# Predicting_Photogrammetric_Uncertainty


### A Bayesian approach for predicting photogrammetric uncertainty in morphometric measurements derived from UAS 



**Materials**

*Data*:    
  - "APE-Dataset.csv" - training/calibration data for different UAS platforms.  
      - Images = image used for measuring calibration object
      - Aircraft = the UAS aircraft used to collect imagery of calibration object to measure
      - Flight = the flight number/name
      - Position = indicates if calibration object is in center of corner of image frame
      - Focal.length = focal length of camera
      - Pix.Dim = pixel dimensions (mm/px), which is the sensor width (mm) / image width (px) of the camera
      - Baro_raw = the raw relative altitude recorded by the barometer
      - Launch.Ht = the launch height of the drone, to be added to the BarAlt to get the absolute barometric altitude above the surface of the water
      - Baro + Ht = the baro_raw + Launch.Ht to get the absolute barometer altitude
      - Laser_Alt = the altitude recorded by the laser altimeter. Blanks spaces/NAs indicate no/false reading
      - Alitude = altitude used in measurement, either Laser_Alt or Baro + Ht
      - Alimeter = which altimeter was used for altitude in measurement; either barometer or laser
      - Measurer = which analyst measured the calibration object (3 analysts total, measured each image once)
      - RRR.pix = the length in pixels of the known sized object (calibration object) 
      
      
  - "humpback_data.csv"  - testing data. Humpback lengths from the Western Antarctic Peninsula. 
      - Animal_ID = unique ID for individual whale
      - Reproductive_Class = determined from biopsy sample or from drone images (mom and calf pairs)
      - Species = the species of whale
      - Image = the image ID used for measuring the whale
      - Aircraft = which UAS platform collected the imagery
      - Focal_Length = focal length (mm) of the camera used
      - LaserAlt = the altitude recorded by the laser altimeter. Blanks spaces/NAs indicate no/false reading
      - BaroAlt = the raw relative altitude recorded by the barometer
      - Pixel_Dimension = (mm/px) the Sensor width (mm) / Image width (px) of the camera
      - Launch_Ht = the launch height of the drone, to be added to the BarAlt to get the absolute barometric altitude above the surface of the water
      - Altitude = altitude used in measurement, either LaserAlt or BaroAlt + Launch_Ht
      - Altimeter = which altimeter was used for altitude in measurement; either barometer or laser
      - 'TL' = total length of the whale in meters


  

*Uncertainty Models*:  
  - Model 1: barometer only
      -  "Model_1-baro.Rmd" - uses altitude from barometer and laser in training data, but only barometer in testing data. 
      -  Generates a diagnostic reports of posterior predictions
      -  Saves outputs as rdata and csv files to be used with the saved outputs from Model 2 to compare uncertainty results from both models using "Uncertainty_Plots.Rmd".
      
      
  - Model 2: barometer and laser
      - "Model_2-baro&lsr.Rmd" - uses altitude from both barometer and laser in training and testing data. 
      - Generates a diagnostic reports of posterior predictions
      - Saves outputs as rdata and csv files to be used with the saved outputs from Model 1 to compare uncertainty results from both models using "Uncertainty_Plots.Rmd".


  Both models located in the 'reports' folder.  
  


*Plots*:    
  - "Uncertainty_Plots.Rmd"   
      - Create figures in manuscript/supplementary material for comparing the outputs from Model 1 & 2. 
      - First run 'Model_1-baro.Rmd' and 'Model_2-baro&lsr.Rmd'. The saved outputs from both of these Models are used as the inputs to compare results and create figures.
      - Figures saved to the "uncertainty_figures" folder in the "output_figures" folder.  
    
  
  - "uncorrected_error_plots.Rmd"   
      - Creates figures in manuscript/supplementary material of unocorrected error (measurement error before running uncertainty model) and length estimates from the training/calibration datal. Includes the uncorrected measurement error for each UAS.
  
  
  
  Both .Rmd files located in the 'reports' folder. 
  
  
  
**Steps**:  
  1. Run 'Model_1-baro.Rmd'   
        - Can knit document to run model and automatically generate posterior diagnostic report.  
        - Model outputs are automatically saved as rdata and csv files in the "model_outputs" folder.    
            - These outputs will be used later along with the saved outputs from 'Model_2-baro&lsr.Rmd' to create figures using 'Uncertainty_Plots.Rmd'.  
            
        
  2. Run 'Model_2-baro&lsr.Rmd'   
        - Can knit document to run model and automatically generate posterior diagnostic report.  
        - Model outputs are automatically saved as rdata and csv files in the "model_outputs" folder.    
            - These outputs will be used next along with the saved outputs from 'Model_1-baro.Rmd' to create figures using 'Uncertainty_Plots.Rmd'.  
        
        
  3. Run 'Uncertainty_Plots.Rmd'   
        - Uses the saved outputs from Models 1 & 2 to create figures in the manuscript/supplementary material comparing the outputs from Models 1 and 2.  
        - Figures automatically saved to the "uncertatinty_figures" folder in the "output_figures" folder.
        
        
  4. Run 'uncorrected_error_plots.Rmd'  
        - Creates figures in the manuscript/supplementary material of the uncorrected error (before running uncertainty model) from the training/calibration data.  
    

  
Contact KC Bierlich kcb43@duke.edu for more details.