Checking for missing content type metadata ...
This resource contains content types with missing metadata required to make it public or discoverable. Show missing content type metadata.
Click on the edit button ( ) below to edit this resource.
Checking for non-preferred file/folder path names (may take a long time depending on the number of files/folders) ...
This resource contains some files/folders that have non-preferred characters in their name. Show non-conforming files/folders.
This resource contains content types with files that need to be updated to match with metadata changes. Show content type files that need updating.
Monthly hydroclimatic and operational data and model outputs for forecasting the storage of the Cantareira Water Supply System (São Paulo, Brazil)
| Authors: |
|
|
|---|---|---|
| Owners: |
|
This resource does not have an owner who is an active HydroShare user. Contact CUAHSI (help@cuahsi.org) for information on this resource. |
| Type: | Resource | |
| Storage: | The size of this resource is 92.5 KB | |
| Created: | Nov 09, 2026 at 1:35 p.m. (UTC) | |
| Last updated: | Nov 09, 2026 at 3:29 p.m. (UTC) | |
| Citation: | See how to cite this resource | |
| Content types: | CSV Content |
| Sharing Status: | Public |
|---|---|
| Views: | 74 |
| Downloads: | 57 |
| +1 Votes: | Be the first one to this. |
| Comments: | No comments (yet) |
Abstract
Monthly input series and model outputs of a study that forecasts the total storage of the Cantareira Water Supply System, the largest water supply system of the São Paulo Metropolitan Region (Brazil), 3 and 12 months ahead with XGBoost and LSTM models. Inputs, per incremental basin of the four Cantareira reservoirs: precipitation and reference evapotranspiration (1961–2025, area-weighted aggregates of the BR-DWGD gridded data), GRACE/GRACE-FO terrestrial water storage anomaly (2002–2024), and Sabesp operational records of storage, natural inflow, release, tunnel and inter-basin transfer flows (1980–2026), plus the operational balance flow derived from them. Outputs: forecasts, benchmark comparisons (persistence and seasonal climatology), skill and reliability scores, Diebold–Mariano tests and explainability results (SHAP, saliency and perturbation). Files keep the paths of the code repository, so the code release reproduces every figure and table of the paper. A codebook describes every column and a manifest lists SHA-256 checksums.
Subject Keywords
Coverage
Spatial
Temporal
| Start Date: | |
|---|---|
| End Date: |
Content
Readme.md
Monthly hydroclimatic and operational data and model outputs for forecasting the storage of the Cantareira Water Supply System (São Paulo, Brazil)
This resource holds the monthly input series and the model outputs used in the forecasting paper on the storage of the Cantareira Water Supply System (São Paulo Metropolitan Region, Brazil). In that paper, XGBoost and LSTM models forecast the total system storage 3 and 12 months ahead. The code that reads these files and reproduces every figure and table is published separately:
- Code: https://github.com/LabSid-USP/ml-drought-modeling-studies, archived in Zenodo (see Related Resources).
- Spatial layers of the study area: companion HydroShare resource (see Related Resources).
Files keep the same relative paths as in the code repository, so
scripts/download_data.py restores them in place.
Contents
| Path | What it is |
|---|---|
Previsao/Dados/Pr/Pr_monthly.csv |
Monthly precipitation (mm/month), 1961-01 to 2025-12, per incremental basin |
Previsao/Dados/ETo/ETo_monthly.csv |
Monthly reference evapotranspiration (mm/month), 1961-01 to 2025-12, per incremental basin |
Previsao/Dados/Q/Qnat2_monthly.csv |
Natural (incremental) inflow to each Cantareira reservoir (m³/s), 1980-01 to 2026-05 |
Previsao/Dados/Vol/Vol_monthly.csv |
Stored volume of each Cantareira reservoir (hm³), 1980-01 to 2026-05 |
Previsao/Dados/Q/Qout_monthly.csv |
Downstream release of each Cantareira reservoir (m³/s), 1980-01 to 2026-05 |
Previsao/Dados/Q/Qbal_monthly.csv |
Operational balance flow per reservoir and for the whole system (m³/s), 1980-01 to 2026-05 |
Previsao/Dados/Q/Qt{5,6,7}_monthly.csv |
Flow through Tunnels 5, 6 and 7 (m³/s) |
Previsao/Dados/Q/Qpbs_net_monthly.csv |
Net transfer between the Jaguari reservoir (Paraíba do Sul basin) and Atibainha (m³/s) |
Previsao/Dados/Q/Qelev_monthly.csv |
Santa Inês pumping station flow (m³/s) |
Previsao/Dados/TWS/TWS_monthly.csv |
GRACE/GRACE-FO terrestrial water storage anomaly (cm), 2002-04 to 2024-08, with mission gaps |
Previsao/Outputs/XGBoost/**/article_*.csv |
XGBoost (H = 3) forecasts, ablation, skill scores, Diebold–Mariano tests, SHAP importances, feature correlations |
Previsao/Outputs/LSTM/**/article_*.csv |
LSTM (H = 3, 12) walk-forward forecasts, quantile forecasts, skill and reliability scores, explainability |
codebook.csv |
One row per file × column: unit and description |
MANIFEST-sha256.txt |
SHA-256 checksum of every data file |
All series are monthly; date is the first day of the month. Missing values are empty cells.
Basin codes
| Code | Incremental basin / reservoir | Drainage area (km²) |
|---|---|---|
CT_JAGJAC |
Jaguari/Jacareí | 1239.25 |
CT_CACH |
Cachoeira | 392.38 |
CT_ATIB |
Atibainha | 315.07 |
CT_PCASTR |
Paiva Castro (includes the Águas Claras basin) | 337.57 |
CT_TOTAL |
Whole Cantareira System (the four above) | 2284.27 |
Pr_monthly.csv and ETo_monthly.csv also carry the basins of the Alto Tietê
(AT_*) and Guarapiranga (GP_*) systems. They are not used in the forecasting
paper; they are kept because these two files are open data.
Feature names (model outputs)
pr precipitation; eto reference evapotranspiration; q natural inflow;
Vol storage; qout release; qbal operational balance flow; qpbs_net
Paraíba do Sul transfer; qelev Santa Inês pumping; tws GRACE storage
anomaly; SPI*, SPEI*, SRI* standardized precipitation,
precipitation–evapotranspiration and runoff indices over 3, 6 and 12 months;
_LagN value N months before the issue month; _RollingMeanN mean of the last N
months; month_sin, month_cos, is_dry calendar encodings (dry season =
April–September). Experiment labels A–D (persistence base) and A–D
(seasonal-climatology base) follow the paper.
Sources, processing and terms of use
| Data | Source | Terms |
|---|---|---|
| Precipitation, reference evapotranspiration | Brazilian Daily Weather Gridded Data (BR-DWGD) v3.2.4 (Xavier et al. 2022). Daily grids summed to months and averaged over each incremental basin, weighted by the intersected cell area (UTM 23S) | Cite Xavier et al. (2022). Only basin aggregates are redistributed, not the grids |
| Terrestrial water storage | JPL GRACE/GRACE-FO mascon solution (Watkins et al. 2015; Wiese et al. 2016). The same mascon cell covers the four basins. Product release used: not recorded. Used only in the correlation screening (Fig. S1); not a model feature | Open (NASA Earthdata) |
| Storage, releases, natural inflows, tunnel and transfer flows | Monthly operational records of the Companhia de Saneamento Básico do Estado de São Paulo (Sabesp), extracted by the authors from the Sabesp Decision Support System (SSD). Restricted to the Cantareira reservoirs | See Sabesp data below |
Operational balance flow (Qbal) |
Derived by the authors from the Sabesp series above (formulas below) | As Sabesp data |
Model outputs (Previsao/Outputs) |
Produced by the authors with the code release cited above | CC BY 4.0 |
Natural inflow. Sabesp's SSD computes the natural (incremental) inflow of each
reservoir from its water balance. The balance combines the change in storage, the
downstream release and the inflows and outflows through tunnels. For example,
Jaguari/Jacareí = ΔV + release + Tunnel 7. The system inflow is the sum over the
basins. The SSD provides two formulations; this study uses the second one
(Qnat2).
Operational balance flow (authors' formulation; each reservoir, or the whole
system, taken as control volume; release = Qout):
- Jaguari/Jacareí: Qnat − release − Q_T7
- Cachoeira: Qnat + Q_T7 − release − Q_T6
- Atibainha: Qnat + Q_T6 + Q_PBS,net − release − Q_T5
- Paiva Castro: Qnat + Q_T5 − release − Q_Santa Inês
- System (
CT_TOTAL): ΣQnat − Σrelease + Q_PBS,net − Q_Santa Inês. The internal tunnels cancel out.
A month with any missing term is left empty.
Sabesp data. The monthly Sabesp series in this resource are the ones the models used. The underlying raw SSD records are owned by Sabesp and are not publicly available. They can be obtained from the authors upon reasonable request and with permission of Sabesp. Credit the Sabesp series as: "Sabesp, Companhia de Saneamento Básico do Estado de São Paulo, via the Sabesp Decision Support System (SSD)".
Licence
Model outputs and the authors' derived series: Creative Commons Attribution 4.0 International (CC BY 4.0). Third-party data keep their original terms, listed above. Credit the original sources when reusing them.
Acknowledgements
The authors thank the Fundação Centro Tecnológico de Hidráulica (FCTH), LabSid (Escola Politécnica, Universidade de São Paulo), and Sabesp for the operational records.
References
- Watkins MM, Wiese DN, Yuan D-N, Boening C, Landerer FW (2015) Improved methods for observing Earth's time variable mass distribution with GRACE using spherical cap mascons. J Geophys Res Solid Earth 120:2648–2671. https://doi.org/10.1002/2014JB011547
- Wiese DN, Landerer FW, Watkins MM (2016) Quantifying and reducing leakage errors in the JPL RL05M GRACE mascon solution. Water Resour Res 52:7490–7502. https://doi.org/10.1002/2016WR019344
- Xavier AC, Scanlon BR, King CW, Alves AI (2022) New improved Brazilian daily weather gridded data (1961–2020). Int J Climatol 42:8390–8404. https://doi.org/10.1002/joc.7731
Package built by scripts/build_hydroshare_packages.py from commit ffd923905b9f031a11415b76e2334d3979824a3a.
Related Resources
| This resource is described by | DOI of the forecasting paper (after acceptance) |
| The content of this resource can be executed by | https://doi.org/10.5281/zenodo.24085808 |
Credits
Contributors
People or Organizations that contributed technically, materially, financially, or provided general support for the creation of the resource's content but are not considered authors.
| Name | Organization | Address | Phone | Author Identifiers |
|---|---|---|---|---|
| Gerald Augusto Corzo-Pérez | IHE Delft Institute for Water Education | ORCID | ||
| Joaquin Ignacio Bonnecarrère | Department of Hydraulic and Environmental Engineering, Escola Politécnica, Universidade de São Paulo | ORCID | ||
| Dário Hachisu Hossoda | Department of Hydraulic and Environmental Engineering, Escola Politécnica, Universidade de São Paulo | ORCID |
How to Cite
This resource is shared under the Creative Commons Attribution CC BY.
http://creativecommons.org/licenses/by/4.0/
Comments
There are currently no comments
New Comment