Thresholds

QC test thresholds for depth were based on historical Coastal Monitoring Program data. Preliminary quality control (QC) was applied to the historical data. Obvious outliers and freshwater stations were omitted.

The thresholds were calculated and applied to different groups for each QC test, as described below.

Gross Range Test

Sensor Thresholds

The sensor thresholds were determined based on the manual for the aquaMeasure and VR2AR sensors (Table 1).

Table 1: Depth sensor thresholds for the Gross Range Test.

User Thresholds

Measured sensor depth is a function of where the sensor is located on the string, and is not necessarily related to the historical measurements (Figure 1). For example, in Antigonish County have typically been moored in relatively shallow depths, but deeper sensors could be deployed there in the future.

To avoid flagging future observations at unusual depths, a single set of user thresholds was manually selected and applied to the pooled data.

Figure 1: Distribution of depth observations (binwidth = 2 m). Dotted orange lines indicate the user thresholds.

Table 2: Gross Range Test user thresholds for measured sensor depth.

Climatological Test

The Climatological Test was not applied to the measured depth data because there was not an expected or observed seasonal cycle (Figure 2; fig-depth-hist-clim).

Figure 2: Mean +/- 3 standard deviations of the monthly depth observations.

Figure 3: Seasonal distribution of depth observations (binwidth = 2 m). Dotted orange lines indicate the user thresholds.

Spike Test

Separate thresholds were calculated for each county to account for the substantial spatial differences in tidal range around the province.

Because of the difference in precision between the types of sensors that record depth, the Spike Test thresholds were calculated by county and sensor type (Figure 4; Figure 5).

For all groups, the distribution of the spike value is skewed right, and so several upper quartile values (90th, 95th, and 99.7th quartile) were evaluated to use as the \(spike_{low}\). There were relatively few large single-value spikes in the depth data, and so the 99.7th quartile was selected to avoid false positives. \(spike_{high}\) was defined as 3 * \(spike_{low}\) to identify especially egregious spike values.

In Digby and Pictou counties, the only depth observations have been from VR2AR sensors. In the future, depth may be measured with aquaMeasure DOT sensors in these counties. If this happens, the VR2AR thresholds will be used to assess the spike test until the threshold calculation exercise is repeated.

Figure 4: Distribution of the spike value of depth observations measured by aquaMeasure sensors (binwidth = 0.1 m). Dotted orange line indicates \(spike_{low}\); dotted red line indicates \(spike_{high}\). Note difference in y-axis scale.

Figure 5: Distribution of the spike value of depth observations (binwidth = 0.1 m) measured by VR2AR sensors. Dotted orange line indicates \(spike_{low}\); dotted red line indicates \(spike_{high}\). Note difference in y-axis scale.

Table 4: Spike thresholds for depth.

Rolling Standard Deviation Test

The Rolling Standard Deviation Test thresholds were calculated and applied separately for each county due to spatial differences in tidal range.

The distribution of rolling standard deviation is relatively normal for each county (Figure 6), and so the mean and standard deviation were used to calculate \(rolling\_sd\_max\).

Figure 6: Distribution of the 24-hour rolling standard deviation of depth observations (binwidth = 0.1 m). Dotted orange line indicates \(rolling\_sd\_max\).

Table 5: Rolling standard deviation threshold for depth.

Depth Crosscheck

Figure 7 shows the minimum measured depth vs. the estimated sensor depth at low tide. Figure 8 shows a histogram of the absolute difference between these measurements.

Figure 7: The minimum measured depth vs. the estimated sensor depth at low tide.

Figure 8: Distribution of the absolute difference between the minimum measured depth vs. the estimated sensor depth at low tide (binwidth = 1 m).

Table 6: Depth crosscheck threshold.

The distribution of the \(depth\_diff\_max\) is skewed right, and so several upper quartile values (90th, 95th, and 99.7th quartile) were evaluated to use as the threshold. The 90th quartile was considered too stringent, while the 99.7th quartile was too lenient. The 95th quartile was considered the most useful threshold.