Reservoir Estimation of Capacity Loss using AI-based Methods (RECLAIM)

First-of-its-kind globally scalable tool to predict reservoir sedimentation, screen vulnerable reservoirs, and pinpoint those struggling the most.

RECLAIM (Reservoir Estimation of Capacity Loss using AI-based Methods) is a globally scalable machine learning framework to predict absolute sedimentation rates in reservoirs using observed records and multi-decadal satellite-based Earth observations.

It is the first-of-its-kind tool to:

• Rapidly assess sedimentation risk
• Provide scalable predictions across diverse climates and geographies
• Enable cost-effective screening to prioritize reservoirs for detailed surveys
• Support decision-making for mitigation interventions where they are most needed

Note: RECLAIM is designed as a screening tool, not a replacement for field surveys. It helps guide where and when to act first.

Installation

Install the package via pip:

pip install pyreclaim

Download Data

To generate features for reservoirs using the RECLAIM framework and the pyreclaim Python package, you will need the global datasets.

You can download all required global datasets from the Zenodo Repository:

Download Global Datasets

These datasets include land cover, soil, terrain/DEM derivatives, and vegetation gain/loss data, which are essential for computing reservoir and catchment features for RECLAIM.

Documentation

The documentation is available here.

Quick Start / Example Workflow

This example shows generating features, loading the pretrained model, predicting sedimentation rates, and evaluating results.

from reclaim.generate_features import create_features_per_row
from reclaim.reclaim import Reclaim

# Step 1: Generate features for a reservoir
reservoir_static = {
    "obc": 150.0,
    "hgt": 45.0,
    "mrb": 4030033640,
    "lat": 25.6,
    "lon": 81.9,
    "reservoir_polygon": reservoir_polygon,
    "aec_df": aec_df
}

catchment_static = {
    "ca": 1200,
    "dca": 50,
    "catchment_geometry": catchment_geom,
    "glc_share_path": "data/glc.nc",
    "hwsd2_path": "data/soil.nc",
    "hilda_veg_freq_path": "data/veg.nc",
    "terrain_path": "data/terrain.nc"
}

features = create_features_per_row(
    reservoir_static_params=reservoir_static,
    catchment_static_params=catchment_static,
    observation_period=[2000, 2020]
)

# Step 2: Load pretrained RECLAIM model
model = Reclaim()
model.load_model()  # loads XGBoost, LightGBM, CatBoost, and metadata

# Step 3: Predict sedimentation rates using ensemble
pred_sr, weights = model.predict(features, return_weights=True)

# Step 4: Inspect predictions
print(pred_sr)
print(weights)

# Step 5: Evaluate (if ground truth available)
y_true = [...]  # replace with true sedimentation rates
metrics = model.evaluate(features, y_true)
print(metrics)  # {'RMSE': ..., 'MAE': ..., 'R2': ...}

Citation

If you use RECLAIM in your work, please cite:

Minocha, S., Hossain, F., Zhao, J., & Istanbulluoglu, E.
RECLAIM: A Globally Scalable Machine Learning Framework to Predict Reservoir Sedimentation and Capacity Loss from Satellite-based Earth Observations (Submitted to Environmental Modelling and Software (EMS)).

License

RECLAIM 3.0 is distributed under the GPL v3 license. You may copy, distribute, and modify the software as long as you track changes/dates in source files. Any modifications or software including GPL-licensed code must also be made available under GPL along with build & install instructions.
For more information, see LICENSE.