报告人：Peter M. Atkinson（杰出教授）

时间：2023.9.13 14:00-15:30

地点：测绘馆206会议室

报告人简介

   Peter M. Atkinson is Distinguished Professor of Spatial Data Science at Lancaster University, Lancaster, UK where he is also Executive Dean of the Faculty of Science and Technology. He is currently Visiting Professor at the University of Southampton, Southampton, UK and at the Chinese Academy of Sciences, Beijing, China. Professor Atkinson has authored more than 350 peer-reviewed articles in international scientific journals, one book and over 50 refereed book chapters. He has also edited 11 journal special issues and eight books. He has held many large grants from a wide variety of funders and supervised over 60 PhD students. His research interests are highly interdisciplinary with a focus on remote sensing, geographical information science and spatial (and space-time) statistics applied to a range of environmental, epidemiological and socioeconomic problems. Professor Atkinson has an H-index on Google Scholar of 81 and an H-index on Web of Science of 60. He is a Fellow of the Learned Society of Wales. He was the 2020 Distinguished Lecturer of the International Association of Mathematical Geosciences (IAMG), Laureat of the Peter Burrough Medal 2016 of the International Spatial Accuracy Research Association (ISARA), was awarded the Belle van Zuylen Chair 2014 with Utrecht University, Utrecht, The Netherlands and was Visiting Fellow at Green-Templeton College, Oxford University in 2012-14. Professor Atkinson is Editor-in-Chief of Science of Remote Sensing, the open access sister journal of Remote Sensing of Environment.

报告简介

        Remote sensing is a tool commonly used for measuring the Earth’s surface with spaceborne and airborne sensors. The images that are produced are mostly not of direct interest, but provide information that is related to the subject of interest. Indeed, this is the basis of most remote sensing – that the properties that produced the recorded signal can be inferred inversely from the recorded signal. For scientific applications, the data that are produced through the inverse process (e.g., LAI estimates) can be then related to other properties that are the actual subject of interest, for example, plant biomass. Often regression-type models are used to make this link. In this talk, Peter will discuss some aspects of the measurement process and, in particular, the sampling strategy. This will lead to consideration of how we choose to represent these measurement elements (or not) and, consequently, what the data themselves may represent. Often these elements are not considered at all. This may not matter where the data being operated on define the ‘universe of interest’ (e.g., natural language processing), but they do matter for environmental data science.