Input data for LSTM and seq2seq LSTM surrogate models for multi-step-ahead street-scale flood forecasting in Norfolk, VA

This resource includes the input data for LSTM and seq2seq LSTM surrogate models for multi-step-ahead street-scale flood forecasting in Norfolk, VA, USA. The data consists of topographic features: topographic wetness index (TWI), depth to water (DTW) and elevation, and environmental features: hourly rainfall and tide level from gauge stations and water depth generated by the physics-based model TUFLOW.

There are three folders in this resource -
1. The \"OriginalData\" folder includes the CSV files for the top 20 daily storm events from 2016-2018 for the streets of Norfolk.
2. The \"FloodproneStreets\" folder includes shapefiles of the street segments (polygons of 7.2 m width x 50 m length) of Norfolk. Alongside, it includes a CSV file containing 22 flood-prone streets selected from the STORM report.
3. The \"RelationalDatabase\" folder includes three CSV files for node_data (varied spatially), tide_data (varied temporally) and weather_data (varied spatially and temporally) for efficient data management. The notebook script \"create_relational_data.ipynb\" is used to convert \"OriginalData\" to \"RelationalDatabase\".

The Python script of the LSTM and seq2seq LSTM surrogate models is available on GitHub https://github.com/br3xk/LSTM-and-seq2seq-LSTM-surrogate-models-for-street-scale-flood-forecasting
The output of the model is forecasted hourly water depth on the 22 flood-prone streets with 4-hr and 8-hr lead.