Human impact modelling typically seeks to assess how strong the cumulative impact (Fig. 1.) of anthropogenic ecological stressors is on the ecosystem, i.e. where nature is experiencing stress due to human activities. Impact modelling might in some cases assess the impact of only one specific stressor or activity; in other cases the ecosystem and its components might not be included at all in the analysis, resulting in a product that estimates the amount of stressors present but not their impact.
The process of human impact modelling can roughly be divided into two stages: data processing and analysis. In the first stage, geographic information system (GIS) software like ArcGIS and spreadsheet programs like Microsoft Excel are usually the main tools used to process ecosystem and human activities data, with custom scripts used as support (though scripts can also be used as a main tool). In the second stage the processed data is entered into a program or script that calculates the cumulative impact.
Typically, the ecosystem is represented by a collection of ecosystem components, each as a separate spatial dataset. Examples of ecosystem components are seabed habitats or different mammal and fish species. Ecosystem components can also in turn be aggregates of several datasets or be composed of only a single dataset. If we take whale occurrence data as an example: the occurrence of all whale species could be aggregated into one ecosystem component, or they could be divided into a couple of ecosystem components based on specific species traits or taxonomy, or each whale species could be left as a separate component. What is the best approach depends on things such as data quality, scale, objectives, computer processing power, and time.
Human activities are often divided into stressors, or pressures, much in the same way as the ecosystem is divided into ecosystem components. A single human activity, ship traffic for example, can cause several different types of stress on the surrounding ecosystem. Thus, the ship traffic dataset might be processed into two or more stressor layers, one maybe representing the underwater noise caused by ship machinery and propellers, the other representing the re-suspension of sediments caused by ship passage. Furthermore, these stressor layers might be aggregated with other stressors of the same type, combining all noise stressors into one dataset for example.
The first step when preparing and processing raw human activities data is usually the conversion data into the GIS format needed for further processing. The desired format in this stage might be raster, polygon, point or line, depending on the activity, scale, data quality, or stress it produces. The simplest way of assessing the extent and intensity of a pressure is to take the spatial data of the corresponding activity, and calculate how many times that activity is present in each location (raster cell). Another simple method is creating a buffer (Fig. 2.) around each feature (activity) to simulate the spatial extent of the pressure caused by the activity.
Simple methods for processing data are in some cases perfectly suitable, for example when calculating how much of the seabed surface is lost from dredged material being deposited there; other times, however, more advanced methods and tools are needed to estimate the extent and intensity. Advanced methods might take into consideration the diminishing effects of a stressor when the distance to the source of the stressor increases or when there are obstacles in the way, weakening the propagation of an expanding stressor like underwater noise. Advanced methods might also account for varying intensities within an activity, for example by considering ship size and speed when calculating physical disturbance caused to the seabed. With ArcGIS, these phenomena can be accounted for using tools such as ring buffer, focal statistics, Euclidean distance, cost path, viewshed, zonal statistics, and many others.
Ecosystem data processing can require similar methods as stressor data processing, but is often more straightforward due to the stationary and present/absent nature of many ecosystem components, such as substrate type or seagrasses. Typically, the majority of ecosystem components are presented as presence/absence data (1/0), while stressor data tends to be a fairly balanced mix of presence/absence and continuous (0.01 – 1.00) data. For both ecosystem components and stressors the final processed data is almost always in raster/grid format. Expert opinions and scientific literature are the main sources when deciding how a dataset should be processed, for example when determining the size of buffers used to simulate the spatial extent of a pressure.
When all datasets are processed, the actual cumulative impact analysis can finally be run. All datasets are fed as input into the software or script chosen to run the analysis. Software suitable or meant for human impact modelling includes ArcGIS, Zonation, and ImpactMapper. At this point, so called sensitivity scores or “weights” (Fig. 4.) are also entered into the software; these determine how strong the effect of a specific stressor is on any specific ecosystem component when they intersect. Sensitivity scores are typically based on expert opinions.
The main final product is usually a quantitative map that shows cumulative impact, i.e. where and how much the ecosystem is under stress. Typically, the more ecosystem components and stressors there are present in the same area, the higher the cumulative impact value will be.
Human impact modelling is a concept that is still in a relatively early stage of development, especially when the focus is on the marine environment. New methods are constantly being developed, better data is becoming more available, and computer processing power is steadily rising, leading to new opportunities almost every year!
Written by Marco Nurmi, Finnish Environmental Institute