How an app evaluates the climate data of different land areas

At the Climate Hackathon, researchers and developers worked on solutions for reducing and assessing carbon footprints by remotely classifying the landscape for the Cool Farm Alliance Challenge [1]. BFH data expert Namrata Gurung reports on the hackathon and her work.

The Climate hackathon was organised and hosted by Stratiteq in partnership with Microsoft was held between 22-26 March, in order to obtain innovative solutions to build a sustainable world.

Our team that won the 2nd prize at the hackathon, was a mix of motivated data-scientists and fullstack developers, namely, Dr. Matteo Jucker Riva, Dr. Namrata Gurung, Dr. Barry Sunderland, Dr. Leticia Fernandez Moguel, Jonas Henrikkson, Oded Winberger and Markus Giger. The LandPro (Land Productivity Monitoring) App ideated by Dr. Matteo Jucker [2] was further developed by the team during the hackathon to provide quick evaluation of any vegetated area by examining the spatial and temporal variability of vegetation by combining image segmentation with near real-time satellite data.

LandPro (Land Productivity Monitoring) App

LandPro App is a machine-learning based web-interface that harnesses the power of earth-observation satellites and translates the obtained data into relevant information for farmers, land managers and more. LandPro App’s workflow includes three steps (see also Fig. 1):

  1. The user indicates the area of Interest through a simple web-interface. We retrieve relevant data from Earth Observation satellite image repositories, including multi-spectral images (SENTINEL-2 satellites), NDVI (Normalized Difference Vegetation Index) time series, and Soil organic carbon (SOC) estimation.
  2. A deep-learning model for semantic segmentation divides the area into polygons with homogeneous land cover.
  3. For each polygon, statistics on CO2 sequestration [3,4] is produced: (a) Vegetation CO2 absorption based on the NDVI is computed as the integral of the previous year (b) Soil organic carbon (SOC) stock which is a vital component of soil with important effects on the functioning of terrestrial ecosystems is estimated using the SOC data obtained from Joint Research Centre European Soil Data Centre (ESDAC) [5].

All the above data is then displayed to the user via a simple and clear map-based interface.

Fig. 1: LandPro App’s three-step process: 1. The user selects area via phone or computer using satellite RGB images as reference or using spatial vector files. 2. The algorithm discriminates between different land cover/vegetation types using semantic segmentation. 3. Infrared & radar data are used to classify the land productivity over time and space.

LandPro App Users

The market for LandPro App is quite wide-range: from farmers and land managers who want to assess or demonstrate their CO2 sequestration to companies in the carbon-credit marketplace and ESG financial sector (Environmental, Social, and Governance) who want to streamline and improve their carbon-credit certification process. Local administration, institutions and nature conservation organizations seeking to understand the carbon sequestration potential of their respective land areas and measure the impact of their interventions could also potentially find this app quite useful.

Outlook for the LandPro App

Further improvements on what we developed will be discussed in this section. Within the scientific research community, the topic of CO2 sequestration and usage of real-time high resolution satellite imaging is a hot topic, and thus we expect methods and models to greatly improve in the near future, which will allow us to obtain better estimates of CO2 absorption/emissions that can be calculated combining landcover type with multi-spectral images. More interactivity such as modifying polygons boundaries, adding spatial elements, describing land management practices for each of the polygons can be further integrated within the user interface. Visualization of CO2 sequestration as a function of time can be mapped and its comparison with major changes in the recent past could also be added. Furthermore, SOC estimates could be updated real-time based on SENTINEL images and latest sampling data. Additionally, comparison with neighbouring areas can be performed i.e., information not only on the area of interest entered by the user, but also for the similar neighbouring areas can be provided. This could enable the user to obtain estimates not only of their own performance, but also suggest locations where better practices can be found.

The LandPro App was very well-received by the community. CarbonSink [6] company confirmed their interest in using this app to evaluate potential of carbon sequestration in their project design phase. European Space Agency’s plan on a high 10m2 resolution global land cover map [7] starting next summer, will enable in making our app a lot less complex, faster to develop and easy to use. We are working on a clickable MVP (Minimum Viable Product) for a limited area in the world. We also have been selected by StartUpWiseGuys [8] to participate to their Startup accelerator program.

We very much welcome further collaborations, and/or suggestions of ideas. Feel free to reach out to us and check out our GitHub Repo: You can also try out the interactive app we built at the hackathon here:


  2. Riva, Matteo Jucker, et al. “Assessment of land degradation in Mediterranean forests and grazing lands using a landscape unit approach and the normalized difference vegetation index.” Applied geography 86 (2017): 8-21.
  3. Castaldi, Fabio, et al. “Evaluating the capability of the Sentinel 2 data for soil organic carbon prediction in croplands.” ISPRS Journal of Photogrammetry and Remote Sensing 147 (2019): 267-282.
  4. Issa, Salem, et al. “A Review of Terrestrial Carbon Assessment Methods Using Geo-Spatial Technologies with Emphasis on Arid Lands.” Remote Sensing 12.12 (2020): 2008.
  5. Lugato, Emanuele, et al. “A new baseline of organic carbon stock in European agricultural soils using a modelling approach.” Global change biology 20.1 (2014): 313-326.

AUTOR/AUTORIN: Namrata Gurung

Dr. Namrata Gurung is a Data-Scientist, currently working as a PostDoc at BFH, IDAS on Diversifier-NLP project with Witty Works GmbH that aims to create inclusive language for job-ads. Her passion lies in bridging tech and social impact to create innovative sustainable solutions. She's always interested in creative ideas, and for collaborations can be connected via Linkedin.

PDF erstellen

Related Posts

None found

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *