-
Notifications
You must be signed in to change notification settings - Fork 1
Home
Static Observation Networks (SONs) are often used to study animal migration and habitat. Since sensors in these networks are stationary, the placement of sensors within this network is critical to achieve maximum efficacy. Currently, no automated mechanism exists to facilitate placement. This program takes advantage of high-resolution topographic data and advanced animal modeling to provide near-optimal sensor placements for these networks. Additionally, this allows for statistical analysis of existing network configurations in order to create efficacy-metrics that can be used to compare various network configurations.
- When planning to create a SON.
- When interested in comparing the efficiency of various sensor setups.
(See a detailed list of parameters here).
- A visualization of user-selected bathymetry.
- A visualization of ping distributions.
- A visualization of the "goodness" of a location.
- A visualization of sensor coverage.
- Near-optimal sensor placements (given as x,y coordinates).
- Statistics about the provided sensor placements:
- Estimated absolute recovery rate (pings are counted each time they are heard).
- Estimated unique recovery rate (pings are counted if they are heard at least once).
- Array sparsity (median distance between sensors relative to detection range). A value below 1 implies a relatively dense array with overlapping detection functions, whereas a sparsity above 1 indicates primarily non-overlapping detection functions.
- Plot of unique recovery rate as a function of number of sensors, and increase in unique recovery rate per additional placed sensor. These are useful when making a cost-benefit analysis of the projected array installation.
Some animals exhibit a perference for a certain part of the water column (on the bottom, near the bottom, or near the surface). This preference can be incorporated into the behavioral model by specifying mean (Preferred Depth) and standard deviation(SD of Preferred Depth) values for the depth off of the bottom the animal prefers. For example, speficying a depth of '0' for "Preferred Depth" indicates that the animal prefers to live on the sea floor, while a value of '5' indicates that the animal prefers to live 5m off the sea floor.
Some animals will live only in a specific depth range. For example, a deep sea fish may live only in depths of 300-400 meters. To incorporate this into the behavioral model, users can specify a minimum and maximum vertical habitat range for their animal. If this option is used, the program will only simulate animals in cells whose depths are between the minimum and maximum depths.
Animals exhibit many different movement models and habitat preferences. This greatly affects their distribution and movement across a given area. To simulate this, we provide two basic movement models: Random Walk, and Ornstein-Uhlenbeck(OU).
-
The Random Walk model assumes that animals move randomly through the environment. As a result, over the entire study period, each grid cell will see roughly the same amount of animal traffic. The result is that every valid cell (as defined by vertical habitat range and depth preference) in the grid will have the same chance of seeing an animal.
-
The Ornstein-Uhlenbeck(OU)model assumes that over time, animals will prefer to gather near a certain point of interest. Users must provide the x & y coordinates for this point, the strength of attraction in the x & y direction, and the correlation between the two as parameters to the program. See more about the process here: http://en.wikipedia.org/wiki/Ornstein%E2%80%93Uhlenbeck_process.
####Sensor Projection Normally users have a set number of sensors to place in the water. However, the question of “How much better would my results have been if I had had just a few more sensors?” often arises. The program allows for the “projection” of additional sensor placements, and graphs how much more data collection would have been possible.
####Definable Goodness Algorithms (Bias) The “Goodness” algorithm is the driving force behind the selection of sensor placements. While users are able to write their own “Goodness” algorithms, three basic algorithms are provided:
-
Animal Only (Option “1”): This option prefers to place sensors in areas of high animal activity, completely oblivious to the surrounding topography.
-
Topography Only (Option “2”): This option places sensors in areas that have the best visibility of the surrounding area. This is useful for experiments where animal habitat is unknown or to be determined.
-
Visible Fish (Option “3”): This option chooses sensor locations that have the best view of areas of high animal activity. Both animal presence and visibility due to topography are considered.