# EXPLORING THE IMPLICATIONS OF MODELING CHOICES ON PREDICTION OF WATER SAVINGS
## CODE AND DATA REPOSITORY

This repository includes the scripts (for analysis and figures) and results for the paper "Exploring the implications of modeling choices on prediction of water savings."

There are five folders in the repository:
1. Input: This folder contains two folders (climate folder, run files folder)
The climate folder holds the precip and temperature data for the hydrologic grid cells.
The run files hold the grid cells' metadata and static information (including hydraulic routing information).

2. Runoff: This folder contains a single file - the runoff observations

3. Scripts: This folder contains three folders (HBVFunctions, helper_functions, and hydrologic_models)
(a) The HBVFunctions FOLDER holds scripts that perform some background work - especially for model calibration. Do not worry too much about this folder. If you want to run the code, you must have this folder in the same directory as the analysis script.
(b) The helper_functions folder is a folder to help process raw outputs from the BORG algorithm
(c)The hydrologic_models folder holds scripts for the hydrologic models that are used in the study. The models are Python implementations of the respective hydrologic equations.
(d) six python scripts start with "figure" plot the figures in this paper. You may need to install packages to run these files. However, the packages are all listed in import statements at the top of the files.
(e) the script "mpi_runFullExperiment.py" is the script to run the entire experiment presented in this paper. You need the hydrologic_models folder and the HBV_Functions folder in the same directory if you want to run this file.
(f) the shell script "test_submission.sh" is a file that I used to submit the experiment on the cluster. This file is formatted for the LSH submit system. If you are using a different cluster system, you may need to modify this file.

4. Output: This folder holds (in theory) four folders (actually, two folders):
(a) RAW_CALIBRATION_PARAMETERS - are the outputs of the BORG algorithm used to calibrate the hydrologic models.
(b) weru_weru_parameters - are the 100 calibrated parameters for each hydrologic model
(c) output_FULL_EXPERIMENT_0604 - are all the results from the experiment (due to space restrictions on the cluster, only hydrologic state variables on irrigated grid cells are returned for the complete experiment - 100 parameter sets for each model) - this folder is over 20 GB in size, so is not uploaded - please create this folder and then run the experiment and save the results here.
(d) output_SAMPLE_WATERBALANCE_0603 - sample results contain (for a few parameter sets) the hydrologic state variables for all the grid cells (irrigated or not) for each model - this folder is over 20 GB in size, so it is not uploaded - please create this folder and then run the experiment and save the results here.
This folder also contains CSV files that are the calibrated runoff results for all the hydrologic models' parameters.

Please email me if you have any questions (celuwa@umass.edu)