Size of the study area

Hi Mam/Sir,

I want to know what is the maximum and minimum study area size that we can use for any InVEST model.
Whether InVEST model computes good for area around 30 sq km. Is there any optimum size of study area that the model works effective?
May I also know whether we can use administrative boundary for Annual Water Yield model and for NDR and SDR model.

Thank You.


There are a few aspects to study area size, and different models will be a bit different, so there’s not a simple answer to your question.

Technically, and in general, InVEST models can handle a wide range of study areas, mainly limited by the size of the input data that you provide, because of computer resources. Most of my experience is with the freshwater models, which do computations at the same resolution as the DEM. For these models, I’ve found that having a DEM that’s more than 1-2GB in size starts to be a problem for my laptop. And of course, they will take increasingly long to run.

We have run many of our models at a variety of scales, from small watersheds or administrative units to countries to globally. And in general, the larger the scale of the area of interest, the coarser the resolution of the input data needs to be for it to be computationally reasonable.

At the small end of the study area scale, we would want to use higher-resolution data, since coarse results won’t be very meaningful. However, most of these models were made to provide only a good first estimate of results, so use general methods that aren’t necessarily created for fine detail. For example, the Annual Water Yield model results are best used at the sub-watershed scale, and I don’t think you’d have many sub-watersheds (or at least many of a significant size) in a 30 sq km area.

For a 30 sq km area, I would definitely say that it depends on the model. While Annual Water Yield might not be made well for that kind of area, the Carbon Sequestration model would work fine, as would some others. That’s probably on the low end for NDR and SDR, but they’d probably work ok, since their main outputs are at the pixel scale, or aggregated to the whole area of interest for assessing water quality at some downstream point.

Regarding using an administrative boundary, that is fine for Annual Water Yield because it does not route water down slope across the landscape. But NDR and SDR do route nutrient and sediment, so we should use watershed boundaries to define the area of interest. Otherwise, we’ll be missing areas of the watershed outside of the administrative boundary that may be contributing significantly to the nutrient or sediment arriving at a stream, and the results will be incomplete, and potentially misleading.

~ Stacie