The Team behind ML4 Earth
Dear visitor of this webpage! My name is XiaoxiangZhu and I am holding the Chair for Data Science in Earth observation (SiPEO) at the Technical University of Munich (TUM). My research focus on AI and data science in Earth observation. Specifically, I develop innovative signal processing and machine learning methods, and big data analytics solutions to extract highly accurate large-scale geo-information from big Earth observation data. My lab aims at tackling societal grand challenges, e.g., global urbanization, UN’s SDGs and climate change, thus, works on solutions that can scale up for global applications. I have enjoyed a lot to (co-)supervise more than 60 PhD students to date and to initiate or lead more than 30 research projects. I also serve in the scientific advisory board in several research organizations, among others the German Research Center for Geosciences (GFZ) and Potsdam Institute for Climate Impact Research (PIK).
I am convinced by the impact of our methodological research to provide society with the data, knowledge, and tools to mitigate the unpleasant effects of the changing climate, or to adapt with the best-possible knowledge that AI models can provide based on global EO data. I am very happy that we are in the position to focus our research in these areas with ML4Earth and to thus add significant value to society. Within ML4Earth, together with our extremely capable team of PhD students and senior staff, we develop novel methods in EO Data Science and directly apply these methods where they are most needed in climate and environmental research.
My name is Andrés and I am an Artificial Intelligence researcher with 10+ years of academic
and professional experience, passionate about solving real-world problems.
Besides my strong technical background, I am enjoying a lot my role as team leader in the
multicultural environment of our research Lab.
+ N.N. (Community Building)
My name is Marcus Langejahn and I coordinate the ML4Earth project at TUM as a science manager. I have a background in observational astrophysics where I obtained my PhD at the University of Wuerzburg in 2022 and gained experience with coordinating the DFG research unit FOR5195. For ML4Earth I aim to connect our community of researchers and students while facilitating a friendly and open working environment. Also, I am interested in the field of science communication, trying to convey the amazing scientific results of today’s research to the public
My name is Jie Zhao. I received my Ph.D. degree in remote sensing between Technische Universität Wien, Vienna, Austria and Luxembourg Institute of Science and Technology, Esch-Sur-Alzette, Luxembourg. Currently, I am a Postdoc focusing on large-scale flood mapping using earth observation data at the department of Data Science in Earth observation, Technical University of Munich. My research interests broadly lie in flood mapping, deep learning, and SAR image analysis. During my Ph.D. studies, I worked on several projects with various partners. With such an academic background, I’d like to contribute to community development in ML4Earth.
I received the Dipl.-Inf.(FH) degree in Computer Science at the University of Applied Science Schmalkalden with the specialization of intelligent information systems in 2010. From 2010-2014 I worked in energy forecasting (on the producer side) in the industry. From 2015-2019, I completed my Ph.D at the Technical University of Munich in the field of Energy Informatics, Energy Information Retrieval and Non-Intrusive Appliance Recognition. After the Ph.D I worked in computer vision for quality control in the industry for one year and started in 2020, a postdoctoral research position at the AI4EO Future Lab, a collaborative project with DLR at the Chair for Data Science in Earth observation of the Technical University of Munich. The current research fields include AI driven population density estimation, mining site detection. My research interests include
● Classical Machine Learning & Representation Learning
● Computer Vision
● Music Information Retrieval
● Time Series Prediction and Event Detection
● Strong AI
My name is ZhitongXiong. I received my B.E. degree in Software Engineering, M.S. degree in Computer Science and Technology, and a Ph.D. degree in Computer Science and Technology from Northwestern Polytechnical University, Xi’an, China. Currently, I am a Postdoc working on geometry-aware remote sensing scene understanding at the department of Data Science in Earth observation, Technical University of Munich.
Geometric and semantic information about cities can be used for urban planning and disaster monitoring, which closely relate to the lives of residents living in cities. Aiming at the automatical perception of semantic and 3D information from image data, my research interests broadly lie in computer vision, deep learning, and remote sensing image analysis.
More specifically, I have research experience in many computer vision tasks, such as object detection, semantic segmentation, scene recognition, and multi-modal feature learning. Currently, I am also working on label-efficient learning to reduce the annotation efforts for training large-scale deep neural networks. Furthermore, I have extensive hands-on experience in deep learning frameworks, intelligent transportation systems, and remote sensing image analysis. For remote sensing tasks, I mainly focus on the automatic understanding of 3D cities using deep learning technologies.
Since I have both a research background in computer vision and remote sensing, I hope I can better combine these two communities to promote a variety of practical applications such as autonomous driving, smart cities, disaster forecasting, and so forth.
I am a Ph.D. student at the Technical University of Munich and part of the ML4Earth project. My path began with an interest in software engineering in my undergraduate studies in computer science, which I pursued as a software engineer for a year after graduation. During this time, my interest in artificial intelligence was sparked and I continued as a master’s student in this direction. Now, during my Ph.D., my passion for machine learning is ongoing and I am applying it to the field of remote sensing.
More specifically, my research focuses on explainable AI in remote sensing, where I try to open up the black-box behaviour of many machine learning methods and make them understandable to humans. Currently, I am working on the impacts of extreme climate events.
Hi everybody, my name is Konrad Heidler. I’m a PhD student at the chair for Data Science in Earth observation at TU Munich. I am interested in applying deep learning and computer vision for the remote sensing of phenomena in polar regions. Some examples for the things I try to capture with my algorithms are glaciers, permafrost, or sea ice.
As part of the ML4Earth project, I am exploring reasoning mehods for tracking the developments in permafrost areas. As Arctic permafrost is one of the earth systems most affected by global climate change, we need to closely monitor these areas in order to detect trends and possible feedback loops. The goal of the project is to incorporate both spatial and temporal cues in algorithms for detecting permafrost disturbances, as these are generally hard to detect without such cues.
Before joining the lab, I obtained a Bachelor’s degree in Mathematics at TUM in 2017 and a Master’s degree in Mathematics in Data Science in 2020. Driven by the desire to apply my skills to real-world problems, I decided to do my research in the field of remote sensing.
I am a PhD student at the chair of Data Science in Earth observation at TU Munich working on uncertainty quantification of sea level rise. Previously, I obtained my masters degree from the University of Amsterdam. Generally, my interests lie in Bayesian Deep Learning, uncertainty quantification and their application to Earth Observational data. I also enjoy contributing to open source software.
I am a Ph.D. student and is a member of ML4Earth. I come from China. In 2018, I attained a bachelor’s degree after graduating from the University of Electronic Science and Technology of China. Later I took my master’s degree in the ESPACE double degree program and received a master’s degree from both Technical University of Munich and Wuhan University. Currently, I am working on machine learning for complex structures, including the application of graph neural networks on climates (e.g., tipping points and critical transition). My research interest includes domain adaption, graph neural networks, spatial-temporal analysis, and remote sensing, etc.
Hi there! I am a PhD student at the Chair of Data Science for Earth Observation at TU Munich and I am working on physics-aware machine learning for the earth’s hydrology. Before joining the chair I obtained my master’s degree in physics from LMU Munich where I worked on predicting efficient materials for solar hydrogen generation using machine learning.
I enjoy working on the intersection of domain sciences and methodologies that can help us to better understand and care for our planet.
Hey there! This is Xiangyu Zhao. I am a first year Ph.D. Candidate in ML4EO supervised by Prof. Dr.-Ing. habil. Xiaoxiang Zhu. Prior to my PhD, I got my master degree at TU Munich and bachelor degree at Tongji University. I am greatly passionate about Artificial Intelligence and especially interested in machine learning algorithms. Using AI to explore the earth is a challenging and promising research direction. Based on the abundant data acquired from different sources, we are looking forward to extracting information for Earth observation. Especially, my goal is to develop computer vision algorithms with only sparse labels. My research interests lie in label efficient learning and transfer learning.
The Partners of ML4Earth
I am primarily working on observing the land surface with an imaging spectrometer on board of Earth Observation satellites. For this project, we want to match specific soil reflectance properties derived from satellite imageries with physical soil samples from the ground. The most interesting research question is about how many soil samples are needed to characterize an area of interest and how machine learning techniques can help to overcome the scarcity of soil samples.
I am a professor of glaciology and Earth Observation at the University of Bristol and the Technical University Munich and Director of the Bristol Glaciology Centre and own a degree in physics and a PhD in geophysics. I specialise in the analysis of airborne & satellite data sets from the polar regions, and in combining these data with models of the Earth system. I have been focusing my research on the ice sheets covering Antarctica and Greenland and their contribution to sea level and published extensively in the general field of geodesy, sea level variations in time and space and measuring mass exchange between the land and oceans due to melting of land ice and the hydrological cycle.
Recently I became interested in capturing the uncertainties in projections of future climate change and associated impacts, especially related to sea level rise. This is particularly challenging because of limited understanding of key physical processes and a short observational record. It is, however, time-critical information needed by policy makers now and consequently requires novel and imaginative approaches to make progress.
I’m interested in utilizing methods from satellite geodesy for Earth system research. Our group in Bonn works on understanding global and regional sea level changes, on mapping 4D thermospheric neutral density and its relation to various drivers, and on assessing long-term and rapid changes in the terrestrial water cycle. In these fields, we develop new retrieval methods and generate data products, confront them with model simulations, and generally seek to generate new inference by combining complementary data sets. The recurring theme is that in these applications, geodetic space techniques provide new and unexplored information which is, however, literally buried in a sea of superimposed signals.
That’s why I’m interested in turning to machine learning methods. In our use case project, we will address precursors of drought, one of the most direct consequences of climate change. We seek to combine satellite-gravimetric data (from the GRACE and GRACE-FO satellites), physical knowledge (from simulation models that represent the terrestrial water cycle), and ML approaches in order to arrive at improved seasonal forecasts of water availability and droughts at the European scale.
I am currently a postdoctoral researcher at the Institute of Geodesy and Geoinformation (Astronomical, Physical and Mathematical Geodesy), University of Bonn, Germany. I studied at the Information Engineering University (China, 2013-2016), Wuhan University (China, 2016-2020), and the University of Bonn (Germany, Guest student, 2018-2019). I completed my PhD thesis in Geodesy and Geomatics Engineering (2020) at Wuhan University. Since April 2021, I have been working as a postdoc at the Institute of Geodesy and Geoinformation (APMG), University of Bonn.
My research focuses on the application of machine learning techniques to Geodetic and Hydrological studies. I developed a machine learning based approach for reconstructing long-term global terrestrial water storage change (TWSC) and for forecasting near-future (up to one year) freshwater availability from the empirical relationships between several climatic/hydrological variables, such as precipitation or temperature, and satellite observations of land water storages. My research interests also include land surface modelling, investigating climate extremes (drought/flood events), and integration of satellite observations into hydrological models (WaterGAP Global Hydrology Model, WGHM) using an ensemble Kalman filter (EnKF).
I am a member of the WIKI project (Reanalysis/forecast of water storage in Europe via AI). In this project, my job is to assimilate the GRACE and Sentinel-1 satellite data into the Community Land Model (CLM) model by combining deep learning and EnKF aims to generate a high resolution (12 km; 1d) water compartment data set over Europe for both historical period (2003-present) and the near future (up to 12 months lead).
I’m a full professor at Leipzig University. My main scholarly interests are on understanding ecosystem responses to climate extremes as well as the human environment nexus during extremes: I have also been working on understanding macroecological dynamics and ecosystem functioning. My research is based on the latest data-driven research methods. The backbone of our work is the joint exploitation of high-dimensional Earth observations.
My aim is always to combine our empirical findings with theoretical and conceptual understanding to understand complex interactions in the Earth system through Explainable AI. This can help to reach a deep understanding of extreme events through knowledge based systems.
I am the head of the Permafrost Research Section at AWI and professor for Permafrost in the Earth System at the University of Potsdam since 2016. Since 2021, I am also Deputy Head of the AWI Geosciences Division. I have a background in Geology and have been conducting permafrost research in the Arctic since 1999. After finishing my PhD at University of Potsdam and AWI in 2005, I conducted postdoc work at the Geophysical Institute of the University of Alaska Fairbanks and became research faculty there with a focus on remote sensing of landscape dynamics, hydrology, and carbon cycling in Arctic permafrost regions. I returned to Germany with an ERC Starting Grant in 2013.
While my large team in the AWI Permafrost Research Section works on many different aspects of Arctic permafrost characteristics and dynamics, I focus on remote sensing observations of permafrost landscape dynamics across broad spatial and temporal scales. I have a strong interest in quantifying climate change impacts in high-latitude terrestrial environments and their complex feedbacks. I strive to support the next generation of permafrost and remote sensing researchers, and I am involved in multiple international permafrost-related networks and research projects, some of which are using Machine Learning methods to better understand rapid changes in the Arctic.
I’m a postdoctoral researcher at the university of Leipzig. With the background of applicable mathematics and artificial intelligence I want to introduce new methods into the research field of earth system science. Especially the combination of high dimensional data and modern machine learning tools is a challenging topic I focus my research on.
Explainable AI is one of the important keys when modern approaches meet data-driven research. It is important to understand the physical models in complex systems to gain detailed comprehension of the results made by machine learning applications. In particular cutting edge deep learning methods can benefit from this approach.
Hi, I am a PostDoctoral reasearcher in the Permafrost Research Sectio at the Alfred Wegener Institute for Polar and Marine Research. I have a background in Geography and Remote Sensing with a strong focus on technical applications. I obtained a BSc in Physical Geopgraphy, MSc in Geoinformation and Visualization. After two years as a research assistant in Ireland I started a PhD at the Alfred Wegener Institute in Potsdam, which I finished in 2018. Since then I worked as a PostDoc in various national and international collaborative research projects.
My primary focus in on quantifying, understanding and prediticing rapid landscape dynamics and disturbances, particularly in terrestrial arctic environments. For this purpose I am applying machine- and deep-learning techniques on spatial and earth observation data of various spatial and temporal scales. I am extremely keen on using latest technology to tackle environmental- and data analysis related challenges.