Projects / Programmes
High-resolution drought monitoring based on satellite and ground data
| Code |
Science |
Field |
Subfield |
| 6.12.00 |
Humanities |
Geography |
|
| Code |
Science |
Field |
| T181 |
Technological sciences |
Remote sensing |
| Code |
Science |
Field |
| 5.07 |
Social Sciences |
Social and economic geography |
remote sensing, drought, data time-series, satellite data, machine learning
Researchers (9)
| no. |
Code |
Name and surname |
Research area |
Role |
Period |
No. of publicationsNo. of publications |
| 1. |
53350 |
PhD Nejc Čož |
Geodesy |
Researcher |
2020 - 2022 |
23 |
| 2. |
38086 |
Aleš Grlj |
Geography |
Researcher |
2019 - 2022 |
36 |
| 3. |
33600 |
PhD Urška Kanjir |
Geodesy |
Researcher |
2018 - 2022 |
90 |
| 4. |
25640 |
PhD Žiga Kokalj |
Geography |
Head |
2018 - 2022 |
377 |
| 5. |
25040 |
Peter Pehani |
|
Technical associate |
2018 - 2022 |
100 |
| 6. |
36950 |
Maja Somrak |
Computer science and informatics |
Researcher |
2018 - 2022 |
29 |
| 7. |
50575 |
PhD Liza Stančič |
Geography |
Junior researcher |
2018 - 2022 |
37 |
| 8. |
20005 |
PhD Tatjana Veljanovski |
Geodesy |
Researcher |
2018 - 2022 |
154 |
| 9. |
52249 |
Petra Vovk |
|
Technical associate |
2019 - 2020 |
5 |
Organisations (1)
Abstract
Drought occurrences are getting more frequent; Slovenia was hit by three catastrophic events and two regional ones in the last decade alone. Droughts have serious economic, social and environmental impacts and affects not only agriculture, but also the energy sector, tourism, water transport and supply of drinking water. The proposed project therefore focuses on developing an innovative model for high-resolution determination of agricultural drought on a national level, using freely available satellite data, in-situ meteorological and hydrological observations, and all available ancillary data on past drought events, crop types, and ground characteristics. The model will be based on deep neural networks.
Slovenia is an excellent test area for developing such a model because its diverse topography, a mix of climatic conditions, fragmented land use, and a range of land management practices reduce the accuracy of existing drought detection models. The country also has very open data policies regarding numerous meteorological and hydrological data, which can therefore be used in the deep learning model. The results can be applied on a national scale, which can promote their use in integral drought management.
We have the following objectives:
to determine and analyse the most suitable and stable drought-related indices based on optical satellite data;
to develop a method based on deep neural networks for merging satellite data with ground measurements for an accurate (overall producer’s accuracy of more than 90%), detection of drought on a local level (individual graphical units of agricultural land use); and
to design a prototype system for an operational comprehensive characterization of drought in agricultural areas at a national level.
We will achieve this by examining, evaluating, and enhancing the satellite data processing techniques; data fusion of high and low-resolution imagery (downscaling) to produce long-term data series; data fusion of imagery and in-situ data; applying time series analyses and deep learning to find statistically relevant variables and the affected areas. In a special work package, we shall define and evaluate the whole procedure in terms of its usability for operational use.
There will be several advancements when compared to already established drought detection models, which mostly operate in the sphere of low and medium spatial resolution. We will tap high-resolution data from the newest generation of satellites – Sentinels of the European Commission’s Copernicus programme. Adaptation of algorithms for learning with deep neural networks is not common in remote sensing and has not been used previously for such detection of drought.
I
f the project is successful, the timely and local level intelligence (extent, duration, severity) on agricultural drought will benefit the work of disaster relief and disaster management personnel, scientists studying the impacts of drought on local ecosystems, natural resource managers, policy makers, and individual farmers.
Significance for science
The project will provide a prototype solution that will enable detailed drought observation and mitigation actions in Slovenia on a local and regional scale. However, we believe that the methodology will be applicable to all similar areas of the world (large landscape diversity, moderate latitudes) with minor modifications. For dissimilar areas, the range of required modifications is expected to be larger.
The project will provide support and expert advancement for the “network” mechanism of the Drought Management Centre for South-eastern Europe, hosted by the Slovenian Environment Agency. In a broader scope, the project will be able to support the United Nations Convention to Combat Desertification with long-term integrated strategies that focus simultaneously, in affected areas, on improved fertility of land, and the rehabilitation, conservation and sustainable management of land and water resources, leading to improved living conditions, in particular at the community level.
The originality of the expected results is in:
better spatial resolution of results – full exploitation of high-resolution satellite data for drought detection at a local level,
switch to newly available sensors both due to better resolution as well as for continuity,
full integration of in-situ data – fusion of freely available automatic local ground measurements with freely available satellite imagery on a national scale,
use of deep learning techniques.
The Sentinel-2 data has a much better resolution that is nowadays used in established continental and (inter)national drought monitoring services therefore using high-resolution data and merging them with long-term archives of available low-resolution data will propose new dimensionality of such services and establish a foundation for new standards in the field. The results will also enable (semi)automatic determination of the boundaries of the drought and will provide a greater consistency of data on drought and its quantitative characterization.
Significance for the country
The project will provide a prototype solution that will enable detailed drought observation and mitigation actions in Slovenia on a local and regional scale. However, we believe that the methodology will be applicable to all similar areas of the world (large landscape diversity, moderate latitudes) with minor modifications. For dissimilar areas, the range of required modifications is expected to be larger.
The project will provide support and expert advancement for the “network” mechanism of the Drought Management Centre for South-eastern Europe, hosted by the Slovenian Environment Agency. In a broader scope, the project will be able to support the United Nations Convention to Combat Desertification with long-term integrated strategies that focus simultaneously, in affected areas, on improved fertility of land, and the rehabilitation, conservation and sustainable management of land and water resources, leading to improved living conditions, in particular at the community level.
The originality of the expected results is in:
better spatial resolution of results – full exploitation of high-resolution satellite data for drought detection at a local level,
switch to newly available sensors both due to better resolution as well as for continuity,
full integration of in-situ data – fusion of freely available automatic local ground measurements with freely available satellite imagery on a national scale,
use of deep learning techniques.
The Sentinel-2 data has a much better resolution that is nowadays used in established continental and (inter)national drought monitoring services therefore using high-resolution data and merging them with long-term archives of available low-resolution data will propose new dimensionality of such services and establish a foundation for new standards in the field. The results will also enable (semi)automatic determination of the boundaries of the drought and will provide a greater consistency of data on drought and its quantitative characterization.
Most important scientific results
Interim report
Most important socioeconomically and culturally relevant results
Interim report
