sleuth-automation

Python wrapper for SLEUTH urban growth model.

Project description

This library is an object-oriented wrapper for the SLEUTH urban growth model.

It will automatically create scenario files from directories containing data layers and it can run simulations through MPI and HT-Condor.

Installation

You may install this library and helper scripts using pip.

$ pip install sleuth_automation

Application Programming Interface

import sleuth_automation as sa

# the library must be configured at least with the path to SLEUTH
sa.configure(sleuth_path='/opt/sleuth',
             use_mpi=True, mpi_cores=32)


# a directory containing input layers is given to a location instance
l = sa.Location('my_location',
                '/path/to/my_location')

l.calibrate_coarse()
l.calibrate_fine()
l.calibrate_final()

l.sleuth_predict(end=2050)

Command Line Interface

A single run may be achieved using the included sleuth_run.py script.

$ sleuth_run.py --sleuth_path /opt/sleuth/ \
                --location_dir /path/to/location/ \
                --location_name my_location \
                --mpi_cores 40 \
                --montecarlo_iterations 50 \
                --predict_end 2060

This will create scenario files for coarse, fine and final stages of calibration, extracting parameters from the control_stats.log files, and run predict.

If one wants to predict for several locations, one may group them in a directory and run them as a batch. Using the create_sleuth_condor_batch.py one may create a batch run for the HT-Condor queue management system.

$ create_sleuth_condor_batch.py --sleuth_path /opt/sleuth/ \
                                --region_dir /path/to/locations_group/ \
                                --mpi_cores 32 \
                                --predict_end 2060

This will create a submit.condor file in the locations directory, setup with the appropiate sleuth_run.py commands.