Methane is the second most important greenhouse gas after carbon dioxide. Although the atmosphere contains significantly less methane than carbon dioxide (1.9 ppm compared to 421 ppm), it is still responsible for around a third of current global warming. With his doctoral thesis, OHB employee Enno Tiemann is contributing to making anthropogenic methane emissions easier to detect and quantify from space.
The main natural sources of methane are swamps and wetlands. However, their emissions are only responsible for a good thirty per cent of the methane in the atmosphere. The rest is caused by human activities. The largest anthropogenic sources include agriculture (especially livestock farming and rice cultivation), landfill sites and the energy sector.
As methane only remains in the atmosphere for a comparatively short period of around 10 years, reducing methane emissions can quickly achieve measurable effects in terms of global warming.
A first step in this direction is the reduction of methane emissions in the energy sector. Methane is released into the atmosphere primarily during the extraction of natural gas and crude oil through controlled authorised leaks (venting or flaring) and through pipeline leaks or malfunctions. Therefore, the EU has recently released stricter regulations for monitoring and reporting of methane emissions.
Earth observation satellites can detect methane plumes
Satellite data can help the energy sector to meet the specified requirements. Satellites have a 24/7 view of the Earth and can detect pipeline leaks even in remote and hard-to-reach areas. Nevertheless, detecting methane sources is anything but easy and the amount of escaping gas can often only be roughly estimated. One of the reasons for this is that there are currently only a few satellites in space that have been specifically developed to detect methane. Although various Earth observation satellites can detect methane, the spatial and spectral resolution as well as the recording frequency of the images are often not sufficient to answer the crucial question: Are there any methane leaks in human-made structures? And if so, where are they and how much gas is escaping?
This is precisely the problem that Enno Tiemann is addressing in his doctoral thesis, which is being funded by the ESA Φ-lab and supervised by OHB Digital Connect in collaboration with the Technical University of Munich. OHB Digital Connect is an expert in data processing and image analysis within the OHB Group and heads the Competence Hub for AI and Big Data. The aim of the work done for the thesis is to use machine learning methods to combine time series data from various active satellites – primarily Copernicus Sentinel-2 and Landsat 8/9 – with information on wind direction and speed in order to recognise anthropogenic methane emissions earlier and quantify them more accurately.
Development of algorithms for the automation of methane detection
"Currently, methane detection usually relies on experts manually checking Earth observation data. This involves using the extent of a methane cloud to determine the emission rate – usually with a high degree of uncertainty," explains Enno Tiemann. "The methodology is often based on a cascading approach from coarse to finer resolutions or AI is used to detect methane in images of individual satellites. However, both approaches still require human analysis. We therefore want to develop algorithms that fuse data from different sources from the outset to improve automated detection and reduce uncertainties in emission rates."
The project will build fundamental knowledge in the use of satellite imagery in conjunction with weather data. Combining data from different sources using data fusion and machine learning will allow scientists to gain new insights from existing satellites. This means that a wide range of new applications can be developed – including in the areas of environmental and infrastructure monitoring, urban planning and security.
