3rd BEACON Business Case

The 3rd BEACON Business Case represents theGreek Case conducted together with KARAVIAS Underwriting Agency from Athens, Greece. The Case is focused on the insurance against flood.

Cotton crops flash-flood damage – validation of the BEACON Damage Assessment Calculator component’s accuracy

KARAVIAS Underwriting Agency, Greece

KARAVIAS Underwriting Agency is founded in 2014. As Coverholder at Lloyd’s, the company has the capacity and experience to provide insurance solutions in the agricultural sector. Crop producers and agribusiness entities around the world are realizing the value of forward - thinking insurance providers who can provide innovative products, financial strength, and superior risk management. KARAVIAS Underwriting Agency is able to provide risk assessment, risk underwriting, policy issuance and claims management. The Agency is already active in all Insurance sectors and is very well recognized among a sales network of almost 2.000 brokers and insurance consultants, seeking to expand its activity through innovative services and pioneering solutions based on new technologies and experienced executives.

  • Region covered: Thessaly, lowland area located in Central Greece.
  • Predominant crops cultivated: winter cereals, as well as summer crops, such as cotton and maize.
  • Total agriculturala arable land: Thessaly plain is the second largest plain of Greece occupying an area of approximately 14,000 km2, or 10.6% of the Greek territory. Out of this percentage, cultivated land covers 36.1%, the rest being distributed to forests, rangeland and other uses.
  • Data included: Flood damaged and non-damaged cotton crop parcels. Damage levels ranged from 40 to 100%.
  • Damaging Event: Due to extensive rainfall caused by the Mediterranean hurricane (Medicane) “Ianos”, many areas of Thessaly region were flooded. The flash flood event took place on the 18th of September at 21:30 local time, just before the cotton harvest season. Extensive damages were reported in agricultural land, urban areas of Farsala, Mouzaki, and Karditsa cities, and on the road network of the wider area.
  • Insured parcels during the summer growing season: 2019-2020.
  • Individual parcel records included information on: field location, field size, risk occurred and date, damage percentage, expected yield.

For flood damage assessment, BEACON employs optical and C-Band SAR satellite data for delineating and mapping a flooded area, for determining the beginning and the duration of the flood, as well as for damage assessment. Natural hazards included in this case are:

  • heavy rains,
  • long periods of rain,
  • snowmelt flash-floodsand floods that could be induced when river overflow their banks and water inundates the surrounding agricultural land.

Optical satellite data were acquired and processed for flood mapping and detection. The first available Sentinel-2 image after the event was captured on the 20th of September with a 25% cloud coverage, partly located over the flooded area. Next optical images were acquired on September 25th and 30th.

  • Index calculation: The modified Normalized Difference Water Index (mNDWI) was calculated.
  • Index thresholding: A land-water threshold of 0.09 was attributed to mNDWI for flood delineation.

Radar satellite data were also acquired and processed for flood detection, delineation and duration estimation.

  • Flooded reference image stack: Five post-event images were used; on September 21st, 22nd, 27th, 28th and October 3rd, since after this date the flooding water had totally withdrawn.
  • Non-flooded reference image stack: This image stack dated two months before the event and included 26 S1A and S1B images. Additionally, for every SAR image in the process, a multi-temporal speckling filter was implemented which required a number of images of the same area in order to achieve noise removal. This number was defined to be 15 image acquisitions before the image in filtering, which means that for the first image of the reference stack (July 18th) 15 previous images were utilized with the first one dating back to the 6th of June.
  • Statistical Analysis of SAR time-series: Statistical analysis of the backscattering Sigma naught (VV polarization) of each pixel is then performed in both multi-temporal image stacks. For each pixel the minimum, maximum and mean Sigma naught is derived.
  • Index calculation: Two indices are calculated, the Normalized Difference Flood Index (NDFI) for highlighting flooded areas and the Normalized Difference Flood in Vegetated areas Index (NDFVI) for highlighting shallow water in short vegetation.
  • Index thresholding: A threshold of 0.70 for NDFI and 0.75 for NDFVI is applied to extract flooded areas.

Machine Learning approach

The methodology for floods that is developed and incorporated in BEACON was tested and evaluated against Copernicus Emergency Management Service (CEMS) in terms of optical and radar EO products. The high overall accuracy reached 94% (Kappa 0.584) revealed very good agreement for the flood delineation automated process considering that CEMS product is created based on a semi-automated approach that also involves visual interpretation in order to draw the flood extent border.

The flooded area in the municipality of Karditsa was detected and delineated following the workflow implemented in BEACON. A number of 13 registered cotton parcels in the BEACON platform during pilot implementation were flooded and damaged.  A great 90% of the total pilot parcels area was unaffected by the flood, while the duration of flooding was estimated up to 6 days.

Flood Duration and extend between the 20th of September and 3rd of October

  •  
  •  
  •  
  •  
  •  
  •  

Leave a Reply

Your email address will not be published. Required fields are marked *

WHAT YOU CAN READ NEXT

1st BEACON Business Case: Agroseguro, Spain

2nd BEACON Business Case Triglav, Serbia

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 821964.

BEACON Boosting Agricultural Insurance based on Earth Observation data. © 2019 All Rights reserved. Privacy and Terms

TOP