EO4Multihazards_CaseStudy4

The Science Case in the Caribbean region presents records on landslides, precipitation, maps used as inputs of hazard models and drone imagery over the region of interest. 
For the Carribean study-case, an analysis of open and proprietary satellite based dataset was used to facilitate the setup and evaluation of physically-based multi-hazard models. These allow for qualification and quantification of spatio-temporal multi-hazard patterns. These form a crucial input into the general hazard and risk assessment workflow. 

Presented here are the datasets employed for Case Study 4 in Deliverable D3.1 with a short description, produced and saved within the following folders:

Dominica_landslide: the landslides datasets mapped by ITC using high-resolution satellite imagery. It is intended to calibrate and validate the flood and landslide modelling. The folder contains four shapefiles:

·       Landslide_Part.shp - Shapefile containing landslide extent, flash flood extents, and their attributes.

·       Cloud.shp – Shapefile represents the cloud-filled areas in the satellite imagery where no mapping was possible.

·       The other two shapefiles are self-explanatory.

GPM_Maria: NASA Global Precipitation Mission (GPM) precipitation maps processed for model input in LISEM. GPM is a hybrid fusion with satellite datasets for precipitation estimates. Mean as input data to represent precipitation in the landslide and flood modelling. 

Maps_Models_Input : Soil and land use and channels, lots of custom work, SOILGRIDS, and SPOT image classification; all the datasets are ready for model input for OpenLISEM and LISEM Hazard or FastFlood. The dataset is meant to calibrate and validate the flood and landslide modelling.  

The raster files are either in Geotiff format or PCraster map format. Both can be opened by GIS systems such as GDAL or QGIS. The projection of each file is in UTM20N. 

Some key files are:

• dem.map -elevation model, the height of the landscape in meters above sea level.
•  lai.map - leaf area index, estimated using empirical relationships based on NDVI (Normalized Difference Vegetation Index)). The source data to calculate NDVI is Sentinel-2.
•   KSat.map - Saturated hydraulic conductivity of the soil, estimated based on a combination of SOILGRIDS soil texture, Saxton et al. (2006) Pedotransfer functions, and a national soil map for Dominica.
• clay.map - Clay texture fraction, SoilGrids resampling
• silt.map - Silt texture fraction, SoilGrids resampling
• sand.map - Sand texture fraction, SoilGrids resampling
• cover.map - Vegetation cover as a fraction, estimated using linear correlation with NDVI.
•   lu_new.map - Spot satellite image classification at 10 meters resolution for predominant land use types.
•   n.map - Mannings surface roughness coefficient, specific value based on the land use type. 
•  ndvi.map - Normalized Differential Vegetation Index, based on Sentinel-2 images in summer.
•  ldd.map - Drainage network map for the island, which can be used for flow accumulation and streamflow detection
• catchments.map - Catchment ID's based on the ldd.map drainage network.
• Channelldd.map - Channel-only drainage network map, calibrated manually to have all channels on the island represented correctly.
• Soildepth - Soil depth in meters, based on a physically-based soil depth model in meters and observational data obtained from landslide-sites during fieldwork in 2018.
• Slope.map - Slope map in gradient of the elevation  model (m/m) in the steepest direction

StakeholderQuestionnaire_Survey_ITC: The stakeholder questionnaires particularly relating to the tools developed partly by this project on rapid hazard modelling.  Stakeholder Engagement survey and Stakeholder Survey Results prepared and implemented by Sruthie Rajendran as part of her MSc Thesis Twin Framework For Decision Support In Flood Risk Management supervised by  Dr. M.N. Koeva (Mila) and Dr. B. van den Bout (Bastian) submitted in July 2024.

·Drone_Images_ 2024: Images captured using a DJI drone of part of the Study area in February 2024. The file comprises three different regions: Coulibistrie, Pichelin and Point Michel. The 3D models for Coulibistrie were generated from the nadir drone images using photogrammetric techniques employed by the software Pix4D. The image Coordinate System is WGS 84 (EGM 96 Geoid0), but the Output Coordinate System of the 3D model is WGS 84 / UTM zone 20N (EGM 96 Geoid). The other two folders contain only the drone images captured for that particular region's Pichelin and Point Michel. The dataset is used with other datasets to prepare and create the digital twin framework tailored for flood risk management in the study area.