Deep Learning for High-dimensional Parameter Retrieval

David Malmgren-Hansen

Research output: Chapter in Book/Report/Conference proceedingBook chapterResearchpeer-review


Modeling the state of the Earth in biology, ecology, climate, or geophysics, require describing it with a set of parameters which we can frame as bio-geophysical parameters. Retrieving these from satellites provides dense and frequent coverage and this task is an important area of Remote Sensing that supports the broader field of Earth Sciences. Parameter retrieval from satellite measurements is an area deep learning (DL) has started having a growing impact on, and this is likely to increase in the future. Despite this, several challenges need to be addressed for DL to solve and improve the current and future retrieval problems. This chapter outlines some of these challenges, which include the computational load of high-dimensional problems and defining the learning objectives for each problem at hand. Examples from two cases of bio-geophysical parameter retrieval which highlight these challenges are provided.

Original languageEnglish
Title of host publicationDeep Learning for the Earth Sciences: A Comprehensive Approach to Remote Sensing, Climate Science, and Geosciences
Publication date2021
ISBN (Print)9781119646143
Publication statusPublished - 2021


Dive into the research topics of 'Deep Learning for High-dimensional Parameter Retrieval'. Together they form a unique fingerprint.

Cite this