There are a number of potental things we can do to make HydroVIS processes easier, quicker, less memory intensitive and cheaper.
1) HydroVIS has a significant inefficiency that they can not easily compensate for and would be easier for FIM dev to do it.
Currently, HydroVIS has to load the REMs directly from S3 as they can not be stored into the PostGres database. They load the catchments at the same time from S3, and use both files against incoming forecast data to inundate and create the vectors. In that process they have to reproject everything to EPSG:3857 to match NWM forecast data.
On the hand side, we certainly want to keep processing as 5070 for CONUS AND 3338 for Alaska, but before we turn it over to HV, reproject it all to 3857. That way they can pull from S3 and not need all of that reproject how many times per day. It is far easier for Fim Dev to do this as a script for Prod runs as part of handing over a new HAND dataset.
2) There are a number of columns in various places that they drop or rename. We can help take care of that ahead of time. More details coming
3) Other efficiencies have been detected and details to come.
This is stuff we can look at implementing for the spring release of HydroVIS 2.1.9 (ish)
There are a number of potental things we can do to make HydroVIS processes easier, quicker, less memory intensitive and cheaper.
1) HydroVIS has a significant inefficiency that they can not easily compensate for and would be easier for FIM dev to do it.
Currently, HydroVIS has to load the REMs directly from S3 as they can not be stored into the PostGres database. They load the catchments at the same time from S3, and use both files against incoming forecast data to inundate and create the vectors. In that process they have to reproject everything to EPSG:3857 to match NWM forecast data.
On the hand side, we certainly want to keep processing as 5070 for CONUS AND 3338 for Alaska, but before we turn it over to HV, reproject it all to 3857. That way they can pull from S3 and not need all of that reproject how many times per day. It is far easier for Fim Dev to do this as a script for Prod runs as part of handing over a new HAND dataset.
2) There are a number of columns in various places that they drop or rename. We can help take care of that ahead of time. More details coming
3) Other efficiencies have been detected and details to come.
This is stuff we can look at implementing for the spring release of HydroVIS 2.1.9 (ish)