C+GDROM Pre-calibrated Reservoir Parameters for 1,758 U.S. Reservoirs

Overview

Numerous dams and reservoirs have fundamentally altered streamflow regimes and terrestrial hydrology. However, large-scale hydrological models still depend on simplified and often unrealistic reservoir operation schemes due to limited understanding of real-world operation complexities and inaccessibility of long-term in-situ operation data. C+GDROM, a hybrid empirical–conceptual model for daily reservoir operation simulation, is developed to address these challenges by reducing data requirements while improving compatibility with hydrological models.

The C+GDROM framework introduces three reservoir operation models: General Model, applicable to all reservoirs regardless of function or capacity, and two specialized models—Flood Control Model and Irrigation Model—designed for reservoirs primarily serving flood control or irrigation purpose, respectively. In all three models, reservoir-specific conceptual storage regulation curve is a required input, which helps maintain physically consistent storage dynamics and guide release decisions. Leveraging a recently published reservoir dataset GDROM v2 (Zheng et al., 2025), this dataset provides pre-calibrated storage curve parameters for implementing the C+GDROM across 1,758 reservoirs in the United States.

Further details of the C+GDROM framework are described in:

A Hybrid Empirical and Conceptual Model for Improving the Representation of Reservoirs with Limited Data in Hydrological Models, published in Water Resources Research (2026).

For full C+GDROM implementation, source code and documentation are available in the GitHub repository:

👉 https://github.com/yananc2024/C-GDROM

Data Source

C+GDROM parameters are calibrated from the dataset:

• GDROM v2, a reservoir dataset that includes historical operation records for 2,017 reservoirs across the contiguous United States

Reservoir storage data in GDROM v2 includes:

• in-situ daily observations
• remotely sensed storage estimates (weekly or monthly)
• gap-filled values where sufficient neighboring information is available

From this dataset, we selected 1,758 reservoirs that satisfy the following criteria:

• Minimum data length: At least 5 years of storage data after 1990
• Sufficient seasonal coverage: When storage records are aggregated to representative medians, data must cover at least 10 months (monthly data), or 44 weeks (weekly data), or 305 days (daily data)

Notes: 1. We use storage data from 1990 onward to ensure that the derived conceptual storage curve reflects recent typical storage patterns. 2. Because the conceptual storage regulation curve is derived from historical characteristic medians, sufficient intra-annual coverage is required. The ≥10-month (or equivalent) threshold is imposed to ensure that key seasonal signals (e.g., refill and drawdown phases) are adequately captured and not biased by missing periods.

Dataset Contents

The dataset includes:

1. Reservoir parameter table (conceptual_S_curve_para_1758reservoirs.csv)

A CSV file containing the calibrated parameters of conceptual storage regulation curve, for each reservoir: - GRAND_ID: Unique reservoir identifier - curve_shape: overall shape of the storage curve, either 'four-piece' (with a clear refill and drawdown cycle) or 'single' (maintaining a relatively stable storage) - A1-A4 : timing parameters for storage transitions (months) - S_A4-A1 and S_A2-A3 : representative storage levels of transition stages (note: when the relative difference between S_A4-A1 and S_A2-A3 is less than 10%, seasonal storage variation is considered negligible, and the curve shape is noted as 'single') - S_median: historical median storage (approximated by 0.5*(S_A4-A1 +S_A2-A3); used only when the curve shape is 'single')

2. Reservoir metadata table (metadata_1758reservoirs.csv)

The reservoir metadata table provides essential information for all 1,758 reservoirs included in this dataset. The columns are grouped into three categories:

1. Basic Reservoir Attributes (mainly derived from the GRanD database, Lehner et al., 2011)
2. GRAND_ID: Unique reservoir identifier in the GRanD database. For reservoirs added in GDROM v2 but not listed in GRanD, an ID is assigned starting from 10000.
3. GRAND_DAM_NAME: Reservoir/dam name in GRanD
4. STATE: U.S. state where the reservoir is located
5. LONGITUDE, LATITUDE: Geographic coordinates of the reservoir
6. MAIN_USE: Primary operational purpose of the reservoir

7. USE_IRRI, USE_ELEC, USE_SUPP, USE_FCON, USE_RECR, USE_NAVI: Operational purpose indicators for irrigation, hydroelectricity, water supply, flood control, recreation, navigation, respectively. "Main" denotes primary operational purpose, "Sec" indicates secondary purpose, and a blank cell means the reservoir is not used for that purpose.

8. Storage Data Coverage

These fields describe the temporal coverage and characteristics of storage data from GDROM v2 used to derive the conceptual storage regulation curve:

• S_YEAR_RANGE: Time span of available storage data after 1990, e.g., 1995-2015
• S_YEAR_NUM: Total number of years with available storage data
• S_FLAG: Dominant temporal resolution of storage data (daily, weekly, or monthly)
• S_LENGTH: Total number of storage records
• S_COVER_MONTH: Months covered by available storage data
• S_MAX_HIST_AF: Historical maximum storage (acre-feet)

• S_MIN_HIST_AF: Historical minimum storage (acre-feet)

• Reservoir Information from National Inventory of Dams (NID)

These fields provide additional reservoir information retrieved from the National Inventory of Dams (NID):

• NID_ID: Unique dam identifier in the NID database
• NID_DAM_NAME: Dam name in NID
• NID_CAP_AF: Reported storage capacity (acre-feet) from NID

Users may use the NID_ID and NID_DAM_NAME to retrieve additional reservoir attributes directly from the NID database. The full NID database contains over 90,000 dams; in this study, we consider a subset (~3,500 dams) with storage capacity exceeding 10 million cubic meters for matching with GDROM v2 reservoirs. Matching between these two datasets is performed using both dam name similarity and spatial proximity (within 5 km). For 87 reservoirs, no corresponding NID record was identified based on these criteria.

To ensure completeness, the column GRAND_CAP_AF is provided as an alternative estimate of storage capacity from the GRanD database.

Derivation of Conceptual Storage Regulation Curve

Reservoir-specific conceptual storage regulation curve is required input for C+GDROM, which is parameterized using six parameters: A1-A4 (timing parameters), $S_{A4-A1}$ and $S_{A2-A3}$ (representative storage levels of transition stages).

Storage curve parameters are derived using GDROM v2 storage data. For reservoirs with daily data, the median storage for each day of year (DOY) is computed. For reservoirs with weekly or monthly data (commonly for remote sensing storage estimates), weekly or monthly median storage values are used for curve derivation.

The derived conceptual-storage-curve parameters for 1,758 studied reservoirs are provided in this dataset. Users may follow the steps in Conceptual_S_curve.ipynb to derive curve parameters for their reservoirs, with necessary functions defined in conceptual_s_curve.py.

Prepare Input Data for C+GDROM

This dataset provides parameters of the conceptual storage regulation curve. To run the C+GDROM model for a reservoir, the following additional information is required:

1. Inflow statistics

Daily inflow statistics, including: - the 99th, 80th, 50th, 30th, 10th (I99, I80, I50, I30, I10), and mean of daily inflow values ($I_{mean}$)

These statistics can be computed from multi-year inflow series obtained from hydrological model simulations.

Using the provided storage curve parameters and computed $I_{mean}$ (in volume), reservoir size ratio, an important reservoir feature used in C+GDROM, can be computed:

$$size\ ratio = {S_{cap}-\min(S_{A4-A1}, S_{A2-A3}) \over I_{mean}*365}$$

2. Storage capacity information

Required parameters: - Total capacity (S_cap) - Dead storage capacity (S_dead) - Flood control capacity (S_flood_cap)

Notes: - Total storage capacity can be obtained from public databases. The file metadata_1758reservoirs.csv already provides total capacity from NID and GRanD database. In addition, if long-term (e.g., >= 25-year) in-situ storage data are available, the historical maximum storage (S_MAX_HIST_AF) can be used as an alternative estimate of total capacity. Notably, we noticed discrepancies between reported total capacities (NID and GRanD) and observed storage values for some reservoirs. Users may need to further verify capacity information for specific reservoirs.

• For dead storage capacity, a common approximation is 10% of total capacity. Alternatively, the historical minimum storage (S_MIN_HIST_AF) can be used when long-term records are available

• For flood control capacity (i.e., capacity to the top of controlled flood control pool), our study used 0.99 × total capacity (based on historical maximum storage) as an approximation. Other studies have used lower fractions (e.g., 0.97; Zajac et al., 2017). If reported capacity significantly exceeds observed storage, a lower fraction may provide a more realistic estimate.

Summary

conceptual storage curve parameters, reservoir inflow statistics, and reservoir storage capacity information consist of all the needed model setup information to run C+GDROM General Model, which is the recommended model for usage under data-scarce conditions. If users want to use Flood Control Model or Irrigation Model for improved operation representation for specific operational purposes, in-situ (or sufficiently accurate RS-derived) operation data (≥ 2 years of daily inflow, release, and storage) are required for further calibration of release module parameters (detailed in the WRR paper).

For full model implementation, including step-by-step examples and guidance on coupling C+GDROM with hydrological models, please refer to the GitHub repository.

References

Chen, Y., Zheng, Y., Cai, X., Bin Y. & Zheng, Z. (2026). [A Hybrid Empirical and Conceptual Model for Improving the Representation of Reservoirs with Limited Data in Hydrological Models], Water Resources Research.

Lehner, B., Liermann, C. R., Revenga, C., Vörösmarty, C., Fekete, B., Crouzet, P., et al. (2011). [High-resolution mapping of the world's reservoirs and dams for sustainable river-flow management], Frontiers in Ecology and the Environment, 9(9), 494-502. https://doi.org/10.1890/100125

U.S. Army Corps of Engineers. National Inventory of Dams. Retrieved March 9, 2026, from https://nid.sec.usace.army.mil/nid/#/

Zajac, Z., Revilla-Romero, B., Salamon, P., Burek, P., Hirpa, F. A. & Beck, H. (2017). [The impact of lake and reservoir parameterization on global streamflow simulation], Journal of Hydrology, 548, 552-568. https://doi.org/10.1016/j.jhydrol.2017.03.022

Zheng, Z., Cai, X., Zhang, L., Li, J. & Chen, Y. (2025). [GDROM v2: An Inventory of Operation Variables Time Series and Rules for 2,017 Large Reservoirs across the CONUS], Scientific Data, 12(1), 1891. https://doi.org/10.1038/s41597-025-06162-7

Citation

If you use this dataset, please cite:

Chen, Y., Zheng, Y., Cai, X., Bin Y. & Zheng, Z. (2026). [A Hybrid Empirical and Conceptual Model for Improving the Representation of Reservoirs with Limited Data in Hydrological Models], Water Resources Research.

Chen, Y., Zheng, Z., Wang, Y., & Cai, X. (2026). [C+GDROM Pre-calibrated Reservoir Parameters for 1,758 U.S. Reservoirs]. HydroShare.

Contact

If you have any questions or would like to contribute, please contact Yanan Chen (yananc2024@outlook.com)