[PITCH]Using EO-Data to detect deforestation of mangroves causing erosion

[angelasindic]

Detect deforestation of mangroves causing erosion with EO-data

The mangrove ecosystem is one of the most productive habitats that support many marine species, help mitigate climate change, and provides natural livelihoods for coastal communities.

Despite this, anthropogenic activities are rapidly degrading and deforesting mangroves worldwide. Additionally, other stressors due to climate change, such as rising sea levels, are increasing rapidly.

To better protect this habitat, it is crucial to map, manage, and monitor this ecosystem.

We can use remote sensing to derive Spatio-temporal information on mangrove forests distribution, forest density, expansion, and contraction in the local extent of mangrove forest coverage.

What would a hack team for this project work towards in 2 days?
First of all, we have to research the most suitable deforestation indicators, how to derive them from satellite imagery, and determine which data products provide the required spectral features.

We can either use those data products directly or calculate a deforestation index to quantify deforestation per image/time frame for the next step.

Alternatively, we could also use a LULC classifier to achieve this, which would require model tuning and training.
We can create a deforestation index over a certain period, aggregate a time series that can be further analyzed (trend analysis, etc.)
Finally, we can correlate forestation changes with erosion events happening after heavy rainfall or other weather conditions.

How would this project use Planet's data & platform?
We can use the planet platform to relatively easy access multispectral sensing data for a chosen area of interest and period. It also supports a quick visual investigation to select time frames we want to use for model creating. It also provides a programming environment to implement a deforestation index and develop the LULC model.

What obstacles or blockers do you think this project might run into?
Choose the best features/bands to detect mangrove deforestation. We probably won't have in-situ data to validate and improve feature selection.
Time and resources to train a land use/land cover model.

And finally... pitch your idea: why should people hack on THIS project in particular?

---

Be innovative, be creative, make a difference: combine remote sensing with deep learning and image processing to help the mangroves save us.

If you're interested in hacking on this project, add a 👍 reaction to this post.

[rqwang]

Hi angelasindic, I am interested in your project and wonder you need a teammate to go.

[kevinlacaille]

I too am interested, I'd be happy to help :)

[angelasindic]

I'll set up a zoom call at 7:15 pm (CEST) to discuss the idea, please feel free to join:
https://zoom.us/j/99926432931?pwd=NllxcFN0ek9vS1gwWGxnd1RPY0UwUT09

Meeting ID: 999 2643 2931
Passcode: veUd7t