1.2. Asset Description¶

This study used a parcel-level inventory of buildings in the Bay Area that was developed by UrbanSim ([Waddell02]) using public resources such as the City and County of San Francisco’s data portal ([DataSF20]) and tax assessor databases.

The raw database, which includes 1.8 M San Francisco buildings, was shared in collaboration with UrbanSim. The UrbanSim database includes two files:

Buildings File: a CSV file that contains building properties including total floor area, number of stories, year built, building occupancy, and parcel ID.
Parcels File: a CSV file including the latitude and longitude of each building defined by the parcel ID.

A parsing application was built to collect the mentioned building properties from the UrbanSim building files. The occupancy ID (in integer) was used to infer the occupancy type and replacement cost per unit area (Table 1.2.1). For buildings with missing or invalid occupancy ID, the building was mapped to the default occupancy type (i.e., residential) with the average area of buildings in the inventory.

Table 1.2.1 Mapping rules for building occupancy type and replacement cost.¶
Occupancy ID	Occupancy Class	Replacement Cost / SQFT	Contents Factor
1/2/3/12	Residential	137.5	0.5
4/14	Office	131.9	1.0
5	Hotel	137.3	0.5
6	School	142.1	1.3
7	Industrial	97.5	1.5
8	Industrial	86.0	1.5
9	Industrial	104.0	1.5
10/11/13	Retail	105.3	1.0
Default	Residential	137.5	0.5

The structure types were also mapped from the building occupancy types along with the year built and number of stories. Unless the building was mapped to a single structure type, the structure type is considered random, in which case the mapped structure types are equally likely. Table 1.2.2 summarizes the structure type mapping.

Table 1.2.2 Mapping rules for structure type.¶
Year Built	Number of Stories	Occupancy	Mapped Structure Types
Older than 1900	Any	Any	Wood (W1) and Masonry (RM1, RM2, URM)
Newer than 1900	1~3	Residential	Wood (W1)
Newer than 1900	1~3	Comercial	Wood, Steel, Concrete, Reinforced Masonary (W1, S1, S2, C1, C2, C3, RM1, RM2)
Newer than 1900	1~3	Industrial	Steel, Concrete, and Reinforced Masonary (S1, S2, C1, C2, C3, RM1, RM2)
Newer than 1900	1~3	Other	Wood (W1)
Newer than 1900	4~7	Residential and Commercial	Wood, Steel, Concrete (W1, S1, S2, C1, C2)
Newer than 1900	4~7	Industrial	Steel, Concrete (S1, S2, C1, C2)
Newer than 1900	4~7	Other	Wood, Steel, Concrete (W1, S1, S2, C1, C2)
Newer than 1900	More than 7	Any	Steel and Concrete (S1, S2, C1, C2)

The available information about location and building geometry was further refined by merging the UrbanSim database with the publicly available Microsoft Building Footprint data ([Microsoft20]) for the testbed area. These data were used to populate two additional attributes, replacement cost and structure type, based on a ruleset that considers local design practice and real estate pricing. For further details about the database and ruleset, see [Elhadded19].

Waddell02: Waddell, P. (2002). UrbanSim: Modeling Urban Development for Land Use, Transportation, and Environmental Planning. J. Am. Planning Assoc. 68, 297–314. doi: 10.1080/01944360208976274
DataSF20: DataSF (2020). Building and Infrastructure Databases. San Francisco, SF: DataSF.
Microsoft20: Microsoft (2020). Microsoft Building Footprint Database for the United States. Washington, DC. https://www.microsoft.com/en-us/maps/building-footprints.
Elhadded19: Elhaddad, W., McKenna, F., Rynge, M., Lowe, J. B., Wang, C., and Zsarnoczay, A. (2019). NHERI-SimCenter/WorkflowRegionalEarthquake: rWHALE (Version v1.1.0). http://doi.org/10.5281/zenodo.2554610