Sample Statistic Subtract
This operator averages each sample (in the linear domain) across all pings in operand 2 and subtracts the result from the corresponding sample in operand 1. The operator can handle no more than 1000 pings. A warning message is displayed when operand 2 pings exceed 1000 pings. Operator performance at this limit is poor but feasible.
It accepts operands with the following data types as input:
Operand 1 
Operand 2 


Notes:
 Only a single ping geometry is supported, being that of the first ping in operand 1.
 Pings in Operand 2 that do not match this ping geometry are ignored.
 Pings in Operand 1 that do not match this ping geometry are filled with no data samples.
 Operand 1 sample values are output when eligible no data samples in Operand 2 are subtracted.
 You can choose to subtract operand 2 from operand 1 in the linear or dB domain. Specify this on the Sample Statistic Subtract page of the Variable properties dialog box.
 For more information about the use of Sample statistic subtract refer to the Background noise subtraction section of Background noise in Echoview and the Sample Statistic Subtract algorithm.
 The Multibeam background removal operator offers more flexibility with a sliding window for a sample statistic (mean, median or percentile) and a minimum SNR threshold.
Settings
The Sample Statistic Subtract Properties dialog box pages include (common) Variable Properties pages and these operator pages:
Operands page
Sample Statistic Subtract page
Setting 
Description 
Subtraction domain 
Specifies the domain for the subtraction:

Enter the factor for percentage range from the mean for data inclusion. 
Algorithm
This operator is useful for the subtraction of noise and/or unwanted signal for multibeam echograms. For more information on options for noise removal, refer to Background noise in Echoview.
The Sample statistic subtract operator removes noise and/or unwanted signal from the first operand by averaging each sample over all pings in the second operand, and subtracting the results from the corresponding sample in the first operand. The resulting virtual variable has the same ping geometry as the first ping in the first operand.
 two input operands specified on the Operands page of the Variable Properties dialog box.
 a Subtraction domain specified on the Sample statistic subtract page of the Variable Properties dialog box.
 a Sample exclusion value, expressed as a percentage of range from the mean sample, specified on the Sample statistic subtract page of the Variable Properties dialog box. Any sample values outside the interval are excluded (for calculation). Refer to Step 1.b. of the algorithm.
Note: If the echogram representing the background noise contains fish (targets), subsequent multibeam target detection on an echogram derived from the Sample statistic subtract variable will be adversely affected (you will get poorer target detection).
The algorithm works as follows.
In each beam of each ping:
Calculate the Mean, Minimum and Maximum sample values
 If Sample exclusion is 0% or 100%, return the Mean sample value (data are unchanged).
 If Sample exclusion has a value other than 0% or 100%.
 Calculate the Adjusted minimum sample value, Adjusted maximum sample value and Adjusted mean sample value as follows:
Adjusted minimum sample value = Mean sample value  Sample exclusion (Mean sample value  Minimum sample value)
Adjusted maximum sample value = Mean sample value + Sample exclusion (Maximum sample value  Mean sample value)
Where:
 Calculate the Adjusted minimum sample value, Adjusted maximum sample value and Adjusted mean sample value as follows:

 Exclude any samples above the Adjusted maximum sample value or below the Adjusted minimum sample value.
 If samples occur within the Adjusted interval, return the Adjusted mean sample value.
 If no samples occur within the Adjusted interval, return the Mean sample value.
