https://doi.org/10.5061/dryad.9cnp5hqtr

Description of the data and file structure

This dataset consists of Micasense RedEdge-3 multispectral images acquired from a UAV platform at 30 m above ground altitude. Its purpose was to assess the exposure settings appropriate for acquiring images of plants, soil, and other objects of interest in agriculture without radiometric distortion. This was done to facilitate the use of fixed exposure settings in multispectral data acquisition, as opposed to the conventionally used auto-exposure settings that is prone to radiometric inaccuracies. More details on the experiments can be found in https://doi.org/10.1002/ppj2.70000.

The dataset contains the following folders:

1. Object based exposure - contains raw images and processed files (.xlsx) for 17 June 2021 and 16 June 2022 to analyze the changes in exposure time and gain in an auto-exposure flight. 
   • Canopy subfolder - contains raw images that predominantly consisted of plant canopy pixels
   • Soil subfolder - contains raw images that predominantly consisted of soil pixels
   • Tarps subfolder - contains raw images that included calibration targets
   • Exposure variations-comprehensive.xlsx - metadata file that contains the image name, latitude, longitude, irradiance intercepted at the time of data capture, exposure time, gain, object type (Canopy, Soil, or Tarps), and multispectral band. 
2. Cross-calibration dataset - contains raw images and processed data (.xlsx) to study potential radiometric errors from auto-exposure settings.
   • Sorted Samples-Batch X (X= 1 to 5) - raw images of in-field calibration targets acquired with different exposure settings for the blue, gree, red, red edge, and NIR bands (sub-folders). 
   • Sorted DLS Reflectance-Batch X (X= 1 to 5) - processed reflectance images of in-field calibration targets acquired with different exposure settings for the blue, green, red, red edge, and NIR bands (sub-folders). 
   • TarpRefl-Batch X.xlsx (X= 1 to 5) - metadata folder containing the gain, exposure time, image name,, uncalibrated (UNC) reflectance of the in-field reflectance targets, correction factor, calibrated reflectance of the infield targets, and ground truth (ASD) reflectance of the in-field targets. This data can be used to derive the ideal exposure settings for fixed-exposure flights. 
   • TarpRefl Regression piecewise-5Batches.xlsx - Contains the linear regression (LR) slope and intercept derived from all five batches combined. 
   • TarpRefl Cross Calibration MAPE-OELM 5Batches.xlsx - Contains the cross-calibration mean absolute percent error (MAPE) matrices for the five multispectral bands, which demonstrates the potential radiometric errors from calibration images with calibration equations derived from reference targets captured under different illumination conditions. 
3. Radiometric accuracy targets -
   • TarpRefl-Fixed vs Auto-Date.xlsx - assess the radiometric accuracy (R^2 and MAPE) for fixed- and auto-exposure orthomosaics captured on 17 June 2021 and 16 June 2022 (separate files). 
   • Date.tif - orthomosaic to derive the different target reflectance values
   • Date_Tarps.shp - shape files to extract panel reflectance from orthomosaics. If the positions are off, they can be adjusted on ArcMap. 
4. Vegetation Indices 
   • Vegetation Index-Fixed vs Auto-N plots gm2 - contains plot-wise total N(g/m2) values and vegetation indices derived from calibrated and uncalibrated fixed- and auto-exposure orthomosaics for 17 June 2021 and 16 June 2022. 

Work referencing this dataset:

Swaminathan, V.,Thomasson, J. A., Hardin, R. G., Rajan, N., & Raman,R. (2024). Selection of appropriate multispectralcamera exposure settings and radiometric calibrationmethods for applications in phenotyping and precisionagriculture. The Plant Phenome Journal, 7, e70000.https://doi.org/10.1002/ppj2.70000

Code/software

Codes: https://github.com/VaishSwami/Mutlispectral-camera-exposure-settings

Software: Python, ArcGIS, and Agisoft Metashape