Considerations on the collection of data from bio-argo floats across sampling scales.
0 likes143 views
The document discusses the optimization of data collection from bio-argo floats with specific reference to mission-specific sampling schemes balancing data collection goals and operational constraints. It highlights the complexities arising from increased sensor integration and presents methods to improve data sequencing efficiency through rule-setting and analysis of field data. The performance of the Navis BGC float and its data transmission capabilities, alongside comparisons between raw and transmitted datasets, are also examined.
Considerations on the collection of data from bio-argo floats across sampling scales.
- 1. CONSIDERATIONS ON THE COLLECTION OF DATA FROM BIO-ARGO FLOATS ACROSS SAMPLING SCALES. ABSTRACT The flexibility of the current generation of float sensor packages peovides an opportunity to craft mission specific sampling schemes that balance the collection of data for specific sampling goals with the practicalities of float operation. Autonomous floats operate within constraints of battery life and data transfer rates. For simplicity of data transfer and handling, most float data sets are transmitted after binning on pressure. Within a given pressure bin different instruments will be sampling within a particular defined sequence. A sampling sequence should be balanced towards minimizing energy consumption while maximizing data accuracy of each instrument. As the number of sensors increases and the breadth of mission parameters expands it becomes more difficult to optimize data sequencing and reporting. We consider methods to reduce the size of the problem by setting rules for sequence development and test those rules relative to field data. We examine a set of data from a float that was equipped with internal memory that captured the full set of sample data taken during the profiling mission. Comparing the ‘raw’ data and the transmitted data we examine the variance around the transmitted data and discuss the impact of data sequencing on the data. NAVIS BGC float 0028 (above) a f t e r d e p l o y m e n t i n t h e Mediterranean Sea. The float used in this study was equipped with and MCOMS and pH sensor. Photo by Christoph Gerigk. Navis BGCi Float Sea-Bird Scientific has developed the Navis BGCi float as a flexible, multi-role scientific platform for autonomous biogeochemical research. The float incorporates Iridium two-way satellite communications for fast data transfer and mission adaptation. Lithium batteries are included for long deployment missions. The SBE 41N CTD measures conductivity, temperature, and pressure (depth). The SBE 63 Optical Dissolved Oxygen sensor is integrated within the CTD flow path, providing optimal correlation with CTD measurements. The MCOMS has three optical sensors, providing chlorophyll a, backscattering, and CDOM, or chlorophyll a and 2 backscattering channels. MCOMS is integrated directly into the float end cap and co-located with DO and physical measurements. The Navis float can be expanded with additional sensors, in this case with an experimental pH sensor. Instrument Sequencing Pressure, temperature and conduc0vity (PTS in the diagram) are measured at 1Hz while serial instruments can have independent update rates up to 1Hz and be mixed and matched, as the deployment requires. Each instrument can have its own inquiry mode: queried for a sample or free running with con0nuous output. In this example, the SBE 63 and pH sensors are con0nuously powered, queried for a value, and put to sleep between each measurement. In contrast, the MCOMS port is powered, the output is recorded, and power is removed at the desired update rate. The MCOMS operates at 1.4 kHz and internally averages packets of LED on/off cycles to remove ambient light. The MCOMS output rate is set at 1 Hz. The sampling technique for each instrument can be tuned to op0mize the data density for the power expended. Profiling Rate and Data Output The Navis float’s ascent rate modulates the number of subsamples within each individual transmitted bin. The horizontal lines connect the negative peaks in the ascent rate with the local increases in the number of subsamples per pressure bin. Since the ascent rate is a function of the density gradient, the float will collect more samples where the gradient is increasing. Data Output Structure Data from each instrument is recorded as the float rises a c c o r d i n g t o t h e programmed instrument sequence. The rise rate of the float determines the data density within the profile as the instruments are driven by the time sequence. For example, as the float rises past 700 dbar (top panel), four complete sequences of data are completed over 4.1 dbar in 32 seconds. Later in the profile as the float passes 100 dbar (bottom panel), the float covers 1.5 dbar in 32 seconds and records four complete sequences. Ian D. Walsh, Ph. D, Joel Reiter, Dan Quittman, David J. Murphy, Thomas O. Mitchell, Ph.D. Sea-Bird Scientific GAIC 2015 Meeting Galway, Ireland 14 – 18 Sept. 2015 Time PTS PTS PTS PTS PTS PTS PTS 63 63 MCOMS pH 0 100 200 300 400 500 600 700 800 900 1000 0 5 10 15 20 25 30 Pressure (dbar) Temperature (deg C) Temperature 0 100 200 300 400 500 600 700 800 900 1000 0 2 4 6 8 10 Pressure (dbar) Samples Per Reported Value nBins MCOMS 0 100 200 300 400 500 600 700 800 900 1000 0 20 40 60 80 Pressure (dbar) Ascent Rate (dbar/min) Ascent Rate 15 sec sm Transmitted v Recorded Data – Temperature and Fluorescence Temperature data (left panel) from both the transmitted and recorded data sets overlaid in the left panel demonstrates no effective difference between the two data sets. Chlorophyll data (right panel) from both data sets demonstrates no effective difference except for at t h e c h l o ro p h y l l m a x i m a . Transmitted data is the red circles. 0 100 200 300 400 500 600 700 800 900 1000 0 0.2 0.4 0.6 0.8 Pressure (dbar) Chlorophyll (ug/l) Chlorphyll Fluorescence 0 100 200 300 400 500 600 700 800 900 1000 0 0.2 0.4 0.6 0.8 Pressure (dbar) Chlorophyll (ug/l) Chlorphyll Fluorescence Transmitted v Recorded Data – Backscattering NAVIS BGCi float 0038 was an experimental float deployed off Hawaii. The float was recovered after 100 profiles and five months at sea. 1 9/23/ 12 101 2/14/14 0 100 200 300 400 500 600 700 800 900 1000 0.00005 0.0001 0.00015 0.0002 0.00025 0.0003 Pressure (dbar) Beta (m-‐1 sr-‐1) TransmiEed bb700 0 100 200 300 400 500 600 700 800 900 1000 0.00005 0.0001 0.00015 0.0002 0.00025 0.0003 Pressure (dbar) Beta (m-‐1 sr-‐1) bb700 Median Filter 0 100 200 300 400 500 600 700 800 900 1000 0.00005 0.0001 0.00015 0.0002 0.00025 0.0003 Pressure (dbar) Beta (m-‐1 sr-‐1) TransmiEed bb700 0 100 200 300 400 500 600 700 800 900 1000 0.00005 0.0001 0.00015 0.0002 0.00025 0.0003 Pressure (dbar) Beta (m-‐1 sr-‐1) Sharpened TransmiEed 0 100 200 300 400 500 600 700 800 900 1000 0 0.0001 0.0002 0.0003 Pressure (dbar) Beta (m-‐1 sr-‐1) Chart Title 0 100 200 300 400 500 600 700 800 900 1000 0.00005 0.0001 0.00015 0.0002 0.00025 0.0003 Pressure (dbar) Beta (m-‐1 sr-‐1) BackscaEering -‐ 700 nm 0 100 200 300 400 500 600 700 800 900 1000 0.00005 0.0001 0.00015 0.0002 0.00025 0.0003 Pressure (dbar) Beta (m-‐1 sr-‐1) bb700 Median Filtered Data -‐ Median T h e t r a n s m i t t e d backscattering data (far left), the recovered data (middle), and both data sets overlain (right). The recovered data demonstrates that the transmitted data smooths the variance in the recovered data. Assuming the ‘spikes’ represent relatively rare large particles, the transmitted data undercounts the large particles. Applying a running median filter to both data sets (far left) generates coherence between both profiles. The residual data, assumed to be the large particles, again shows the undercounting in t h e t r a n s m i t t e d d a t a ( m i d d l e ) . U s i n g a sharpening algorithm ( r i g h t ) , m u c h o f t h e recovered data can be g e n e r a t e d f r o m t h e transmitted data.