Ecological Impacts of Solar Farms

Background & Objectives

Demand for solar power is growing. North Carolina is currently 2nd in solar power production with large scale solar farms appearing more frequently across the state. What are the unintended environmental consequences of these new solar farms?

Evaluate the environmental impact of solar farm installations (present | future) in southeast North Carolina focusing in on aspects of fragmentation and connectivity.
[A possible other angle on this] Are there configurations of new solar farm installations that would minimize ecological impacts - on fragmentation, connectivity, and perhaps other considerations?

Narrowing in on an actionable project

♦ Define the study area

To make our project more manageable, we'll narrow in on a workable study area. Looking at a single county will be more feasible than looking at all of southeastern NC, and it should be easy to scale up our analyses later on if we should want to.
We will choose Pender County to start off with. Perhaps we'll find reason to change that later, but we'll select this to start.

♦ Refine the study question and data requirements: Fragmentation

To evaluate the "impacts of fragmentation", we need to further define two key elements: what is being fragmented, and how do we measure fragmentation. Our client, Liz Kalies, has offered no specifics on either of these, so we must choose meaningful values for each ourselves.

What landscape features are being fragmented?
- Based on our understanding of how fragmentation may be ecologically harmful, we'll want to select a landscape feature that represents a natural refuge that benefits wildlife or ecological functioning.
- As time is limited, we might narrow our choices to those landscape features for which we can readily find data. Some candidates include: TNC resilient areas, South East Regional GAP land cover (SEReGAP), National Land Cover data (NLCD), National Wetlands Inventory (NWI), or perhaps a derived dataset such as riparian areas. Can you think of any others you might want to use?
How to measure fragmentation?
- We've reviewed a number of means for measuring fragmentation in class, using both ArcGIS and FRAGSTATS. Recall that these measures occur at the patch, class, and landscape levels.
- For these measures to be meaningful, however, we need to put them into context. In our case, the context is evaluating the impacts of solar farms installation, so we'll want to generate the above measures of fragmentation across different landscape scenarios: before vs after solar panel installation, and perhaps scenarios where new solar farms may be likely to go.

From the above, you have a clearer picture of the specific analyses involved and should now begin to assemble a spatial analysis workflow to execute the task.

Data requirements
- First, we'll need the dataset that includes the features on which we'll be evaluating fragmentation. Suggested datasets are listed above, but you are welcome to introduce your own.
- Second, we'll need the datasets required to depict the different scenarios. This will likely include the locations of existing solar farms and something that indicates what was there prior to installation. If the latter is not available, a viable alternative would be to examine a "worst-case" scenario where all lands prior to installation are assumed to be "natural" lands.
- To evaluate a "future" scenario, we will need a way to identify likely locations of new farms. For this, we could engage our habitat modeling skills to identify which parcels have attributes that fall within ranges that favor the installation of a new solar farm. This could be rule based (e.g. it's vacant, within a distance to existing power lines, aspect between X and Y, etc.) or statistically based. (I caution on the latter as that could involve a significant time investment with the possibility of a lousy return, but I encourage the investigation...)
  - Be sure to contemplate the datasets you'd need to run these analyses...

♦ Refine the study question and data requirements: Connectivity

To evaluate connectivity, we'll again need to further define the question: what landscape features are we evaluating, and what determines whether they are connected or not? And once again, these specifics are not provided by our client, so we have freedom to define them ourselves.

What landscape features are we evaluating?
- A logical choice is to select the same landscape features we are evaluating for fragmentation. However, we may also want to consider connectivity to a subset of these locations (e.g. large hubs) or to know conserved areas (e.g. SNHAs). Recall that connectivity analysis is time/CPU intensive, and reducing the nodes we are connecting may be necessary when time is limited.
What determines whether features are connected?
- This can also be asked as, what is the measure of connectivity? Euclidean distance between patches/nodes? Cost distance? If cost, what defines the costs? A cost or distance threshold?
How do we measure the changes in connectivity associated with solar farms?
- As above, this is perhaps best handled by evaluating costs across different scenarios: before solar farms vs with solar farms, or perhaps the connectivity changes associated with additional solar farms, wherever you decide those might go.

♦ Contemplate you analysis and output options

With the main bits out of the way, it's also useful to think about your project more broadly and how it will proceed:

Data analysis
- As mentioned above, you will want to sketch out the logical workflow going from data to results prior to diving in. This will allow you to perhaps foresee any potential flaws in your analysis or data gaps. This is just a sketch, of course, and you can deviate from it later, as needed.
Presentation of results
- You will also want to consider how you will present your results most effectively. Tables? Plots? Maps? Which is the best format to communicate your findings, to answer you initial question, given your intended audience.

♦ What's next?

From the above, you should have a better idea of the tasks at hand. The first step is usually to accumulate the data you need. I find that if you are working in teams, building a shared web map in ArcGIS online can be useful. In fact, I have created on myself, which you are welcome to use: http://arcg.is/jjiyW

Next, you'll want chip away at your analyses. If you get stuck, seek help from your classmates, TAs, and instructor. Remain focused and keep in mind that time is limited. A well formed and presented simple analysis is better than an incomplete and sloppy complex analysis. You may find it necessary to make some sweeping assumptions which is fine as long as you are transparent and allude to how these assumptions might bias your results.