Open-Earth Monitor Science Webinars - 11th edition

• Damiano Oldoni

Open Earth Monitor OEMC 2024 webinar
Image by wallpaperflare
On this page

General

Both talks are focussing on research ongoing at the International Institute for Applied Systems Analysis (IIASA).

Exploring In-Situ Data: Definitions, Sources, and Training Applications

Presenter: Steffen Fritz

In-situ data are very important for completing and testing EO satellite data and the software built on such data.

Some interesting links to in-situ data:

Geo-Wiki toolbox is a collection of desktop/mobile applications FotoQuest Go: mobile app for citizen scientists to report land use and land cover. Possibility to earn (a little of) money during the campaigns (2015 and 2018). CropObserve: crop type, phenological stage, damage, management activities; all data is open Picture Pile: Rapid image assessment. For example, do you see deforestation in the picture, Crowd2Train Project: Rapid image assessment again, but more about labelling.

Crowdsourcing land cover and land use reference data using the Geo-Wiki application

Presenter: Linda See

What is the importance of in-situ data? Providing a ground truth for land cover maps. Examples: there are huge disagreements among land cover maps created by software applied to satellite data. And so, the idea of Geo-Wiki was born. The importance of this disagreement analysis is to get the places where in-situ data collection is important. Early campaigns were designed to collect data to solve some questions like “can we validate a map of land availability for biomass and adjust the global estimate?””. This campaign (2011-2012) added 33k data points above the 20k used by the original land cover map.

This talk is strongly based on a letter published as open access: Lessons learned in developing reference data sets with the contribution of citizens: the Geo-Wiki experience.

To motivate people there were prizes, e.g. authorship in papers, Amazon vouchers. These were thought to motivate people to join for long periods like 6 months.

Examples:

  • Global Cropland Mapping, data in doi.PANGAEA.de 873912, 80 participants
  • GLobal Field Size Campaign, 130 participants, 4 week,s 390 K obs, 110k unique locations validated at least 5 times each, pure.iiasa.ac.at/id/eprint.15526
  • Human Impact on Forests Campaigns, 130 participants, 7 weeks for 4 campaigns, 459k obs, 110k unique locations validated at least 5 times each. More info here: Lesiv et al. 2022, citizen scientist co-authors, in Scientific Data
  • GLobal Built-up Surface Campaign: 61 participants, 7 days, 276k obs, 50k unique locations validated at least 5 times each. Data published in IIASA-PURE repository: dare.iiasa.ac.at/112/. Paper published in Scientific Data.
  • Drivers of Tropical Forest Loss: 58 participants, 2 weeks, 400k obs, 120k unique locations validated at least 3 times each. Data published in IIASA-PURE repo: pure.iiasa.

Additional Tools in Geo-Wiki:

  • links to Google Earth Links to Google Earth Engine applications
  • NDVI time series tool
  • Links to Sentinel Hun and display of imagery time series from Sentinel-2 and Landsat: it helps by giving context
  • Feedback tool
  • Branches for quality checking of visual interpretation
  • Ability to create your own validation sample migrated to LACO-Wiki

Ongoing work:

  • Validation of changes in CORINE (LAMASUS project)
  • Drivers of forest/biomass change (EYE-CLIMA project)
  • Updates to forest management map
  • New version of Geo-Wiki