What’s this Kriging Business?!

Thanks to advances in Geographic Information Systems (GIS) computing technology, evaluating changes to lake bottoms over time has gotten much easier!  Prior to GIS, biologists and surveyors would go through great pains to ensure that repeated data collection in study areas of interest would precisely fall on the same area or transect.  If this condition was not met, data would have to be thrown out because biologists could never be sure that the difference seen between two time periods was real, an artifact of sampling a different area, or a product of sampling in a different way.  Consequently, efforts from multiple groups collecting similar data in the same system but in a slightly different way could not be leveraged.  This is an unfortunate missed opportunity that BioBase uniquely handles.

First, BioBase uniformly interprets acoustic signals and the output is the same regardless of the skill level of the individual collecting the data.  Second, BioBase employs kriging to create a statistically robust uniform map output that figuratively turns Survey 1 by Bob Smith from an orange into an apple and Survey 2 by Amy Johnson in the same area from a grapefruit into an apple.  This is unique to kriging which is a geostatistical procedure.  All other standard interpolation methods are simply 3D representations of the input data and each map will look different depending on the precise location of your survey points.  Only kriging turns different fruits into apples.

Kriging takes irregularly spaced data points and creates a smooth GIS map (also called a raster grid) based on the geostatistical properties of the input data.  Generally, points close together are more related than points farther away but the precise relationship can vary from location to location.  Kriging uses the actual statistical relationship of neighboring data points to make predictions in unsampled locations.  Other popular methods such as Inverse-Distance-Weighted (IDW) interpolation make simple assumptions of relatedness and does not use actual data to influence predictions in unsampled locations.

Through its use of kriging, BioBase removes the concern of precisely following the same path from survey to survey; which is very difficult to do on moving water even for the most seasoned surveyor.  Further this process can leverage passive data collection while doing other survey work, fishing, or simply enjoying a pleasure cruise and turn it into useful information for water resource management and protection (Figures 1-4).

Figure 1. Fisheries biologists can collect fish habitat data passively while conducting electrofishing fish surveys.
Figure 2. Passively collect depth and vegetation abundance data while enjoying a pleasure cruise with the kids or fishing.
Figure 3. Result of merged BioBase trip path data from passive data collection (above) resulting in a uniform vegetation map (below).
Figure 4. Zoomed in area of Figure 4 showing merged trip paths (above) and the uniform map output (below).  The heat map represents density of aquatic vegetation.  Blue is no vegetation growth and red represents vegetation growth that is all the way to the water surface.


Revised BioBase automated summary reports 

At 15 pings per second coming out of Lowrance depth finders, data quickly add up and without any help, users can be drowning in data and be worse off than when they started.  This issue was the topic of a previous blog post (What to do with all this data?). BioBase handles the data deluge by using kriging and creating automated summary reports.  Our recently revised summary reports now include statistics based on coordinate point data (i.e., your trip path) and data from the bathymetric and vegetation grids created by kriging (Figure 5).  When survey data collection is structured with straight transects of a uniform speed (as in the case with Figure 5), the differences between the point and grid summaries is small.


Figure 5. BioBase automated summary report excerpt showing both coordinate point and kriging grid summaries
In circumstances where lake managers are primarily interested in monitoring vegetation along standard transects, the point summaries may suit them best since the points are often uniformly spaced a part and along a straight path (Figure 6).
Figure 6. Example automated summary report showing results from a standard transect survey.  Because data lie along straight paths and are mostly uniformly spaced, point data summaries should be used.

However, if you idle for long periods time collecting samples on the lake or trying to entice finicky fish to bite, many data points amass in one location (Figure 6) and can bias the statistics from the point data (Figure 7).

Figure 7.  Vegetation point data along trip path (blue) exported from BioBase and displayed in GIS at two zoom levels overlayed on uniform kriging grid data (blue-red).  Notice the accumulation of data points over areas where the boat is only slightly moving.  Kriging creates a uniform grid of points no matter how the data are collected.

In the situation above, the differences between the point and grid data are larger and the grid data becomes more important to use for formal statistical summaries and reports (Figure 8).  The upshot is when in doubt, use the grid statistics for your data summaries.

Figure 8. Differences between point and grid statistical summaries when data along trip path are not uniform. In these situations, use the statistics from the uniform grid for report summaries and lake management decision making.

A better use of your time
By automating the complexities of creating maps, BioBase users do not need to spend precious time and money dealing with manual data collection with survey rods and hand held GPS’s, entering data with a pencil onto a datasheet, and then figuring out how to display the data in GIS and run Geostatistics models to get a map.  Before BioBase’s launch in 2011, bottom and vegetation mapping was a costly endeavor and often just wasn’t done.  BioBase is changing the game and is empowering all citizens regardless of technical expertise with the ability to see what is below the water’s surface, how it’s changing over time, and how to best manage that change.

Author: biobasemaps

BioBase is a cloud platform for the automated mapping of aquatic habitats (lakes, rivers, ponds, coasts). Standard algorithms process sonar datafiles (EcoSound) and high resolution satellite imagery (EcoSat). Depth and vegetation maps and data reports are rapidly created and stored in a private cloud account for analysis, and sharing. This blog highlights a range of internal and external research, frequently asked questions, feature descriptions and highlights, tips and tricks, and photo galleries.

Leave a Reply

Translate »
%d bloggers like this: