Abstract:

This data comprises imagery collected from an unmanned aerial system (DJI M600 Matrice) over oyster reefs in the Bay of St. Louis near Gulfport, Mississippi from 2018-12-17 to 2018-12-18. This dataset was collected with the goal of observing water quality from a remote platform. In particular, the hyperspectral data collected to observe relevant phenomena, and is meant to be evaluated along with the MicaSense RedEdge data collected simultaneously, although that data is not required to achieve all mission goals. The imagery was collected using a Headwall Nano-Hyperspec sensor, a pushbroom hyperspectral sensor that produces a line of pixel data 640 pixels wide. These are stored in chunks of data that can contain a user-specified number of lines per hyperspectral image file. The spectral range of the imager is from 400-1000 nm in wavelength, although the lower and upper 50 nm tend to be less accurate. The spectral resolution is approximately 2-3 nm per band, and the sensor records 270 of these bands in total. Related processed multispectral data is available under GRIIDC Unique Dataset Identifier (UDI): MS.x839.000:0010 (DOI: 10.7266/n7-qrp5-6479).

Suggested Citation:

Robert Moorhead, Lee Hathcock. 2020. Headwall Photonics Nano-Hyperspec processed imagery collected from an unmanned aerial system over oyster reefs in the Bay of St. Louis near Gulfport, Mississippi from 2018-12-17 to 2018-12-18. Distributed by: GRIIDC, Harte Research Institute, Texas A&M University–Corpus Christi. doi:10.7266/SDX8GEZY

Purpose:

Collection of Headwall Nano-Hyperspec hyperspectral imagery to correlate with simultaneously-collected MicaSense RedEdge imagery. The goal is to determine whether the relevant water quality phenomena can be detected using a cheaper 5-band sensor.

Data Parameters and Units:

All data is Reflectance (%). The imagery was collected using a Headwall Nano-Hyperspec sensor, a pushbroom hyperspectral sensor that produces a line of pixel data 640 pixels wide. These are stored in chunks of data that can contain a user-specified number of lines per hyperspectral image file. The spectral range of the imager is from 400-1000 nm in wavelength, although the lower and upper 50 nm tend to be less accurate. The spectral resolution is approximately 2-3 nm per band, and the sensor records 270 of these bands in total. The horizontal field-of-view of this sensor is 52 degrees (4.8 mm lens). Sensor orientation is recorded by a separate GPS/inertial measurement unit (IMU) mounted to the sensor. The orientation is recorded in the "imu_gps.txt" file. Note that there are two sensors that have been used; post-July, a newer sensor with a wide-angle lens and more accurate GPS/IMU was used. Data values are stored as 32-bit floating-point numbers, representing reflectance, a unitless measurement that codifies the percentage of light reflected from a surface within the given light wavelength. These should range from 0.0 to 1.0, although values less than or greater may be encountered in saturated pixels or those within the range of the imager noise threshold, depending on when the calibration tarp image was taken and if lighting conditions have changed in the time between line captures. Images are stored in the ENVI image format, which takes the form of a binary file and image description pair. In some cases, the processed data will be in a single file, for example, "multi_or_rf/multi_or_rf.hdr", with the "or" and "rf" extensions denoting that the imagery has been both orthorectified and reflectance-corrected. In other cases, the image set will be broken into chunks based on flight lines, which both helps process the data more efficiently and avoids mixing overlapping data on each flight line, producing more usable data.

Methods:

Data was calibrated to 11% reflectance target. This allows observation of water quality (less than 10% reflectance) without saturation while saturating on unimportant areas. Headwall's processing software converts the raw digital numbers (DNs) into reflectance and orthorectifies the data.

Error Analysis:

The orthomosaic process places the data close to where it should reside on a map, but due to inaccuracies in both the GPS and IMU, this will not be perfect. The lack of visible tie points makes this impossible to post-correct. Lighting condition changes during flight will affect the accuracy of reflectance values. If there is cloud cover, the data collected during that period will not be accurate.

Individual Files: