CTD thermocline and DCL automated detector

This folder creates a tool to analyze CTD (conductivity, Temperature, and depth) profilers, specifically for SeaBird CTD profilers.

It can:

• Detect lake stratification (the location of thermocline, epilimnion, hypolimnion) using piecewise linear segmentation (HMM with maximum gradient is currently turned off to save compuation time)

• Detect deep chlorophyll layers by fitting two half gaussian curves on peak points

• A Flask Python app is available, run

  python application.py

---

To Install: Installl anaconda and bokeh

To use:

from seabird.seabird_class import seabird
import json
import matplotlib.pyplot as plt

config=json.load(open('config.json'))
mySeabird = seabird(config = config)  # config is in json format
filename = "sample.cnv"
mySeabird.loadData(dataFile = filename)  # filename is the cnv file, taking data from database is optional, see the source code
mySeabird.preprocessing()
mySeabird.identify()
features = mySeabird.features  # features are in dictionary format.

print features

The output features are:

DCL features:

• allConc: the sum of Chlorophyll concentration with 0.25m sampling interval

• allConc_upper: the sum of Chlorophyll concentration above LEP, with 0.25m sampling interval

• DCL_upperDepth_fit: the upper boundary of the DCL peak. This value is always deeper than LEP since LEP is used as a threshold to cut.

• DCL_bottomDepth_fit: the lower boundary of the DCL peak

• DCL_upperConc_fit: the concentration at DCL_upperDepth_fit

• DCL_bottomConc_fit: the concentration at DCL_bottomDepth_fit

• DCL_leftSigma: the sigma of the Gaussian shape for upper shape

• DCL_rightSigma: the sigma of the Gaussian shape for lower shape

• DCL_conc: the concentration at the maximum peak

• DCL_depth: the depth with maximum concentration for the DCL peak

• DCL_concProp_fit: the ratio of Chlorophyll in DCL, compared to the whole water column

• DCL_exists: whether DCL exist

• peakNums: how many significant peak detected

• DCL_leftShapeFitErr: the error of fitting a Gaussian shape for upper DCL part

• DCL_rightShapeFitErr: the error of fitting a Gaussian shape for lower DCL part

Lake Stratification features

• TRM_gradient_segment: the gradient of temperature at TRM (unit C/m)
• TRM_segment: the depth of thermocline using PLR method
• UHY_segment: the depth of upper hypolimnion
• LEP_segment: the depth of the lower epilimion
• doubleTRM: whether double thermocline exist
• positiveGradient: whether abnormal temperatuer increasing with depth exists
• firstSegmentGradient: the gradient of the first segment (C/m)
• lastSegmentGradient: the gradient of the last segment (C/m)

To plot:

mySeabird.plot_all(interestVarList=["Temperature","DO","Specific_Conductivity","Fluorescence","Par"]) # plot the water quality from raw data
mySeabird.plot() # plot the detected depth
plt.show()

---

File structure:

--Seabird:
	--seabird_class.py: seabird class
	-- deepChlLayers.py: DCL class
	-- thermocline.py: thermocline class
	--tools: contain general file parser, database connection
	--models: Piecewise linear segmentation, HMM and threshold methods.