Dataset and Scripts for: Deep Learning and Stochastic Mapping Framework for Continuous Reservoir Capacity Monitoring using Multi-Mission Satellite Data: The Case of Poechos, Peru

# Dataset and Scripts for: Deep Learning and Stochastic Mapping Framework for Continuous Reservoir Capacity Monitoring using Multi-Mission Satellite Data: The Case of Poechos, Peru **Authors:** Juan Carlos Breña Aliaga (1,*), Joel Cruz Machacuay (2), Jorge Luis Breña Ore (3), Diego Alonzo Tacusi Cuadros (4), and Daniel Marcelino Rosales Hurtado (2) **Affiliations:**(1) Department of Water Resources, Universidad Nacional Agraria La Molina (UNALM), Lima 15012, Peru(2) National Water Authority (ANA), Lima 15036, Peru(3) Faculty of Chemical and Textile Engineering, National University of Engineering (UNI), Lima 15333, Peru(4) National Service of Meteorology and Hydrology of Peru (SENAMHI), Lima 15072, Peru*Correspondence: [e-mail] **Related Publication:** This dataset is associated with the manuscript *"Deep Learning and Stochastic Mapping Framework for Continuous Reservoir Capacity Monitoring using Multi-Mission Satellite Data: The Case of Poechos, Peru"* (Currently under peer review). --- ## 1. OverviewThis repository contains the complete dataset, high-resolution reference masks, deep learning models, and Python scripts required to reproduce the methodology and results presented in the associated research article. The framework reconstructs the Elevation-Area-Volume (EAV) curve of the Poechos Reservoir dynamically, resolving the orbital asynchrony between Sentinel-1 SAR imagery and SWOT altimetry using an FPN-InceptionV4 segmentation ensemble and Monte Carlo stochastic quantile mapping. --- ## 2. Repository Structure The repository is organized into 6 compressed folders (`.zip`), corresponding to the sequential execution of the methodology: * **`1_Scripts_and_Code.zip`** Contains the Jupyter Notebooks required to execute the full pipeline. * `01_PlanetScope_GroundTruth.ipynb`: NDWI and Bmax-Otsu thresholding for reference masks. * `02_SAR_Preprocessing.ipynb`: Sentinel-1 cropping, co-registration, and tensor normalization. * `03_DeepLearning_Segmentation.ipynb`: FPN+InceptionV4 training, cross-validation, and continuous area inference. * `04_Stochastic_Quantile_Mapping.ipynb`: SWOT data filtering, datum correction, Monte Carlo simulation, and EAV curve integration. * **`2_DeepLearning_Dataset.zip`** Contains the training/validation data for the deep learning models. * `Sentinel_1_Tensors/`: 256x256 normalized patches (VV, VH, VV-VH) derived from Sentinel-1 GRD imagery. * `PlanetScope_Masks/`: Corresponding 256x256 binary ground-truth masks (3m resolution). * **`3_Trained_Models.zip`** Contains the PyTorch state dictionaries (`.pth`). * `All_CrossValidation_Models/`: The 45 models evaluated during the 5-fold cross-validation. * `Best_Ensemble_FPN_InceptionV4/`: The optimized weights for the selected architecture (Threshold = 0.64). * **`4_SWOT_Altimetry.zip`** * `SWOT_Validated_Observations.csv`: The 53 validated Water Surface Elevation (WSE) records (Passes 035 and 188) filtered via the 7-criteria quality control and corrected to the local OLSA datum (-8.343 m). * **`5_InSitu_Validation_Data.zip`** * `ANA_Operational_Volumes.csv`: Daily historical reservoir storage records reported by the Proyecto Especial Chira-Piura (PECHP) to the National Water Authority (ANA), used for global volumetric validation and bias calculation. * **`6_Study_Results_and_Metrics.zip`** Contains the tabular data supporting the article's figures and conclusions. --- ## 3. System Requirements and Dependencies To execute the scripts locally or on cloud platforms (e.g., Google Colab), the following core libraries are required:* Python >= 3.9* PyTorch >= 2.0* Segmentation Models PyTorch (`segmentation_models.pytorch`)* GeoPandas, Rasterio, NumPy, Pandas, SciPy --- ## 4. Usage Instructions 1. **Environment Setup:** Install the required dependencies. If using Google Colab, mount your Google Drive or upload the unzipped folders to the local runtime.2. **Path Configuration:** Update the relative file paths in the `01` to `04` Jupyter Notebooks to point to the unzipped directories (`2_DeepLearning_Dataset`, `4_SWOT_Altimetry`, etc.).3. **Sequential Execution:** Run the notebooks in numerical order (01 -> 04) to replicate the study from raw tensor processing to the final EAV curve generation.--- ## 5. License and Citation This dataset and code are provided under the Creative Commons Attribution 4.0 International (CC-BY 4.0) license. If you use these materials in your research, please cite the corresponding peer-reviewed article. (Citation details will be updated upon final publication).