil.

The six biomes across Brazil studied in this workshop.
The six biomes across Brazil studied in this workshop.

Participants, including national experts, collected data using Collect Earth Online (CEO) to support this effort. Technical experts supported by SERVIR-Amazonia provided support as needed before and during the workshop.

The workshop was focused on improving planning, results analysis, and quality control to reduce the cost of fieldwork and ensure the field team’s safety.

A hybrid workshop approach allowed participants to join remotely while national experts and organizers gathered in a central location near Brasília (Capital of Brazil). This group, composed of 6 experts with great experience in identifying specific types of vegetation in each biome, together with the Brazilian National Forest Inventory team with field experience, allowed the participants to get proper training in order to collect data with accuracy. Approximately 80 people participated online in the data collection efforts.

🌄 A diverse ecosystem

Across the six biomes being studied there are a wide range of forest types classified by IFN-BR. They include Floresta Tipica (Rainforest), Cerrado, Caatinga arborea, Restinga arborea, Palmeiras (Palm forest), Bambus (Bamboo forest), Campinarana, and Mangues (Mangrove).

Participants in the workshop used CEO to classify different points. The first decision for participants was whether an area was a forest or not. Those that were considered forests were further split into natural forests and planted forests. 

A decision tree showing how participants classified different points in CEO.
A decision tree showing how participants classified different points in CEO.

Using CEO, the workshop participants were able to collect land use and land cover data using remote sensing in both forested and non-forested areas that are inaccessible because they lack road infrastructure.

Participants were also able to classify other land uses including water, urban areas, or agricultural fields. By using remote sensing, IFN-BR data collection was made more cost-effective without the need to invest in field data collection. 

🌟Samples collected to improve the national forest inventory

The aim of the workshop was to collect as many samples as possible (out of a total of 22,000 samples) distributed all over Brazil in a regular grid of 20 x 20 kilometers. Following a day of training in the use of the CEO platform, the remote team of experts enabled the collection of approximately 1500 data samples in four days.

The workshop also facilitated new collaborations between participants. A new network of collaborators for the national forest inventory has been created. Productive debates occurred between specialists and participants (who can be contacted in the future for additional data collection) about the peculiarities of Brazilian vegetation types which generated new knowledge on how to interpret and label these using remote sensing.

In support of CEO’s user community, the participants are now aware of the potential of the CEO platform as a tool for collaborative forest data collection. Further, technical experts from SERVIR also participated in the workshop to understand the needs and processes of workshop participants in order to inform the future development of CEO and training resources.

🛰Using CEO to inform field data collection

Using CEO in conjunction with field data collection allowed for important improvements to planned field sampling. 

First, the team will be able to reduce field data collection costs. Field data collection is a difficult and costly endeavor. 

Teams of field data collectors must first travel to the chosen sampling location. This is often in the middle of a forest, sometimes in areas with no nearby road infrastructure. Then, field teams need to collect data, which may take many hours, before traveling to the next sample site. 

With CEO, the team was able to collect a lot of data about their sample sites ahead of time. 

Using three concentric circles in CEO, the team collected important information about the area surrounding the sampling location. The radii for these circles was 150 m, 2 km, and 10 km from the proposed field sample location. 

At these scales, the team collected information on how the sample area could be accessed (e.g. road, river, landing strip), impediments to collecting samples (e.g. topography, rivers), and the presence of homes or villages near the samples.  

This data will help the team plan their field data collection to be more efficient–a huge benefit.

(L) The three concentric circles used in the CEO projects to collect information about the surrounding area. (R ) The 150 m radius circle and the field plot square locations.
(L) The three concentric circles used in the CEO projects to collect information about the surrounding area. (R ) The 150 m radius circle and the field plot square locations.

🚧Using CEO and Planet imagery to improve safety

In addition to cost reductions, the team will be able to use data collected in CEO to improve the safety of field team members who, in the past, came up face-to-face with people practicing  illegal gold mining and illegal timber harvesting. These encounters can quickly turn dangerous, threatening the safety of team members.

This year during the workshop, participants used CEO and the Planet data to determine if there were recent disturbances indicative of illegal gold mining and illegal timber harvesting. This crucial information will be used to avoid dangerous situations for field teams.

Near-real time satellite imagery provided by Planet with financial support from NICFI was crucial to this effort. 

🤝CEO’s flexible shapefile plot and sample design

The IFN-BR data collection effort uses unique plot and sample designs. CEO’s flexible plot and sample design allowed for successful data collection. 

An example field sample location with the 150 m circle from the project in CEO.
An example field sample location with the 150 m circle from the project in CEO.

CEO’s shapefile feature allows users to create any shape and configuration for their plot and sample design. Users can create shapefiles using QGIS, ArcGIS, or another program. These files are then uploaded into CEO’s easy to use project wizard interface.

To support IFN-BR’s data collection effort, the team was able to create plot and sample designs reflecting their specific needs. The square shaped samples they used are field units used to collect field data. These include tree measurements, species identification, soil samples and  botanic sample collection.

🌎Using Geo-Dash to support data collection

The IFN-BR data collection also leveraged CEO’s Geo-Dash interface. 

The Geo-Dash widgets used to support data collection.
The Geo-Dash widgets used to support data collection.

CEO’s Geo-Dash interface allows project organizers to create multiple imagery widgets. These widgets provide additional imagery and information for data collectors to use while classifying different points. Here, imagery from the Sentinel-2 and Landsat satellites were used for the last 20 years, to allow users to assess the changes in land use over the plots. 

CEO’s ability to copy Geo-Dash widgets and layouts from other projects was much appreciated by the team, as it greatly reduced the time required to set up projects for the workshop.

 

CEO would like to thank its ongoing funders FAO, NASA, USAID, SERVIR hubs, and SilvaCarbon, a US government program. Thanks also to CEO’s technology partners: Norway’s International Climate & Forests Initiative for funding open high-resolution data availability; Planet for providing high-resolution imagery; and the Google Earth Engine team for creating a platform for Earth science data and analysis.

Collect Earth Online is working constantly to improve the user experience, and your feedback is invaluable. If you have ideas to share, please write to support@collect.earth.

Thank you!

Search Our Blog

Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors