Spatial cellular model for mangroves

Simulate mangrove response to sea-level rise.

Load aligned raster data, define the classes and scenario, and inspect how each cell changes year by year in BR-MANGUE Studio.

Download for Windows or Linux Read the guide

Windows 10/11 stable · Linux x86_64 research preview · No separate Python installation

Real model transition Ilha de São Luís · 75 steps

Animated BR-MANGUE Studio demonstration run on Ilha de São Luís from 2025 to 2100

Demonstração real gerada pelo Studio: 1.083.556 células, de 2025 a 2100. O exemplo mostra o funcionamento do modelo e não é uma previsão da ilha.

What the Studio does

One workflow from raster input to annual results.

The application makes the modelling decisions visible: you can inspect the source classes, coordinate reference, parameters, and transitions before interpreting a map.

Cellular rules

Each cell evaluates its elevation, state, and up to eight neighbours to determine the next annual state.

Two processing engines

Use continuous processing when the grid fits in memory, or persistent blocks for larger domains and recoverable runs.

Inspectable outputs

Review annual maps, trajectories, class changes, transition tables, CSV data, GeoTIFFs, and animation frames.

Animated example of annual mangrove state changes — Fictitious data from the coastal zone of Maranhão State, Brazil. This example is for illustration, not a forecast.

See the model in motion

Water level, elevation, and land-cover state meet in one map.

Use the animation to check the direction of change, then open the annual tables and figures to quantify it.

How to run a scenario →

Performance

Choose the engine for your grid.

Continuous processing is usually faster when the active raster fits safely in RAM. Persistent blocks trade speed for capacity and recovery on very large coastal domains.

Continuous~0.631M updates/sReference run: 78.01 million cell-updates in 123.6 s, with the active grid kept in memory.

Persistent blocks~0.164M updates/sReference run: 78.01 million cell-updates in 476.7 s using 10,000-cell chunks and persistent state.

Same model state0 cell differencesFinal maps and annual trajectories matched when both engines used the same inputs and settings.

The reference area is the Costa de Manguezais de Macromaré da Amazônia (CMMA), at 30 m, with a 2024 land-cover map and ANADEM digital terrain model. Read the benchmark note →

Desktop builds

Download BR-MANGUE Studio 1.0.0.

The stable Windows release supports Windows 10/11, 64-bit. A Linux x86_64 research preview is available; it was built on Ubuntu 22.04 and launched on Ubuntu 26.04.1 LTS. See the download page for tested systems and compatibility notes.

Open the download page

Start with three filesLand cover · elevation · optional suitabilityPrepare aligned GeoTIFFs, map the classes, set the scenario, and run the model.

Project context

From a research model to a practical tool.

The first BR-MANGUE version was developed between 2010 and 2014 in Lua with TerraME by Dr. Denilson da Silva Bezerra. BR-MANGUE Studio reimplements the model in Python as an integrated desktop application developed at the Federal University of Maranhão by MSc. Felipe Martins Sousa, in partnership with the original author.

Research softwareVisible inputs.
Reproducible outputs.Documentation, benchmarks, and release files are available on this site.

Documentation

Start with the guide.

Find the input requirements, class mapping, coordinate reference, processing engines, outputs, and troubleshooting steps in one place.

Studio workflowProjects, mapping, simulation, and outputs↗ Raster input requirementsAlignment, masks, and optional suitability↗ Processing enginesContinuous versus persistent blocks↗ Audit checklistRecord inputs, rules, and scenario parameters↗