rodrigo amado garcia silva - eventos fgv · coastal and oceanographic engineering area (ocean...

ADVANCES IN NUMERICAL MODELING AND DATA

ANALYSIS ON WATER RESOURCES RELATED ISSUES

Rodrigo Amado Garcia Silva

Contact: rodrigoamado@oceanica.ufrj.br

SUMMER SCHOOL ON DATA SCIENCE

Coastal and Oceanographic Engineering Area (Ocean Engineering Program)

➢ Research Engineer (postdoc)

➢ Issues: Environmental hydrodynamics, wave generation and propagation, sediment

transport, coastal vulnerability, water quality

HidroAmb – Water Resources and Environmental Engineering

➢ Director and Environmental Engineer

➢ Issues: the same

Water resources related issues

➢ Environmental Hydrodynamics

➢ Wave generation and propagation

➢ Sediment transport

➢ Pollutant transport and dispersionHydrodynamics

Morphodynamics

Wave propagation

Effluent dispersion

Water resources related issues

➢ Environmental Hydrodynamics

➢ Wave propagation

➢ Sediment transport

➢ Pollutant transport and dispersion

ALL NONLINEAR PHENOMENA!!

Water resources related issues

➢ Environmental Hydrodynamics

➢ Wave propagation

➢ Sediment transport

➢ Pollutant transport and dispersion

ALL NONLINEAR PHENOMENA!!

Simulation requires numerical solution

of nonlinear differential equations

➢Computational modeling systemprovided by COPPE/UFRJ

➢Environmental modeling of waterbodies with complex geometry, asrivers, estuaries, lagoons, bays, coasts,reservoirs, etc.

➢Open code (Fortran) free software

➢Finite elements and finite differencenumerical models

SisBaHiA – Environmental Hydrodynamics Base System

How does it work?

Example: Ilhabela – SP

Santos

Ilhabela

Bertioga

SP

How does it work?

Example: Ilhabela – SP

➢ Digital terrain model

Santos

Ilhabela

Bertioga

SP

Finite elements mesh

Bathymetry data

How does it work?

Example: Ilhabela – SP

➢ Digital terrain model

How does it work?

Example: Ilhabela – SP

Simulated phenomena: wind waves generation

Input data:

➢ Bathymetry (DTM)

➢ Winds

➢ Boundary conditions

Wind field

Open boundary(ocean)

Land boundary(zero flux)

How does it work?

Example: Ilhabela – SP

➢ Results

Wave height (m)

How does it work?

Example: Ilhabela – SP

➢ Results

Wave period (m)

How does it work?

Example: Ilhabela – SP

➢ ResultsWave inside the harbour

Data analysis

Bathymetry data

Sea bottom surface

Surfer(2D and 3D mapping)

Interpolation methods

• Kriging• Delaunay triangulation• Minimum curvature• Natural neighbor• etc

Google Earth(model domain)

Georeferenced map layers

Data analysis

• Wind data acquisition

Wind dataanalysis

Grapher(2D and 3D graphing

and analysis)

Global athmospheric models

• Wind data

• NETCDF format

• Python interface

Results analysis

Model setup and execution

Microsoft Access(or any similar database tool)

Grapher(temporal results)Surfer

(spatial results)

➢ Nearshore morphological processes

➢ Wave induced sediment transport

Research and Development

Mucuripe

Port

Fortaleza - CE

Coastal erosion

Port sedimentation

Pereira et al. (2017)

NMMBom

tempo

Perfil de bom tempo

Beach erosion

NMMBom

tempo

Perfil de bom tempo

Beach erosion

NMMBom

tempo

Perfil de bom tempo

Beach erosion

NMMBom

tempo

Perfil de bom tempo

Beach erosion

NMMBom

tempo

Perfil de bom tempo

Beach erosion

Berma de bom tempo

Duna

Beach erosion model

Beach Profile

water level

Model result

Measured

Initial

Wave tank

Berma de bom tempo

Duna

Beach erosion model

Beach Profile

water level

Model result

Measured

Initial

Wave tank

Duna

Beach erosion model

➢ Good results

Berma de bom tempo

Beach Profile

water level

Model result

Measured

Initial

Duna

Beach erosion model

➢ Good results

➢ Highly dependent on case to case calibration

Berma de bom tempo

Beach Profile

water level

Model result

Measured

Initial

Duna

Beach erosion model

➢ Good results

➢ Highly dependent on case to case calibration

➢ Several parameters involved

Berma de bom tempo

Beach Profile

water level

Model result

Measured

Initial

Duna

Beach erosion model

➢ Good results

➢ Highly dependent on case to case calibration

➢ Several parameters involved

➢ Nonlinearity expressed in the data

Berma de bom tempo

Beach Profile

water level

Model result

Measured

Initial

Duna

Beach erosion model

➢ Good results

➢ Highly dependent on case to case calibration

➢ Several parameters involved

➢ Nonlinearity expressed in the data

➢ Upcoming research: how to find a pattern?

Berma de bom tempo

Beach Profile

water level

Model result

Measured

Initial

Duna

Beach erosion model

➢ Good results

➢ Highly dependent on case to case calibration

➢ Several parameters involved

➢ Nonlinearity expressed in the data

➢ Upcoming research: how to find a pattern?

➢ Machine learning

Berma de bom tempo

Beach Profile

water level

Model result

Measured

Initial

Duna

Machine learning

Several methods:

➢ Artificial neural networks;

➢ Genetic algorithms;

➢ Bayesian networks;

➢ Regression trees;

➢ etc

Goldstein et. al (2019)

Thanks a lot!

Rodrigo Amado G. Silva

rodrigoamado@oceanica.ufrj.br