You are here

The technology of in-depth training for satellite monitoring of terrestrial issues

Work number - M 39 AWARDED

Presented by the Institute of Space Research of the NAS of Ukraine and SSA of Ukraine

Authors: Kolotii Andrii, Lavreniuk Mykola, Yaylimov Bohdan

The purpose of the work is to research and improve methods for identifying Essential Variables (indicators) that can be used for satellite environmental monitoring and assessment of Sustainable Development indicators (in particular land degradation) for the territory of Ukraine, using products based on the developed deep learning technology.

The authors, based on theoretical studies and experiments, have identified an approach based on the methods of deep learning for land cover mapping and used them to assess the indicators for achieving of sustainable development goals.

The scientific basis for creation of land cover maps for the whole territory of Ukraine for several decades (1990, 2000, 2010, 2016-2018) with a high spatial resolution of 10-30 m was created.

The basic conceptual framework for assessing the achievement of Sustainable Development Goals for the territory of Ukraine with use of satellite data and machine learning techniques has been formed.

The technology of deep learning for land cover mapping and its use for the estimation of indicators of achievement of the Sustainable Development Goals is proposed.

The technology of deep learning for the construction of land cover maps is created.

Comparison with world counterparts. The analysis of existing methods and models land cover mapping was carried out. Advantages of the proposed method are confirmed on test independent data.

Deployment: Ukrainian Hydrometeorological Center, GEO International Committee in the context of GEOSS Global Earth Observation Systems, AG-07-03: Global Agricultural Monitoring, JECAM, UN-SPIDER, ERA-PLANET Project of the European program HORIZON 2020, UN Convention on Combating Desertification.

Economic effect of deployment - assessment of the state of land cover, detection of signs of degradation processes, detection of crop rotation violation. Such information will be useful for assessing the value of land resources for land tenants and may be used as the basis for independent land evaluation in case of a land market launch. Use of such automated information technology allows to save money on the data collection from the population and enterprises, it allows small traders to evaluate the situation of a competitor with the production of a particular agricultural crop.

Number of publications: 60, including 1 monograph, 59 articles (39 - in foreign journals). According to the Scopus database, the total number of references to authors' publications, presented in the work, is 267; according to the Google Scholar database, the total number of links is 957. Under this topic, 1 PhD thesis was defended