Field Spectroscopy – Part Two: Innovative data logging protocols to meet increased demand
Authored by James Caudery, Geospatial Analyst with 2Excel Geospatial – James has an academic background in Earth Sciences (BSc HONS by thesis) from Monash University, Australia, and several years of experience in Geographic Information Systems. James is skilled in data processing, spatial analysis, and cartographic data visualisation. He is also lead for Field Operations and has successfully run many expeditions in locations around the world.
Last week, in Part One, we demonstrated how we increased our capacity to collect high-quality reflectance spectra in the field. In this Part Two, we consider the challenge of matching thousands of field spectra to other details about the target.
Single Spectrometer Data Logging
“Manual data logging is prone to error; assisted logging tools mitigate these errors and improve efficiency”
A critical element of field spectroscopy is to record the details (metadata) about the target at the time of measurement. Ideally, collecting data using a single spectrometer (as described in Part One), is conducted with two operators, one operating the spectrometer and one to log metadata. However, manual data logging in the field is time-consuming and prone to error; for example, if the scan numbers and logs become unsynchronised, this error can prove to be unrecoverable.
To address these issues, as part of our wheat disease campaign, we developed an application within Microsoft Excel to assist data logging, which is depicted in the figure below.
An example of an early assisted data logging tool
This application allows the operator to record the actions taken by the spectrometer operator. This includes optimizing the [ASD FieldSpec 4] instrument to the intensity of light conditions, taking a white reference measurement (to be used to correct the target measurement/s), taking a quality control white reference measurement, and taking an actual target measurement. Each time an action was triggered, the scan (measurement) number would auto-update as appropriate. Other buttons were also included, which related to the target being measured. In the example shown above, this included the specific wheat cultivar and its treatment regime.
As described in Part One, white reference measurements corrected with a previous white reference measurement used to check the change in illumination in the intervening period. If the result varies more than a few percents, then the target measurements would be marked as poor and may prompt a further collection attempt.
This procedure improved efficiency over traditional form filling, and vastly reduced the occurrences of omissions and errors. Back in the office, the data log was used to group spectra into various categories and ensure that only reliable spectra were passed on for exploitation.
Automated Data Logging for the [ASD] Dual Spectrometer System
“Our system reduces time spent in the field, maximizes collection capacity and eliminates data logging.”
In Part One, we discussed the introduction of a Dual Spectrometer System. This system is capable of rapidly collecting thousands of spectra per hour. In order to realize this potential in a field context, we needed to upgrade our data logging abilities and so decided to use a spatial solution. Rather than actually recording the metadata in the field, we would record the spatial position of each measured spectrum and use this to match the measurement to the target’s metadata, post-collection.
The spatial accuracy required was on the order of centimeters. To obtain this accuracy we procured the Reach RS RTK GNSS system by Emlid. In this configuration, one receiver with an accurately known location acts as a base station, which transmits real-time corrections to the roving receiver positioned close to the spectrometer’s fiber optic cable.
From left to right: The RTK GNSS base station; the operator with a controller, roving spectrometer, and RTK GNSS unit; and the base [ASD FieldSpec] spectrometer measuring the standard reference panel synchronously with each roving target measurement.
This setup allowed roving spatial and spectral measurements, constantly referenced to a known position and standard reference respectively. To mitigate the influence of shadows and spectral contamination caused by the operator, the fiber optic cable and roving GPS are mounted on a side boom.
This system enables the constant recording of field spectra, with each measurement being geotagged with a highly accurate position. Quality checks to mitigate the effects of instrument drift were made periodically, by using both spectrometers to measure the standard reference simultaneously.
Post-processing associates each spectrum with a specific plot and its associated metadata.
Using a Geographic Information System, the spectral measurements can be spatially matched to known attributes of each target, completely eliminating any manual data logging in the field. The outcome of this development is a system that reduces time spent in the field, maximizes collection capacity and eliminates manual data logs.
The equipment and operators described in this article are available as a stand-alone service or as a component of a remote sensing project.