Using Earth Observation data for measuring economic indicators

Aakash Gupta (~aakash70)


Policymakers require swift feedback for measuring the impact of their policies. The delay in feedback can have major recuperation to the lowest segment in society. This is where high-frequency economic indicators derived from satellite imagery, can be a life-saver

1# Talk outline (30-min) breakdown


We discuss the various methods with which economic activity can be measured and also discuss the pros & cons of this approach. We will show how we can use freely available Sentinel imagery to highlight areas of interest. We can then access high-res images to gather more detail information. Infra-red and radar sensors can provide information which is hidden from optical imagery.


We will examine some common methods to identify economic activity and identify areas of high growth vs poorer areas.

  1. Using VHR (30cm resolution) imagery to assess changes in car distribution

  2. SAR (Radar) for activity monitoring of building construction

  3. Measuring quality of water in major rivers in Europe

  4. Impact on agri-production in the USA

  5. Monitoring airport activity by counting aero-planes parked in airports

  6. Measuring pollution over major industrial belts

  7. Measuring deforestation using vegetation indices


Questions from the audience


Basics of Python & Pandas

Interest in Geo-science and working with satellite imagery

Video URL:

Content URLs:

Draft slides:

An older version of the talk at PyCon Ng 2021

Speaker Info:

Aakash considers himself to be a life-long learner. Over the last 12+years he have helped companies build their Data Science teams from scratch. He has a rich experience in designing innovation strategies for new-age startups. Along the way he has also taken part & won multiple hackathons. If you would like to connect with him, drop a message on his LinkedIn profile.

Speaker Links:

LinkedIn Profile

Medium Blog

Section: Scientific Computing
Type: Talks
Target Audience: Intermediate
Last Updated: