Practical Use of Earth Observation Data for Official Statistics
Course Description:This intensive workshop is designed for the staff of National Statistical Institutes (NSIs) and scientific institutions working with spatial datasets. The training has been tailored to deliver the most crucial and practically applicable skills within a single day. Participants will acquire the tools necessary to download, process, and analyze both optical and radar satellite data using free, open-source GIS software (QGIS, SNAP) and cloud-based computing platforms (Google Earth Engine). The training script is highly practice-oriented, focusing primarily on real-world Use Cases.
Shortened Training Script:
Copernicus Ecosystem and Data Downloading
The workshop begins with setting up accounts in the Copernicus Data Space Ecosystem. Participants will explore the data browser interface, define their Area of Interest (AOI), and learn how to effectively search for and download relevant optical scenes from the Sentinel-2 satellite, as well as radar scenes from Sentinel-1.
Processing and Analysis of Optical Data in QGIS
The next module of the script focuses on Sentinel-2 optical data. It covers the visualization of images using RGB color composites with dedicated plugins, and the calculation of key vegetation indices (e.g., NDVI, SAVI) using the Raster Calculator. This section concludes with participants performing a simple image reclassification based on established NDVI thresholds, which allows for tasks such as the automatic identification of healthy versus stressed vegetation.
Cloud Processing – Google Earth Engine
The final part of the training introduces cutting-edge cloud technology. Participants will transition their analytical workflows to the Google Earth Engine (GEE) cloud. The script demonstrates how to instantly generate a cloudless image mosaic from a long time series and how to calculate radiometric indices on the fly over large geographical areas. The module finishes with a practical Machine Learning task – performing a supervised land cover classification using provided training points (utilizing algorithms like Support Vector Machine or Random Forest).
Course programme
Trainer
Artur Łączyński, Przemysław Slesiński, Tomasz Milewski
Location
Kraków, Poland
Date and time
10 July 2026; 8:30 a.m. – 2:30 p.m local Kraków time
Description
The course is dedicated to NSI staff, other national authorities and scientific institutions working with EO data. The basic goals are increasing knowledge on EO datasets, their characteristics, methodology of data processing (including machine learning), analysis and presenting.
Tool for practical session
Workshop in groups, GIS software (QGIS), internet access to the COPERNICUS data hub.
10 July 2026
8:30 – 10:00 a.m.
Introduction to Cloud Processing for Earth Observation
Available satellite data and their basic characteristics: Copernicus Program (SENTINEL 1-6), LANDSAT, MODIS, NOAA, commercial satellites, orthofotomaps:
Optical and radar sensors – data characteristics
EO classifiers build on optical data e.g. NDVI, NDWI, TCI
EO classifiers build on radar data e.g. scatter back, polarimetry
Scientific partners
Projects – ESA, EUROSTAT, COPERNICUS, UN SC, FAO
Confidentiality of EO data for statistical purposes
Reference data
Auxiliary data for EO data interpretation and validation
Administrative data – IACS, maps, CORINE, LUCAS, cadaster
In situ data – statistical collections, IACS
10:00 – 10:15 a.m.
Coffee break
10:15 – 12:15 a.m.
Use cases with demonstration using QGIS
Cloud removal e.g. preparing of composite 10 days images
Computation of NDVI • SAR data visualization
Agriculture - crop recognition, monitoring of plant growth, yield estimation
Land cover - recognition of land cover classes
Environmental statistics
12:15 - 1:15 p.m.
Lunch break
1:15 - 2:30 p.m.
Demonstration of processing and analyzing of EO data using Google Earth Engine or Copernicus Space data Ecosystem
Pre-processing of EO data – preparation of radar and optical data, ready to use data
Pixel and object classifications – which method to choose? – demonstration of examplesclassification methods – pixel classification,
Object classification (polygons/objects), classifiers in optical and radar data
Supervised classification algorithms – machine learning e.g. random forest, support vector machine, decision tree, ANN, maximum likehood
Fusion of radar and optical EO data
For more details on registrations and submissions for the 10th International Conference on Agricultural Statistics, please first login to your account.
If you do not have an account then you can create one below:
Make sure to register for this event to avoid missing out.
