Hazard class should be "none" when there is no hazard

GFDRR / thinkhazard

ThinkHazard!

http://thinkhazard.org/

GNU General Public License v3.0

32 stars 19 forks source link

Hazard class should be "none" when there is no hazard #854

Closed matamadio closed 5 years ago

matamadio commented 5 years ago

For some hazards, "nodata" is interpreted as "very low" hazard.

Some examples: Coastal flood classified as "very low" where there is no coast (e.g. Switzerland, landlocked regions) Similarly, cyclones are classified as "very low" on nodata areas (e.g. Italy) Volcanic hazard classified as "low" where there are no volcanoes (e.g India)

stufraser1 commented 5 years ago

This requires discussion.

India is low, because of volcanoes located on Andaman Islands (its just not very visible on map due to small islands)
Coastal flood does need updating to show land-locked ADM units as NoData
I would agree, lets change CY to be NoData (so then it is also clear it is cyclone only, not other winds included)

This requires updating the raw hazard layers to a nodata value and reharvesting the data.

matamadio commented 5 years ago

Coastal flood and cyclones rasters have been corrected: zeroes turned into nodata, should solve the issue with "low hazard". But the geonode map does not update the nodata value.