Locations of upstream processes

[dt-woods]

With the new regionalized upstream renewable construction processes (and further regionalization anticipated), it is worthwhile revisiting locations associated with these processes. Currently, the process dictionary skips location in upstream_dict.py (snippet below)

ElectricityLCI/electricitylci/upstream_dict.py

Line 277 in 60399f7

def _process_table_creation_gen(process_name, exchanges_list, fuel_type):

However, there is a placeholder for adding location (see commented out code).

ElectricityLCI/electricitylci/upstream_dict.py

Line 331 in 60399f7

ar["location"] = "" # location(region)

At the plant stage, locations are listed in a CSV file in the repository's data folder and comprises of a surprisingly short list of location codes. It's notable that none of the UUIDs in this CSV file are found in the JSON-LD.

https://github.com/NETL-RIC/ElectricityLCI/blob/development/electricitylci/data/location_UUIDs.csv

Any location that shows up without its own UUID gets one created in the _location method found in olca_jsonld_writer.py, which expects ISO two-letter country codes. The reality is that locations are created using balancing authority names, FERC codes, and 'US'.

Since eLCI v2, all generation upstream processes have their locations associated with their region (e.g., 'Electricity - BIOMASS - Bonneville Power Administration' is associate with its BA location).

Upstream processes missing location include

• coal extraction and processing (e.g., 'coal extraction and processing - Northern Appalachia, BIT, Underground')
• coal transport (e.g., 'coal transport - Barge')
• natural gas extraction and processing (e.g., 'natural gas extraction and processing - Midwest')
• power plant construction (e.g. 'power plant construction - coal_const - Duke Energy Carolinas')
• petroleum extraction and processing (e.g., 'petroleum extraction and processing - DFO PADD 1')
• nuclear fuel extraction, processing, and transport

The immediate quick solution would be to associate all processes without a location to 'United States'. Additional regionalization is possible given the naming of certain processes.

[dt-woods]

Note there is a slight disconnect between the United States UUID in the location_UUID.csv

ElectricityLCI/electricitylci/data/location_UUIDs.csv

Line 39 in 71a3f6d

United States,91477fbe-7ee0-4795-a92b-260ebd0201b5

as compared the the UUID provided in the standard background data from openLCA

• Name: United States of America (the)
• Category: Country
• Code: US
• UUID: 0b3b97fa-6688-3c56-88ee-4ae80ec0c3c2

Not sure if these should be aligned.

[bl-young]

Not sure if these should be aligned.

This should match the standard US UUID, otherwise regionalized LCIA methods won't align (I think)

[dt-woods]

I'm guessing openLCA locations have changed since 8 years ago.
81ac301
Maybe we abandon this CSV and use Fed Commons to get a location JSON (similar to how we get the pedigree matrix from the Core Database and the flow properties from the elementary flow list on the Commons).

Is there a repository from which we can pull locations?

[bl-young]

I'll point you to two functions that we've used to construct these for various applications:
esupy.location extract_coordinates to get the standard geojson data. And here is an example of building a dictionary of location objects with that geojson data.

Note that GreenDelta publishes their list here: https://github.com/GreenDelta/data/blob/master/refdata/locations.csv

And they also use the ecoinvent data that esupy draws on.