Applying InVEST to agri-environmental schemes

Hi there,
I am considering using the InVEST model to model changes in ecosystems on agricultural land, specifically in regards to agri-environmental measures. For example, I would perhaps want to compare changes in farming practices such as flower strips, cover cropping, etc. to see how this alters ecosystem service provision when compared to a baseline. I haven’t found any research using InVEST within agriculture, and I am not sure if this is possible or not with this model. Any insight as is appreciated. Thanks!

Hi @michelle.vanhilten -

Which InVEST model(s) are you thinking about using? Each one will have a different way of handling agriculture. The freshwater models in particular (SDR, NDR, annual and monthly water yield) are often run to compare a baseline versus scenarios of different agricultural management practices, to see how they impact water quality or quantity.

One of the inputs to these models is a land use/land cover map, which can include agriculture, along with natural and developed land cover types. There is also a table of model-specific values that differentiate, for example, vegetation cover, root depth, fertilizer input, etc, and this table is where we can provide different values for current ag practices, versus cover cropping or others.

As one place to start, I can recommend looking through our Publications Library, whose Search function is unfortunately still rather limited to paper titles and authors, so a bit difficult to find things, but if you search for “management” or “sediment” etc, you will hopefully find some related papers. Or, if you give us more of an idea of the models you’re interested in, we might be able to point you to specific projects or publications.

~ Stacie


Hi Stacy,
Thank you for this information. More specifically I am considering the carbon sequestration, habitat quality, and/or one of the freshwater models, but I am open to others. I am interested to compare a conventional agricultural region (industrial) as a sort of business as usual or baseline, and then run a scenario where improved management would be implemented, for example cover cropping, in field/field edge diversity, etc. I would also be interested to know about synergies between ecosystems services, if a practice implemented to enhance one service also enhances another, or vice versa with trade offs.

What I am a bit confused or not was whether or not it’s feasible to focus on agriculture specifically, and not necessarily a land use change. The studies I’ve read all accounted for a change in land use but I just want to know how adjustments with in agriculture affects ecosystem services, if that makes sense.

Thanks again for the feedback/insight. I will absolutely use your recommendation for searching, indeed I struggled with finding appropriate articles earlier.

Kind regards,

Hi Michelle -

I’ll say a few notes about agriculture with the different models you mentioned.

The Carbon model is extremely simple, it simply maps a table with 4 carbon pool values to different land cover classes and adds them up. Depending on the agricultural management activities you’re considering, it’s likely that the main change would be in soil carbon, in which case you could do that analysis manually by starting off with a map of current soil carbon, then adjusting based on whatever you learn/know about how the practices affect soil carbon content and where they are done. If you’re also considering something like agroforestry, then the other carbon pools could also come into play, and the model might offer some convenience for doing those calculations.

For habitat quality, I’d imagine that the main model adjustment would be regarding how suitable the ag areas are for your species of interest. Again, this could be as simple as a change in the input tables required by the model, not necessarily a change in the land cover map.

But, you may want to apply one management practice to one ag area or crop type, and a different practice to another, and in that case you may end up altering the land use map, to distinguish, say, perennial tree crops where interplanting is done, from annual crops where a winter cover crop is done. Of course, this depends on how agriculture is defined in the land cover map you’re starting with. All of these models require a land cover map as main input. You may end up only changing the biophysical table values associated with that land cover map (for a change in carbon pools, habitat suitability or nutrient application), and that’s ok, that works well for creating scenarios. It’s also fine to only change the ag areas and leave everything else static between scenarios.

One thing of note for the water models is that we recommend running them on an entire watershed, so not, say, only on one field, or only on the ag areas within an area of interest, because of how landscapes are hydrologically connected.

~ Stacie


Hi Stacie,
Thank you so much for this insight. This is really helpful in guiding which ecosystem services modules to consider and perhaps which conservation practices would work with the modules based on the required data. I do like the idea that the habitat quality, nutrient, and carbon modules would only require changes in the data tables if I wanted to model different management practices, it seems this may be simpler than creating new maps.

An additional question for you, as I would focus on management practices withing agricultural land consistently and no land use change would occur, to account for changes to the agricultural land would I then just have different levels of agricultural land associated with different lucode classes? As opposed to having different land use types with different lucode classes. This is referring to the LULC class CSVs that are required for data needs. The examples in the user manual always show 1 Forest, 2 Coffee, 3 Pasture, etc.

Thanks again for your help it has been incredibly insightful.

Hi Michelle -

If you have different levels of agricultural land, you will need each of those to have a distinct land cover class and integer ID Value in the land cover map. The models map biophysical table values to the land cover map via the ID numbers, so you must have one assigned to each class in both the map and the table. For example, if you want to map industrial corn and permaculture corn, you will need to have two different land use classes defined in your land cover map, each with a different ID Value. So, along the lines of the User Guide example, you might have raster value 30 is Industrial Maize and raster value 31 is Permaculture Maize. Then you’ll have a biophysical table where “lucode” value 30 has model parameters specific to management practices on Industrial Maize and “lucode” value 31 has parameters specific to Permaculture Maize. I recommend looking at the sample data that comes with InVEST and you’ll see how it works.

It may be that your original land cover map already has these distinctions. If not, or if you want to try different configurations than what is already in the land cover map, then you’ll need to use GIS tools to define which places are industrial corn and which places are permaculture corn and assign the correct ID values.

~ Stacie

1 Like