Spatial Modeling: A Chess Board Analogy

This week was a quieter week at the lab, with lots of tasks running in the background, while we prepare ourselves for future endeavors. However, since this week was quieter, I thought I would spend this week's blog explaining the spatial component of our ecosystem model, and do so in a way that is comprehensible, regardless of your academic background. For reference, I've created this chess board figure, which does a nice job of breaking down the spatial component of our modeling efforts.

Our first step - after defining a geographic space of interest - is to create a grid structure within the model domain. These grid cells represent the spaces that all the organisms in the model can inhabit, move into, and move away from, as shown by the arrows in the figure. The model divides the amount of animals for each model group equally across each grid cell, regardless of if the animal can survive in that initial space. Take for instance the spotted seatrout I have pictured here. These animals are mostly nearshore, coastal animals, using estuaries - possibly seagrass meadows - for nursery habitat. Therefore, even though they prefer nearshore areas, some spotted seatrout will get assigned to open ocean habitat at the start of the model run - perhaps the bottom left corner of the chess board here. As the model runs through time, the organisms can move to the nearby grid cells through swimming or drifting. Imagine these organisms doing a line dance and moving around the chess board to find a better space. For our fixed organisms, like oysters and seagrasses, they cannot move to nearby grid cells and must try to tolerate the local conditions. In our model, we provide new environmental conditions every month, which inform the suitability of each grid cell to each model group. Again, those that can avoid harsh conditions will attempt to escape, while those that are not mobile may perish. 

While organisms at the start of the model run will be evenly dispersed among the grid cells, soon we will see distributional patterns, where some areas appear most suitable for groups of organisms. Take our oysters, for instance; we can include a oyster reef locations in our grid, which we know are the most suitable places for oysters, and the individuals that do not start on those reefs will likely die off, while those on the reefs can survive and reproduce. For some of our mobile organisms, we may see them move around a lot within these grid cells, expending energy to find the most suitable habitats. 

The two other items at play in this spatial modeling component are the fisheries and the changes in the land. While the fisheries are slightly more difficult to depict in this visual, we have fishing fleets that remove organisms from the model - one way of maintaining the balance in the ecosystem. We know how much fish the boats are catching due to data reporting and logging efforts, so we aren't removing animals artificially. We can also include changes in land cover in the spatial modeling efforts. Here, imagine that over time the chess board shape changes ever so slightly, where the bottom left white square starts to become multicolor. While not a perfect analogy, over time the geography of our model domain can change where we have land where land previously wasn't, or where we have loss of marsh habitat, like we are seeing in coastal Louisiana. We can include changes in the shape of our model domain over time, which can influence the suitability of the habitat for our organisms. Perhaps the largemouth bass prefer marsh habitat and decline when marsh cover decreases, or perhaps the dolphins are unaffected by changes in the land if they are more offshore swimmers and foragers. For all these aspects of our model, the software uses mathematics to analyze how environmental and physical changes affect the organisms, without requiring days of computing. I'm always amazed by how quickly the model runs, especially since computing power and computing time was an often discussed aspect of my graduate school modeling coursework. 

I hope this snapshot into the spatial modeling component of my work helps you understand the big picture of what we accomplish with our software. As I've mentioned before, I can't share outcomes, results, a lot of details about this work, as it is still ongoing. However, once we have final results and peer reviewed materials, I will be sure to share what we've learned. Stay tuned for another blog next week!

---