pyINSYDE

pyINSYDE

Python port of the INSYDE synthetic flood damage model (Dottori et al, 2016).

* original author: Cella Schnabel, rms428@cornell.edu * corresponding author: Vivek Srikrishnan, vs498@cornell.edu

Purpose

Estimating damages from flooding are often based on simple, deterministic depth-damage functions (DDFs) which rely on historical claims or reporting data and expert assessments. While large-scale damage estimates using DDFs are readily scalable, the underlying estimation methods can be subject to input data errors or the typical potential biases associated with expert assessments. While aggregate damage estimates may be robust to these errors, they may matter more for individual properties. Moreover, there may be considerable uncertainties associated with how both building characteristics and flood dynamics influence the depth-damage relationship.

INSYDE is a synthetic damage model using a reduced-form but mechanistically-motivated representation of building damages and associated repair and recovery costs. The goal of pyINSYDE is to provide a Python port of this model to faciliate assessments of depth-damage uncertainty.

Journal reference

Original INSYDE reference:

Dottori, F., Figueiredo, R., Martina, M. L. V., Molinari, D., & Scorzini, A. R. (2016). INSYDE: a synthetic, probabilistic flood damage model based on explicit cost analysis. Natural Hazards and Earth System Sciences, 16(12), 2577–2591. https://doi.org/10.5194/nhess-16-2577-2016

Installation

Dependencies

pyINSYDE was developed using Python 3.9.

Package	Version
numpy	>=2.0
scipy	>=1.9

Instructions

For Users

To use pyINSYDE, install using pip install git+https://github.com/vsrikrish/pyinsyde. We recommend installing pyINSYDE in a virtual environment.

For Developers

If you would like to edit the source code:

1. Clone the repository:

   git clone https://github.com/vsrikrish/pyinsyde.git
   cd pyinsyde

2. Install the dependencies above (ideally in a virtual environment). pyINSYDE is designed to be lightweight and should work with most versions of Python 3, NumPy, and SciPy. If you use older versions, you can test functionality with test/insyde_example.py.

3. Install pyINSYDE in development mode: pip install -e ..

Getting Started

Buildings

pyINSYDE is structured around Building objects, which can then be subjected to a variety flood events. To create a new Building, construct a dictionary containing the relevant attributes and unpack this dictionary into a BuildingProperties dataclass. Finally, call the Building constructor using the BuildingProperties object.

Example:

from pyinsyde import BuildingProperties, Building

property_dict = ...
bp = BuildingProperties(**property_dict)
b = Building(bp)

A BuildingProperties object can also be constructed without the initial dictionary, but this approach increases the flexibility of specifying attributes.

In addition to the structural characteristics for the building, there are a few optional settings which can be set in the initial dictionary.

• Specifying a value for sd will add uncertainty to the component-wide damage estimates based on that number of standard deviations relative to the associated fragility curve range.
• By default, pyINSYDE damage accounting will use the replacement value and unit price data from the original INSYDE model. These can be replaced by passing (absolute) paths as rp_val and up_val. We assume that these will have the same formatting as the original INSYDE data (found in data/replacement_values.txt and data/unit_prices.txt).

Usage Examples

Checkout the examples folder to work through a basic sensitivity analysis example and learn how to get started with pyINSYDE.

We are always working on creating more usage examples. Please feel free to contribute some from your own work!

Documentation

Coming...

Contribution Guidelines

Contributions are welcome, especially those aimed at increasing the flexibility of pyINSYDE to new locations or datasets! Please fork, submit pull requests, or report bugs.

License

pyINSYDE uses the GNU GPLv3 license. For more details, see GNU GPLv3 on Choose A License.