HASARD-MM service specifications
The HASARD Multi Mission (HASARD-MM)_ service performs a mapping of standing waters from a co-pol backscatter image. It supports backscatter images from multiple missions. This service relies on the “HASARD”1 flood mapping algorithm that enables systematic, automatic and reliable SAR mapping of terrestrial water bodies. HASARD implements the Hierarchical Split-Based Approach (HSBA), described by Chini et al. 20172, and is specifically designed to support flooding-related disaster risk reduction at global scale. The underlying end-to-end image processing chain is based on a patented technology from the Luxembourg Institute of Science and Technology (LIST)3 that combines and fully integrates innovative hierarchical image splitting and open water backscatter modelling.
HASARD-MM exploits a single satellite image (Sigma Nought) to rapidly produce accurate standing water maps. The service employs Height Above the Nearest Drainage (HAND) auxiliary data to exclude false alarms in slope areas due to layover. HASARD-MM generates a water extent map in both raster and vector format.
The HASARD-MM processing service applies the workflow below:
Input of the HASARD-MM service is a single band Sigma0 asset in dB from a Calibrated Dataset [CD] and the single band asset from an HAND Auxiliary Dataset. Thus to employ the service you need the following two Datasets:
An event SAR Calibrated Dataset including a co-pol Sigma Nought asset in dB (e.g. s0_db_x_hh for TerraSAR-X) acquired during crisis,
HAND Auxiliary Dataset including the
handsingle band asset providing relative elevations in m.
HASARD-MM requires as input the HAND auxiliary dataset. The hand single band asset must cover at least the area of interest defined within the footprint of the input sigma nought asset.
This service does not support SAR complex data (e.g. Sentinel-1 SLC).
The HASARD-MM service requires a specified number of mandatory and optional parameters. All service parameters are listed in the below Table 1.
|Radar calibrated dataset||Input SAR Calibrated Dataset including an event Sigma0 single band asset providing backscatter values in db||YES|
|Sigma nought asset name||Name of the co-pol Sigma0 single band asset providing backscatter values in db (e.g. s0_db_x_hh for a Calibrated Dataset in X-Band)||YES||
|HAND auxiliary dataset||Input Auxiliary Dataset including the hand single band asset providing relative elevations in meters||YES|
|HAND asset name||Name of the HAND single band asset providing relative elevations in meters||YES||
|Area of Interest||Area of interest expressed in WKT||NO|
|Starting tiling level||Hierarchical level of splitting. When the level is equal to
|Check bimodal distribution (Ashman D)||Ashman D coefficient to be used to check bimodal distribution||YES||2.4|
|Minimum number of pixels per tile||Minimum number of pixels per tile for checking the bimodality hypothesis||YES||1000|
|Number of pixels for small object removal||Number of pixels for small objects (blobs) removal||YES||20|
|Threshold for HAND asset||Relative elevation threshold to be considered when mapping water extent. Pixel classified as water that have height above the nearest drainage above this threshold in meters (e.g. 15m) are masked out.||YES||15|
|Backscatter threshold||Backscatter threshold to be considered when mapping water extent. The threshold must be given as a signed decimal in decibel (e.g. -20 db).||YES||-20|
Table 1 - Service parameters for the HASARD-MM processor.
Input SAR calibrated dataset and asset name
First two mandatory parameters are dedicated to specify the input calibrated dataset and define which single band backscatter asset from this dataset shall be used for the water mapping computation. The system does not automatically select the co-pol asset from the given Calibrated Dataset to be used in the computation.
The drag and drop of the single-band asset (e.g. "s0_db_c_vv") is not possible. Users must drag and drop first a Calibrated Dataset (e.g. "[CD] SENTINEL-1A C-SAR IW VV/VH 134 2021/11/06 21:43:07") into the
Radar calibrated dataset parameter and then specify the STAC key (name of the asset) in the
Sigma nought asset name one.
Input single band asset specified into the
Sigma nought asset name parameter must be a co-pol one. As an example from a full pol calibrated dataset in C-band you must insert either
s0_db_c_hh as asset name.
To quickly find the names of all single band assets in a Calibrated Dataset, click on the
Show assets button available near the feature title. A list with all single-band assets (CBNs) included within the Calibrated Dataset will appear under the feature title.
All CBNs available in the ESA Charter Mapper can be found here.
Input HAND auxiliary dataset and asset name
With the third
HAND auxiliary dataset and fourth mandatory parameters, ‘HAND auxiliary dataset
andHAND asset name
respectively, the user must specify the input auxiliary dataset and specify the name of the single band asset to be used for the water mapping computation. The name of the asset in the forth parameterHAND asset name
is pre-filled by default and the user must just drag and drop the auxiliary dataset in the third parameterHAND auxiliary dataset`.
The drag and drop of the single-band asset (e.g.
hand) is not possible. Users must drag and drop the Auxiliary Dataset (e.g. "[CD] MERIT Hydro HAND - Hand") into the
HAND auxiliary dataset parameter and then check that the STAC key (name of the asset) in the
HAND asset name is set as
Specifications of the
hand single band asset can be found here.
The fifth parameter (optional) can be used to define the area of interest expressed as a Well-Known Text value.
The input area of interest must have an area higher than 1600 skm or 160000 ha. If a smaller area is inserted the job will not be executed and a warning will appear.
An AOI higher than 160000 ha can be easily drawn in the ESA Charter Mapper by using the Spatial Filter button available in the Navigation and Search Toolbar in the upper left side of the map. During the drawing of the rectangular area, a popup showing the size of the current area in ha will appear in the map next to the cursor.
In the definition of “Area of interest as Well Known Text” it is possible to apply as AOI the drawn polygon defined with the area filter. To do so, click on the button in the left side of the "Area of interest expressed as Well-known text" box and select the option AOI from the list. The platform will automatically fill the parameter value with the rectangular bounding box taken from the current search area in WKT format.
The following four mandatory parameters:
Starting tiling level,
Check bimodal distribution (Ashman D),
Minimum number of pixels per tile,
Number of pixels for small object removal represents the parameters of the retrieval algorithm. Default values given in Table 1 are the ones recommended for a general use.
In the tenth mandatory parameter the user must specify a relative elevation threshold to be considered when mapping water extent. Pixels classified as water that have values of Height Above the Nearest Drainage (HAND) above this threshold in meters (e.g. 15m) are masked out. In particular, this threshold is used to derive a binary mask of potential floodplains from the HAND dataset which is then applied for masking the water extent.
To better define your HAND threshold, select under the Results tab in the left panel the HAND auxiliary dataset, open the Details of this auxiliary dataset, select the
hand single band asset, and inspect its histogram. In the map you will see by default the HAND dataset with a grayscale color bar with the min max of the entire dataset. Under Expression insert the following
and click on the underneath Apply button to visualize in the map the binary mask of potential floodplains applied in HASARD-CD. If you are not happy with the result, change the value of 15 in the expression with another one and click again on the Apply button. The map will be immediately refreshed with the updated binary map. Repeat this binarization on the fly of the HAND asset as much as you need. Once you are satisfied with the binary mask displayed in the map, you can replace the default value of 15 in the
HAND threshold parameter with the one you identified.
In the last mandatory parameter the user must insert a threshold value in decibel (e.g. -20 db) to be used for the computation of water extent from the backscatter signal.
To better define your backscatter threshold, select under the Results tab in the left panel the Radar Calibrated dataset that you defined as input of HASARD-CD (e.g. [CD] TSX-1 EEC SM HH 86 2012-09-26 17:41:16), open the Details of this calibrated dataset, select the
s0_db_B_PP (where PP is the polarization [HH, HV, VH, VV] and B the SAR-band [X,C,L]). Inspect backscatter value using the Get Point Value function after a right click on the map. Employ the Get Point Value function over potential water pixel having low backscatter values. Define the suitable threshold by choosing a value in dB around the average of the left lobe values from the Sigma Nought histogram distribution. More information can be found in Chini et al., 20172.
The HASARD-MM processor provides as output the following products:
Water extent overview
Water extent single band
Water extent polygon vector
HASARD-MM Product Specifications can be found in the below tables.
|Attribute||Value / description|
|Long Name||Water Extente RGB composite|
|Description||Water bitmask in the blue channel including transparency|
|Data Type||UnSigned 8-bit Integer|
|Projection||UTM or EPSG:4326 - WGS84|
|Valid Range||[1 - 255]|
|Attribute||Value / description|
|Long Name||Water mask from HASARD-MM|
|Description||Water extent bitmask: 1=Water, 0 =Not-water|
|Projection||UTM or EPSG:4326 - WGS84|
|Valid Range||[0 - 1]|
|Attribute||Value / description|
|Description||Polygon vector converted from HASARD-MM water mask|
|Geospatial Data Type||Vector|
|Vector Data Type||Polygon|
|Feature properties||DN values of source raster|
M. Chini, R. Hostache, L. Giustarini and P. Matgen (2017), "A Hierarchical Split-Based Approach for Parametric Thresholding of SAR Images: Flood Inundation as a Test Case," in IEEE Transactions on Geoscience and Remote Sensing, vol. 55, no. 12, pp. 6975-6988, Dec. 2017, DOI: 10.1109/TGRS.2017.2737664. ↩↩