![[Skip Navigation]](/images/pixel.gif){kind=tracking}
|
Benthic Habitat Mapping
The term spatial analysis refers to the process of converting survey and sample data into a coherent benthic habitat map through analysis and integration. Each set of sample data points can be displayed and visualized in a variety of ways. Data collected using different methods and measuring different parameters can be combined to form a more complete picture of a particular benthic habitat. A variety of computer hardware and software packages are available to assist in the analysis of benthic habitat data. The developing science of Geographic Information Systems (GIS) has been well established as a useful tool for benthic mapping applications. One of the most important features of GIS is that it allows researchers to transform benthic habitat data from simple data display to integrated data layers and visualization surfaces. Data Display vs. Data Analysis The difference between simple data display and data analysis is important to understand when creating a benthic habitat map. Once benthic data are collected, results can be easily displayed using GIS. Point data, such as that acquired from single-beam acoustic systems and sediment grabs, provide information regarding the habitat at that particular sampling station. A simple map showing sample locations and the parameters that were measured at specific locations can be helpful, but ultimately it is more useful to analyze the data further and determine the relationship between those data and broader-scale benthic structure. Various interpolation procedures facilitate the extrapolation of point data to larger areas by converting point data to gridded raster data layers. Interpolation techniques recognize and present benthic parameters as spatially continuous features by using point data from sampled stations to predict parameter values at unsampled locations. Interpolation is based on the idea that points that are close to one another in space have more similar characteristics than ones further away. For any given mapping program, sampling takes place at discrete locations within a study area. To create a continuous picture (or visualization) of the habitat within the study area, there must be a means to predict the values of the parameters at points that have not been directly sampled. These models assume that sampled locations closest to the unmeasured point have more similar values than samples further away. They predict intermediate values and create a continuous surface of the habitat.
Interpolation Methods Three of the most commonly used interpolation methods are inverse distance weighting (IDW), spline, and kriging. Inverse Distance Weighting IDW is a deterministic technique that calculates the parameter value at an unmeasured point using a distance-weighted average of the data points. It uses samples located within the designated neighborhood surrounding the unmeasured point to estimate the value. The weight assigned to a particular value decreases as distance from the prediction location increases; thus sample data closest to the unmeasured point contribute more to the calculated average. This technique works well for sparse or irregularly spaced data. It is commonly used in GIS programs to convert point data into raster overlays.
Spline The spline method is also a deterministic technique. A spline interpolation is calculated such that each part of the resulting raster curve is fitted exactly to a small number of data points. All parts of the curve are then joined such that the connections between one part of the curve and another are continuous. Spline algorithms can create very smooth surfaces from moderately detailed data and provide exact interpolation within smoothing limits. They are able to reveal trends within the entire data set (global trends) and within a small neighborhood of sample points around the predicted value (local variation).
Kriging Kriging is a stochastic interpolation method. It is similar to IDW in that surrounding measured values are weighted to predict values at unmeasured locations. Unlike IDW, however, kriging weights are estimated based on spatial autocorrelation between sample points. That is, a statistical relationship between values at sampled points is determined and this relationship is then applied to make predictions about unmeasured points. Kriging can be used with larger data sets than IDW or spline. The ability to compute and assess error, unique to stochastic methods, is another advantage over IDW and spline.
Although not all benthic data analyses require point interpolation, it is a useful technique for representing continuous habitat information from point samples. Once point data for a given parameter have been converted to raster data using an interpolation procedure, the data can be readily compared with other parameters. GIS is a powerful tool for integrating raster data of one type with raster data of another parameter. |
