U-net Flood Depth Mapping Project geo-spatial expert
Budget: $30 – $250 USD
I'm seeking a geo-spatial expert with U-net experience for detailed flood depth mapping.
Study Area:
- South Punjab Region, Pakistan
Data Sources:
- Satellite imagery (Sentinel 1 Data & Sentinel 2 Data)
- DEM
- Flood extent maps (2015-2023)
Final Deliverable:
- Analysis results formatted as Raster images (TIFF)
- Flood depth maps
- Validation and accuracy assessment
Ideal candidates should be proficient in U-net and geo-spatial data analysis, with experience in interpreting satellite imagery and historical flood data. Knowledge of topographical maps and their relation to flood depth is crucial.
Expected Project Outcome
Data Preparation:
Pre-processing: Ensure all data sources (satellite imagery, topographical maps, DEMs, and flood extent maps) are aligned and pre-processed correctly. This may involve resampling, normalization, and ensuring consistent spatial resolution.
Annotation: For training the U-Net model, you’ll need annotated data showing flood depths. This may involve creating ground-truth labels based on existing flood extent maps or expert annotations.
Model Development:
U-Net Architecture: Set up and customize the U-Net architecture to handle your specific data and task. This involves configuring the model layers, loss functions, and optimization techniques.
Training: Train the U-Net model using the prepared dataset. Monitor the training process and adjust parameters as needed to improve accuracy and performance.
Flood Depth Mapping:
Inference: Use the trained U-Net model to predict flood depths from new satellite imagery and DEMs.
Post-processing: Refine the model outputs to enhance clarity and accuracy, including filtering, smoothing, or correcting artifacts.
Output Formatting:
Raster Images: Export the flood depth maps as TIFF files. Ensure the final output is georeferenced correctly and aligns with your spatial data requirements.
Validation and Analysis:
Verification: Validate the flood depth maps against known flood events or ground-truth data to assess accuracy.
Analysis: Perform any additional analysis required to interpret the results or integrate them into broader flood management systems.
Study Area:
- South Punjab Region, Pakistan
Data Sources:
- Satellite imagery (Sentinel 1 Data & Sentinel 2 Data)
- DEM
- Flood extent maps (2015-2023)
Final Deliverable:
- Analysis results formatted as Raster images (TIFF)
- Flood depth maps
- Validation and accuracy assessment
Ideal candidates should be proficient in U-net and geo-spatial data analysis, with experience in interpreting satellite imagery and historical flood data. Knowledge of topographical maps and their relation to flood depth is crucial.
Expected Project Outcome
Data Preparation:
Pre-processing: Ensure all data sources (satellite imagery, topographical maps, DEMs, and flood extent maps) are aligned and pre-processed correctly. This may involve resampling, normalization, and ensuring consistent spatial resolution.
Annotation: For training the U-Net model, you’ll need annotated data showing flood depths. This may involve creating ground-truth labels based on existing flood extent maps or expert annotations.
Model Development:
U-Net Architecture: Set up and customize the U-Net architecture to handle your specific data and task. This involves configuring the model layers, loss functions, and optimization techniques.
Training: Train the U-Net model using the prepared dataset. Monitor the training process and adjust parameters as needed to improve accuracy and performance.
Flood Depth Mapping:
Inference: Use the trained U-Net model to predict flood depths from new satellite imagery and DEMs.
Post-processing: Refine the model outputs to enhance clarity and accuracy, including filtering, smoothing, or correcting artifacts.
Output Formatting:
Raster Images: Export the flood depth maps as TIFF files. Ensure the final output is georeferenced correctly and aligns with your spatial data requirements.
Validation and Analysis:
Verification: Validate the flood depth maps against known flood events or ground-truth data to assess accuracy.
Analysis: Perform any additional analysis required to interpret the results or integrate them into broader flood management systems.