/bayarea_urbansim

UrbanSim implementation for theoretical second bay crossing, CP218 Transportation Studio at UC-Berkeley, Dept of City and Regional Planning, Fall 2016

Primary LanguagePython

DRAFT Bay Area Urbansim Implementation

This is the DRAFT UrbanSim implementation for the Bay Area. Documenation for the Bay Area model is available at http://metropolitantransportationcommission.github.io/baus_docs/ and documentation for the generic UrbanSim model is at https://udst.github.io/urbansim/index.html

###Install Overview

###Data

We track the data for this project in the Makefile in this repository. The makefile will generally be the most up to date list of which data is needed, where it goes in the directory, etc.

To fetch data with AWS CLI and Make, you can: make data.

Below we provide a list to links of the data in the Makefile for convenience, but in general the makefile is what is being used to run simulations. If you find that something below is out of date w/r/t the makefile, please feel free to update it and submit a pull request.

####Data necessary for run.py to run

These data should be in the data/ folder:

https://s3.amazonaws.com/bayarea_urbansim/data/2015_06_01_osm_bayarea4326.h5
https://s3.amazonaws.com/bayarea_urbansim/data/2015_08_03_tmnet.h5
https://s3.amazonaws.com/bayarea_urbansim/data/2015_12_21_zoning_parcels.csv
https://s3.amazonaws.com/bayarea_urbansim/data/02_01_2016_parcels_geography.csv
https://s3.amazonaws.com/bayarea_urbansim/data/2015_08_29_costar.csv
https://s3.amazonaws.com/bayarea_urbansim/data/2015_09_01_bayarea_v3.h5

Because the hdf5 file used here contains one table with proprietary data, you will need to enter credentials to download it. You can request them from Tom Buckley(tbuckl@mtc.ca.gov). Or if you already have access to Box, you can download the hdf5 file from there.

####Data Description

How To

####Set Up Simulation and Estimation
Install dependencies using standard pip requirements install: pip install -r requirements.txt You may also need to install pandana pip install pandana

####Set up using a Virtual Machine For convenience, there is a Vagrantfile and a scripts/vagrant/bootstrap.sh file. This is the recommended way to set up and run Simulation.py on Windows.

####Enter Amazon Web Services credentials to fetch data.

See Installing and [configuring] (http://docs.aws.amazon.com/cli/latest/userguide/cli-chap-getting-started.html)

Each of the following just runs a different set of models for a different set of years.

####Run a Simulation
In the repository directory type python run.py

####Estimate Regressions used in the Simulation In the repository directory edit run.py and set MODE to "estimation" and type python run.py

####Run a Base Year Simulation In the repository directory edit run.py and set MODE to "baseyearsim" and type python run.py. A base year simulation is used to run a few models and make sure everything matches the first year of the control totals but not to add any new buildings. This is then used in comparison of the year 2040 to the base year for all future simulations (until the control totals change) and this mode is rerun.

####Review Outputs from Simulation

#####Runs Directory

ALL OUTPUT IN THIS DIRECTORY IS CONSIDERED DRAFT. PLEASE CONTACT MTC FOR OFFICIAL FINAL OUTPUTS.

# = a number that is updated in the RUNNUM file in the bayarea_urbansim directory each time you run Simulation.py.

Many files are output to the runs/ directory. They are described below.

filename description
run#_topsheet_2040 An overall summary of various housing, employment, etc by regional planning area types
run#_parcel_output.csv csv of parcels that are built for review in Explorer
run#_parcel_data_diff.csv A CSV with parcel level output for all parcels with lat, lng and includes change in total_residential_units and change in total_job_spaces, as well as zoned capacity measures
run#_simulation_output.json summary by TAZ for review in Explorer (unix only)
run#_taz_summaries A CSV for input to the MTC travel model
run#_urban_footprint_summary A CSV with A Summary of how close the scenario is to meeting Performance Target 4

Browse results here

######Other Directories Below is an explanation of the directories in this repository not described above.

configs/

The YAML files in this directory allow you to configure UrbanSim by changing the keys and values of arguments taken by urbansim functions. See the UrbanSim Defaults docs for more details.

Note that even the values taken by data can be and are configured with these config files (e.g. values in settings.yaml).

data_regeneration/

The scripts in here can be used to re-create the data in the data/ folder from source (various local, state, and federal sources). Use these to re-create the data here when source data change fundamentally.

scripts/ This is a good place to put scripts that can exist independently of the analysis environment here.

####Parcel Geometries

The parcel geometries are the basis of many operations in the simulation. For example, as one can see in this pull request, in order to add schedule real estate development projects to the list of projects that are included in the simulation, one must use an existing geom_id, which is a field on the parcels table added here.

Parcel geometries are available at the following link:

https://s3.amazonaws.com/bayarea_urbansim/data/09_01_2015_parcel_shareable.zip

#####Geom ID

Please be aware that many ArcGIS users have found that ArcGIS automatically converts and then rounds the geom_id column, effectively making it unusable. Therefore we recommend using QGIS, which does not exhibit this behavior with delimited files by default.

Also, in Microsoft Excel, you will need to make sure that the data type of the geom_id column is set to number and that the number of decimal points is set to 0. Otherwise when you save the CSV again the geom_ids will be unusable.

What is the geom_id field and why does it exist?

In short, this is a legacy identifier. The geom_id field was introduced as a stable identifier for parcels across shapefiles, database tables, CSV's, and other data types. It is an integer because at some point there was a need to support integer only identifiers. It is not based on an Assessor's Parcel Numbers because there was a perception that those were inadequate. And it is based on the geometry of the parcel because many users have found that geometries are the most important feature of parcels.