Working Paper Article Version 1 This version is not peer-reviewed

Knowledge and Data Driven Mapping of Environmental Status Indicators from Remote Sensing and VGI

Version 1 : Received: 27 December 2019 / Approved: 29 December 2019 / Online: 29 December 2019 (08:24:59 CET)

A peer-reviewed article of this Preprint also exists.

Goffi, A.; Bordogna, G.; Stroppiana, D.; Boschetti, M.; Brivio, P.A. Knowledge and Data-Driven Mapping of Environmental Status Indicators from Remote Sensing and VGI. Remote Sens. 2020, 12, 495. Goffi, A.; Bordogna, G.; Stroppiana, D.; Boschetti, M.; Brivio, P.A. Knowledge and Data-Driven Mapping of Environmental Status Indicators from Remote Sensing and VGI. Remote Sens. 2020, 12, 495.

Journal reference: Remote Sens. 2020, 12, 495
DOI: 10.3390/rs12030495

Abstract

The paper proposes a human explainable artificial intelligence approach for mapping the status of environmental phenomena from multisource geo data. It is both knowledge and data driven: it exploits remote sensing expert’s knowledge to define the contributing factors from which partial evidence of the environmental status can be computed. Furthermore, it aggregates the partial evidences to compute a map of the environmental status by adapting to a region of interest through a learning mechanism exploiting Volunteered Geographic Information (VGI), both from in situ observations and photointerpretation. The approach is capable to capture the specificities of local context as well as to cope with the subjectivity and incompleteness of expert’s knowledge. The proposal is exemplified to map the status of standing water areas (i.e. water bodies and river, human driven or natural hazard flooding) by considering satellite data and geotagged observations. Results of the validation experiments were performed in three areas of Northern Italy, characterized by distinct ecosystems. Results of the proposed methodological framework showed better performances than traditional approaches based on single spectral indexes thresholding. The use of expert’s knowledge, possibly imprecise/uncertain and incomplete, the need of few ground truth data for learning, and finally the explainability of learned rules are the distinguishing characteristics of the proposal with respect to traditional machine learning methods.

Subject Areas

soft constraints; Ordered Weighted Averaging Operators; Volunteered Geographic Information; standing water area mapping; decision attitude modeling

Comments (0)

We encourage comments and feedback from a broad range of readers. See criteria for comments and our diversity statement.

Leave a public comment
Send a private comment to the author(s)
Views 0
Downloads 0
Comments 0
Metrics 0


×
Alerts
Notify me about updates to this article or when a peer-reviewed version is published.