Sentinel-1 data as operational tool for Flood Monitoring in Transboundary River Basins Theodora Perrou, PhD Candidate Harokopio University of Athens (HUA), Dept. of Geography, Athens 70, El. Venizelou str., Athens, GR-17671, email: [email protected]

Abstract Methods Discussion Flood disaster is one of the heaviest disasters in the The methodology for flood mapping/ monitoring of the The significant contribution of the study in Kerkini-lake world. It is necessary to monitor and evaluate the flood affected areas is based on a change detection and is the identification of the causes inducing the flood disaster in order to mitigate the consequences. As thresholding procedure of radar satellite images. Binary phenomenon in the downstream of the Strymon IRB floods do not recognize borders, transboundary flood water/non-water and RGB multitemporal products were using archive data and creating retrospective products monitoring and management is imperative in shared generated and interpreted. based on Sentinel-1 free data. In combination with river basins. Flood risk management is highly Kerkini lake-reservoir regulated in Strymon river basin Transboundary river basin hydro-meteorological and topography data support 15 Level-1 GRD SAR scenes, in Descending Interferometric Wide 21 Level-1 GRD SAR scenes, in Descending Interferometric Wide swath (IW) mode, with polarization VV and VH were collected swath (IW) mode, with polarization VV and VH and relative orbit 102 dependent on early information and requires data from spanning the period from October 2014 to October 2015 were collected spanning the period from October 2014 to May 2015 the hypothesis that flood event was provoked mainly

Reference Reference the whole river basin. The utility of Synthetic Aperture Flood images Flood images image image by the contribution of the increased water amount (Slave) (Slave) (Master) (Master) Radar (SAR) images for flood mapping, was Apply orbit Apply orbit flow in the lake coming from the upstream part of file file demonstrated by previous studies but the SAR systems Subset Calibration Strymon river catchment. The increase of water level

Calibration processing -

Subset processing

in orbit were not characterized by high operational - in the lake from upstream coincides with the start of

Pre Pre Speckle filtering Coregistration DEM- capability. Copernicus system will fill this gap in (Gamma map 5x5) Assisted XCorr the flood event in the downstream. The eventually Coregistration Terrain correction operational service for risk management. The DEM-Assisted Range Doppler inappropriate management of water release through operational capabilities have been significantly Binarizarion Binarizarion Image the dam are the main reasons why the flood event RGB composition of (Band math, (Band math, difference 3 processed images threshold)

threshold) Processing improved by newly available satellite constellation, Processing could be attributed to. Sentinel-1 imagery provided Water/ non water Change detection Water/ non water Change such as the Sentinel-1A&B mission, which is able to image image detection excellent spatiotemporal dynamic flood hazard maps Terrain correction Terrain correction provide systematic acquisitions with a very high Range Doppler Range Doppler Flood extent map for the case of Evros IRB, offering the possibility of Binary flood Multitemporal temporal resolution in a wide swath coverage. The extent product RGB product processing the entire transboundary area due to the present study deals with the monitoring of Figure 4. Methodology flowchart Figure 5. Methodology flowchart Case study: Kerkini lake-reservoir in Strymon IRB Case study: Evros IRB wide spatial coverage of each image. The flood transboundary flood events occurred in Strymon and phenomenon occurred almost simultaneously in the Evros river basins using free Sentinel-1 data and ESA’s Greek and Turkish territory. The affected areas extend open source SNAP software. Results north of the city of () and south to the river's delta. There is no evidence for a specific Kerkini lake-reservoir regulated in Strymon RB (October 2014 – October 2015) spatiotemporal pattern of flood at least during this Introduction event. Floods are probably the most frequent and disastrous natural hazards of the world. Mapping their extension is fundamental to assess the damages1. Flood Conclusions monitoring and management is complicated enough in . The Copernicus Sentinel-1 mission can be river basins controlled by a single authority, and considered as a unique operational tool for flood becomes even more challenging when dealing with monitoring. 2 transboundary floods . Satellite Earth Observation (EO) . SAR's inherent capability to observe during cloud are a unique source of synoptic information at global cover and Sentinel-1's frequent revisits makes it scale that can supply regular, detailed updates on the ideal for flood monitoring. It can be used to 3 status of hazards . The most common approaches assess the extent of flooded areas and the impact making use of EO for flood mapping are based on SAR on human, economic and environmental loss. data. The potential of the data provided by SAR Figure 6. Binary images water/non water from October 2014 –October 2015 . Sentinel-1 repeat cycle of six days, and its systems for large-scale flood detection has been Chart 1. precipitation diagram Chart 2. Water level measurements of Kerkini lake (mm) (mm) 160 300 systematic acquisition of dual-pol SAR data, demonstrated by several previous investigations. The 140 250 120 100 200 provides an unprecedented chance to develop 80 150 synoptic view and the capability to operate during 60 40 100 20 50 automatic, high frequency flood mapping daytime and nighttime and in almost all weather 0 0 algorithms for complex environments. conditions, contrary to visible/infrared sensors, are the Serres Station Kerkini Station Achladochori Station key features that make SAR images useful for . NRT flood monitoring could support authorities inundation mapping4. In addition, the great sensitivity especially during emergency response phase. to water of the microwave band permits SAR to . In conclusion, the case of flood monitoring in a distinguish between land and water. Near real-time transboundary river basin became critical and an (NRT) flood monitoring can be considered as a unique essential requirement to monitor the overall operational tool in the hand of authorities. spatial catchment area. The needs revealed by this research are overall satisfied by Sentinel-1 mission in terms of revisit, wide coverage and systematic acquisition of free available data and

Objective Figure 7. RGB Multitemporal SAR images open source software.

The objective of the present study is to create the Transboundary Evros river basin (October 2014 – May 2015) “migration story” of the flooded areas on the basis of Chart 3. Accumulated monthly precipitation the evolution in time for flood events occurred in 180,00 Acknowledgements 160,00 Turkey Strymon and Evros International River Basins IRB 140,00 120,00 100,00 Author would like to thank the ESA RSS team for 80,00

rainfall (mm) 60,00 40,00 supporting the processing work of this study by 20,00 0,00 providing a high performance Virtual Machine. Oct Nov Dec Jan Feb Mar Apr May 14 14 14 15 15 15 15 15

Alexandroupolis station (47m) station (35m) Orestiada station (22m) Major References Chart 4. Water level measurements [1] Pulvirenti, L., Pierdicca, N., Chini, M.., Guerriero, L., ‘An algorithm for operational 8,00 7,00 flood mapping from Synthetic Aperture Radar (SAR) data using fuzzy logic’, Nat. 6,00 Hazards Earth Syst. Sci. 11(2), 529–540 (2011). 5,00 4,00 [2] Bakker, M. H. N., ‘Transboundary River Floods and Institutional Capacity’, JAWRA J. Figure 2. Strymon IRB 3,00 Am. Water Resour. Assoc. 45(3), 553–566 (2009). 2,00

Water level Waterlevel (m) 1,00 [3] Ivan Petiteville, Stephen Ward, George Dyke, M. S. & J. H., ‘Satellite Earth 0,00 Observations in support of disaster risk reduction, Special 2015 WCDRR Edition’

(2015).

2-Apr-15 6-Apr-15

1-Dec-14 5-Dec-14 9-Dec-14

5-Nov-14

27-Oct-14

15-Apr-15

13-Dec-14 18-Dec-14 23-Dec-14

17-Mar-15 21-Mar-15 27-Mar-15 11-Mar-15 [4] Smith, L. C., ‘Satellite remote sensing of river inundation area, stage, and Pythio station Petalon station discharge: a review’, Hydrol. Process. 11(10), 1427–1439 (1997). Figure 1. Location map Figure 3. Evros IRB Figure 8. Flood extent maps in Evros IRB

7th ADVANCED TRAINING COURSE ON LAND REMOTE SENSING 4–9 September 2017 | Szent István University | Gödöllő Hungary