I am assessing the impact of land use changes on some ES. I have LULC maps for present and future (optimistic and pessimistic) scenarios, reflecting different levels of environmental law compliance.
Regarding the coastal vulnerability model, I am uncertain if only the vegetation from the pixels directly touching the ocean influences the model outputs, or if pixels at a greater distance from the coast also affect the model.
I aim to calculate the gains and losses of vegetation along the coastline in different scenarios to understand the impact of land use changes on coastal vulnerability.
I am running the model with a 30 m resolution. If I create a 30 m buffer along the coastline to extract the LULC, would be enough? Or should I use a wider buffer?
You can include vegetation at greater distances from the coastline, both into the ocean, and inland. For each vegetation type, you’ll specify a distance over which it provides protection. So the model will look in a radius of that distance around the vegetation type to see if/where it intersects the shoreline, and if so it will provide some sort of protection to the places it intersects.
We often include habitats such as coastal wetlands and coastal forests, which can extend a bit inland. I wouldn’t go too far inland though, since the model is really focused on the effects to the shoreline itself, and forest that’s 10km inland isn’t going to protect a community located right on the coast.
I’ve used inland vegetation within about 1km of the coast for a few projects, but it would be worth looking at other studies that have used this model to see what distances they’ve used for including land-based habitat types. You can check out both NatCap studies and wider InVEST community studies from the Publications page of our website.
On the topic of resolution, 30m sounds rather small to me. What is your total area of interest? If it’s large, the model will take a very long time to run at 30m. And given the simplicity of the model, it’s best not to use it at too fine a resolution, so we don’t give the impression that it is providing more specific detail about a particular place than it really is. The smallest resolution I’ve seen it used in our work is more like 100m. It can work at 30m, and it’s fine to run it at that scale, just be aware of its limitations when you’re reporting results and making decisions.
Thank you for your elaborate response. So, we have decided to create a buffer along the coastline using “protection distance (m)” values from the Habitats Table for each native vegetation class. For example, we calculated forest and mangrove change between scenarios considering a 2000 m buffer along the coast and 100 m for the beach.
Our study area is the state of Rio de Janeiro, Brazil, covering around 43,500 km². We recognize that the 30 m resolution may not be useful for this coastal protection model, but we are also running other InVEST models and using some refined inputs, such as 30 m land use maps. We ran the model using 30m resolution and it takes around 4 h. In any case, we will test running the model at 100 m resolution to assess its impact on the results.