Training EcoSat Vegetation Classifications: User tips

What is EcoSat?
EcoSat delivers a one-of-it’s-kind semi-automated cloud processing of very high resolution satellite imagery to map nearshore vegetation and coastal benthic habitats.  EcoSat uses the latest multi-spectral imagery from reputable providers such as Digital Globe (World View 2,3 and 4), Airbus Defence and Space (Pleiades), and ESA’s Sentinel program and industry standard image processing techniques.  Sophisticated Amazon Web Service cloud infrastructure rapidly processes imagery, creates reports and imagery tiles, and delivers detailed habitat maps to user’s BioBase dashboard where it can be analyzed and shared.  Average turnaround time from imagery tasking order to delivery of results is 90 days.  The rapid and standard processing methods are allowing entities like the Florida Fish and Wildlife Conservation Commission to establish regular monitoring programs for emergent vegetation.  The extremely long and expensive one-off nature of conventional remote sensing mapping projects using non-repeatable tailored techniques has prevented natural resource entities from assessing the degree that habitats are changing as a result of environmental stressors such as invasive species invasions and climate change.

Continue reading “Training EcoSat Vegetation Classifications: User tips”

Consumer Sonar for Bottom Mapping: Updated Reference List

Another FAQ we get is wondering if there are published studies using BioBase technology? There are many legacy applications on which the BioBase technology is based. Further, now that a sufficient passage of years has accumulated to support the “research to publication” cycle, we’re happy to share several BioBase-specific studies published in the peer-reviewed literature.  This is far from an exhaustive list and we’ve intentionally left out the niche growth in consumer side-scan technology for creating habitat maps.  If there are good published papers you know of that are not on this list, please share in the comments.

Continue reading “Consumer Sonar for Bottom Mapping: Updated Reference List”

BioBase 3 Step Process: Important Details!

A primary strength of BioBase EcoSound is its simplicity and that is reflected in the easy 3 step process of “Collect,” “Upload,” and “Analyze” (Figure 1).

Figure 1. The core process of EcoSound depicting the 3 Steps of “Collect,” “Upload,” and “Analyze.”

But there are many strategies that users can employ that will ensure that they will get the best EcoSound outputs possible.  We’ll focus on several questions under each of the three categories

COLLECT

What is collected?
BioBase EcoSound analyzes the downlooking 200 kHz broadband frequency from any Lowrance, Simrad, or B&G sonar/gps chartplotter capable of recording files as .slg, .sl2, or .sl3 to high capacity storage media (e.g., SD, MicroSD).  Each of the 10-20 acoustic signals per second has detailed information about the depth (depth to the top of the substrate), bottom hardness, and the height of aquatic vegetation canopies if present.

How do you collect data?
After you open your BioBase account, go to the Support and Resources page and print off the quick start guide for your unit (Note: Elite Ti users should reference the HDS quick start guide).  These guides gives you important instructions about setting configurations and how to record your data.

Choose the hardware, watercraft, and mount that will best suit your needs.  You will get the most consistent and clearest signal from a fixed transducer mount either on the transom or through the transducer hull.  Take careful measure to ensure the proper transducer angle and to minimize the creation of cavitation (air bubbles) near the transducer face.  Your unit will come with a detailed transducer installation guide – read it and follow it carefully!  Cavitation can create false vegetation detects or missed bottom detects and greatly affect the speed at which you are able to map.  Click here to learn more about transducer selection and installation.

Once you have a good understanding of how to mount your transducer, you have a range of portable options available if you need to either get into small water bodies or transfer the equipment to multiple vessels.  This post has a great gallery of photos from the BioBase mapping community demonstrating a range of portable options.   With any mount, ensure that the GPS position is very near the transducer such that X,Y, and Z positions are all aligned.  For console-steer boats, consider a Point-1 GPS external antenna.  Learn more about GPS here.

How far apart do I space my transects?
Using geostatistics, EcoSound creates bathymetric, aquatic vegetation, and bottom hardness digital models with point data much like how land topography is mapped (e.g., “Digital Elevation Models” or DEM’s; Figure 2)

Figure 2.  Schematic simulating the actual bottom area quantified by a 20-deg Lowrance Transducer (model HST-WSBL – white box) and 6-deg Airmar Transducer (model TM-260 yellow box).  At 5 ft, the 20-deg transducer insonifies a 1.76 ft area, whereas the 6-deg insonifies only a 0.5 ft area,  Given the rapid ping rates, most applications will have very high overlap of adjacent pings and thus be quantifying the entire bottom directly under the mapping vessel.

Once you understand where actual measurements are being recorded, you can make an informed judgement on the appropriate level of bottom “sampling” with transects.  As with any kind of sampling, the appropriate level of sampling depends greatly on the study or management objectives. Users might be able to use painting as a useful analogy for a starting place to guide transect spacing decisions.  For instance, if you are painting a large flat wall, you use a large roller and your objective is to get the best coverage as fast as possible.  Similarly, if you are mapping a large lake or bay with gently changing bottom features, you can space transects relatively wide (e.g., 100-300 m) and drive a modest speed (e.g., 8 mph or 12 km/h) if you have a good install with your transducer.  In contrast, if you are painting an ornate cabinet with detailed carvings, then you have to use a small brush and spend a lot of time getting that small brush into all cracks and crevices.  Same goes with a highly complex lake or reef bottom.  In these cases, close transects (e.g., 5-m) with a narrow beam transducer, and slow speed (2 mph or 3 km/h) might be called for.  Our Transect Design Blog goes into these issues in more detail.

Log Sonar while doing other activities
Running BioBase EcoSound surveys does not need to be a dedicated activity that you now have to layer onto your other obligations.  Rather, EcoSound is a tool for passive data capture during activities you are already on the lake conducting.  For instance, biologists are on lakes and bays everyday throughout the globe bouncing spot to spot taking samples of aquatic plants.  Most have a gps and sonar on their boat to find where they are going.  EcoSound allows them to be recording and capturing detailed spatial information about exactly how much plants are growing not only at their sampling spot, but between sampling location.  Many Aquatic Plant Biologists using EcoSound for the first time are shocked by what they missed with their original surveys.  This analysis provides possibly the most eye opening example of how spatially dynamic aquatic plant growth is in glacial lakes.

The second example is sampling fish with electrofishing.  Habitat degradation is often cited as one of the primary causes of global fisheries declines, yet rarely are habitat data collected in conjunction with fisheries data.  Recently, Lowrance and BioBase teamed up with the global leader in Electrofishing Smith-Root to demonstrate how Fisheries biologist can integrate Lowrance HDS into their electrofishing boat and collect important fish habitat data while electrofishing (Figure 3).  BioBase EcoSound allows you to simultaneously collect both fish and habitat data, thus leading to better informed Fisheries Management decisions.  Read more about this partnership here.

Figure 3. Integrate Lowrance HDS into your Electrofishing Boat to monitor boat function, view structure real-time, and record important fish habitat data.  Inquire with Smith-Root about how to retrofit old boats or outfit new ones.

UPLOAD

Cloud-computing has quickly become the new standard for managing large datasets.  Remote servers and databases are better able to manage the large data-rich .sl2 files than most desktop programs and external storage media.  Use our desktop upload tool to select your .sl2 file from your card and upload.  After your upload is complete, the file hits one of our dozens of servers that goes to work analyzing every signal.  Our algorithms perform many automated tasks including:

Data Cleansing
Signal quality is reviewed for every signal.  Those not passing certain tests, are discarded (at 10-20 data points per second, you can afford to discard some bad ones).  These tests include whether the operator was going too fast (e.g., 20 mph for depth, 12 mph for vegetation, 10 mph for composition), whether the signal was too noisy, or whether depth was lost for some reason.  Bad soundings are removed automatically.

Feature Extraction
With 10-20 data points per second, aggregation is necessary to create manageable outputs that aren’t “too busy.”  Therefore, EcoSound algorithms create georeferenced data “points” that represent averaged areas (see Figure 2).  These X,Y,Z points can be exported from BioBase and imported into GIS for important exploratory analyses (e.g., to understand data coverage and spacing of attribute points) or statistical analysis (e.g., statistics from transect studies).  Most other hydroacoustic processing software stops here.  BioBase takes it one step further and creates robust geostatistical maps.

Geostatistical Interpolation
Finally, once the coordinate points are created for all layers, the points are sent to a generic kriging algorithm that uses geostatistical models to make predictions of depth, vegetation, and bottom hardness in areas not sampled.  Kriging creates a uniform grid (grid cell size is user configured and 5 times less than the track buffer) for each EcoSound layer.  Redundant point data while idling gets averaged into one grid cell value.  By knowing the value of neighbor points, kriging can predict the value in an unsampled location.  Read more here…

Quality Control with Real People!
Although automation has led to huge productivity gains in the last century, quality control by humans is a critical component to a full solution.  Accordingly, BioBase Quality Control Engineer staff review every uploaded sonar log and review the quality.  Staff ensure the signal looks clear, that the output is free of evidence of improper installation (e.g., slanted transducer, Figure 4), and that the digital shoreline (lake polygon shapefile) aligns with the Bing aerial imagery and has sufficient zero values along shore such that contours do not intersect shore.  In general, QC staff will review your trip and make adjustments to sensitivity or shorelines within 24 business hours of your upload.  Maps should be considered “provisional” prior to QC review.  After QC staff check off, users can do their own QC checks prior to analyzing the map data in depth.

Figure 4.  Sample output with a improperly installed transducer that would be flagged by BioBase’s Quality Control Team.  QC Staff check every file and will alert you to verify the output and possibly make corrections.

ANALYZE

Trip Replay
After you receive an email informing you that your trip is ready for viewing and QC have completed their review, you can replay your sonar log synced with your track and interpolated map and accompanying data.  You can add or delete data by clicking on areas on the map, on the sonar log, or in the data table below the map and choosing delete.

Polygon Tool
Digitize an area of interest such as a bed of an invasive aquatic plant species.  The EcoSound polygon tool will use the polygon like a cookie-cutter and clip the statistics for just your specific area.  Further, BioBase has a partnership with United Phosphorus Inc. to help aquatic plant managers create precise herbicide treatments with their polygons using the UPI Treatment Tool. Finally, these polygons can be exported as shapefiles and converted to .gpx for viewing on your Lowrance Chartplotter.  This feature enhances precision of aquatic plant management and dredging activities.


Data Offset
Gain accuracy in water elevations by using the offset tool to correct for your transducer depth or create a map with a benchmark elevation as discussed in this blog.  If you are working in a coastal area with tide stations, your trip will be automatically offset to Mean Lower Low Water (MLLW) every 5 minutes as described here.

Trip Reprocessing
If you make edits to your map or want to fill in gaps in your map with a buffer, you’ll need to send the trip back to the BioBase servers to update the map and datasets.  You can increase map detail by decreasing the buffer, with a minimum buffer of 5-m and 1-m grid cell resolution.  You can fill in gaps and create a generalized map by increasing the buffer.

Merge Trips
With multiple-use subscriptions or special single-lake subscriptions, you can combine files from anyone using a Lowrance HDS or Elite HDI/Chirp to slowly build the best map ever created.  One amazing example is the effort put forth by citizens on Ten Mile Lake in Northern Minnesota simply by passively recording their sonar while they were out boating or fishing during a 2 year span.  Most commonly, BioBase users use the merge function to combine multiple smaller sized files into a larger aggregate for an entire waterbody.

Export Data
The primary strength of BioBase is its ability to rapidly process very large raw datasets and produce valuable spatial data.  As we’ve discussed, BioBase has many “turn-key” features that are valuable to the everyday practitioner who may not be trained in Geographic Information Systems (GIS).  However, the value of BioBase outputs increase dramatically for those who export BioBase data and run spatial analyses or create custom map layouts with BioBase and other data layers.  As ESRI Silver Partners, we support use of BioBase data layers in ESRI products like ArcMap and actually have step-by-step tutorials that will walk you through how to create GIS data layers from your BioBase outputs

Automated Reports

Another popular feature of BioBase EcoSound is that automated summary reports are produced with every upload or merge and stored on a dedicated file server for sharing with partners.  Partners just see the report and do not have access to your account.  Further, if you add waypoints to your map, they can also be shared.  For example, if I want to include a BioBase aquatic vegetation summary report in a larger .pdf report that I am sending to a client or study sponsor, I just include the link in the report and now the person with whom I am sharing has dynamic html with which they can interact and find the statistic(s) that most interest them.

These reports have many numbers and might be confusing at first, but simply hover over the question mark with your mouse, and it will tell you what the numbers mean.  Data summaries are created from the “point” data along your track and also the kriging interpolated “grid” data that is created from the point data.  If you are primarily interest in monitoring repeated transects or the max depth of vegetation growth, use the point data statistics.  If you are doing back and forth mapping (most EcoSound applications), use the grid statistics.

COLLECT – UPLOAD – ANALYZE!

Analysis of Alternative Mapping Methods

Budgets are tight, time is short, labor resources and technical know-how are scarce.  These truths are the motivating force behind the ciBioBase system.  Recently, we ran an analysis that demonstrates the cost-effectiveness of ciBioBase.  We selected 3 peer-reviewed studies that demonstrated three alternative methods for whole lake assessments of vegetation abundance and compared the costs of producing a vegetation biovolume map with ciBioBase.  The first two studies Valley and Drake (2007) and Sabol et al. (2009) used a scientific-grade echosounder, associated software, and required expertise in hydroacoustics and Geographic Information Systems (GIS).  Hardware and software costs were adjusted to 2012 dollars which actually brought costs down to a total of $18,400.  These costs were amortized over 5 years at 5% interest and scaled to daily costs assuming use in a season would not typically exceed 45 days.  For both methods, hardware and software costs amounted to approximately $84 a day.  We did not factor in time on the water for any of these analyses, or the cost of training in hydroacoustics, geostatistics, and GIS.

Labor costs were relatively large in the Valley and Drake (2007) study because the authors were working in environments that exceeded the capability of the vegetation-detecting algorithm they were using.  Specifically, noisy signals generated in surface-growing vegetation canopies were thrown out and thus biasing biovolume (i.e., percent of the water column occupied by vegetation) downward.  Consequently,  Valley and Drake did ping-by-ping verification and reclassification where signals were obscured by surface-growing vegetation.  Summing the modest hardware and high labor costs to manually verify thousands of pings, the cost of producing a vegetation map in a 500-acre lake using methods described in Valley and Drake (2007) was approximately $1,288.

Labor costs were significantly lower in the Sabol et al. (2009) investigation because we assume vegetation did not grow to the surface in the Wisconsin study lake during the investigation and thus the vegetation algorithm processed individual files relatively quickly.  Taking the labor costs (10 hrs @ $25/hr) in Sabol et al. (2009) and adding in adjusted amortized hardware and annual maintenance costs, the costs of producing a map on a 500-acre lake was a much lower $357 compared with Valley and Drake (2007).
The third study evaluated the LAKEWATCH volunteer lake monitoring program administered by the University of Florida.  LAKEWATCH utilizes commercial-grade Lowrance sonar units to log data on bathymetry and vegetation height/biovolume (otherwise known as percent volume inhabited; Hoyer 2009).  Entry-level technicians analyze 100 random points from pooled transect files and record depth and estimate plant height to get a lake-wide estimate of percent area covered by vegetation and percent volume inhabited with aquatic plants.  Although the objective of LAKEWATCH is not to create high resolution vegetation maps, in order to make apples-to-apples comparisons, we had to scale-up the Hoyer (2009) method to reflect the same survey resolution (16,383 points) of the previous two methods.  This resulted in an incredibly high cost of $6,884 to produce the same type of vegetation map as described with the previous two methods.
ciBioBase 
Because we automate the analysis and mapping of vegetation, there is very little labor outside of conducting the survey, save for a recommended hour of reviewing the data after a trip and verifying the output.  Also, the hardware and software costs are minimal because we analyze data from Lowrance HDS-line sonar systems that are coupled with differentially corrected GPS systems and retail for $700-$2200.  Running the same calculations as the other methods, we estimated the per survey day cost of mapping a 500-acre lake was a very low $125; 2.8 times cheaper than the next lowest described by Sabol et al. (2009).

Daily Costs
Method Amortized Hardware Maint-enance Labor Subscription Cost Total      Cost
Valley and Drake (2007) $84 $23 $1,181  NA  $1,288
Sabol et al. (2009) $84 $23 $250  NA  $357
Hoyer (2009)* $3 $0 $6,881  NA  $6,884
ciBioBase $3 $0 $25 $97 $125
*High resolution vegetationmapping was not an objective of Hoyer (2009) and thus the following scaled-up cost estimates should be viewed as a hypothetical scenario for an equal comparison to other methods

The low rate of ciBioBase doesn’t consider any of the value-added features of ciBioBase such as:

·       Automation: No training needed in hydroacoustics, geostatistics, or GIS.  Our cloud-based software analyzes patterns in the acoustic signal and uses standard geostatistical techniques to produce accurate maps.
·       Centralization: As data from more systems is uploaded, algorithm performance is continually verified and enhanced.  These enhancements are constantly refined in the cloud and are pushed universally to all users, free of charge.

·       Crowd-sourcing: Multiple subscribers from an organization can contribute their data to an optional shared repository.  Organization members can leverage each other’s efforts and data to produce a single output.

·       Speed: Lowrance sonar units occupy little space on board (and actually are portable!) and come with a skimmer transducer that allows data collection of up to speeds of 10 mph.  As such a 500-acre lake may take half the time to traverse 25 mi of transect compared with methods 1 and 2.

·       Efficiency: Because there’s no “set-up and break down” with our method, hitting “record” is the extent of the effort you need to do to start logging data.  While doing so, you can be collecting other important fisheries, aquatic plant, or water quality data on the lake.
·       Data Visualization and Verification: We offer visual, geospatial tools to replay your trip and verify the automated output.

Log in and see for yourself! Go here and type demo@cibiobase.com for the login email and for the password enter “demo.”  You’ll first need Microsoft Silverlight, click here to check to see if you already have it installed on your PC or Mac or need to download it.
Literature Cited
Hoyer, M.V. 2009. Calculations for successful planning. Lakeline Spring 2009: 39-42.

Sabol, B.M., Kannenberg, J., and Skogerboe, J.G. 2009. Integrating acoustic mapping into
              operational aquatic plant management : a case study in Wisconsin. Journal of Aquatic Plant
              Management 44-52.

Valley, R.D. and M.T. Drake 2007. What does resilience of a clear-water state in lakes mean for the spatial heterogeneity of submersed macrophyte biovolume? Aquatic Botany 87: 307-319.