LASER-driftcards

This is a python toolbox that allows to process and analyze data from drift cards deployed during LASER Phase 3. The library containts scripts that permits to work on bamboo plates identification, rectification and calculation of basic dispersion statistics.

The toolbox is organized in two main modules: XY and INS

XY

Includes all the subroutines needed to work on the aerostat images. Drift cards position can be extracted and converted into x,y positions. These can then be georectified given appropriate INS data.

INS

Includes all the subroutines required to deal with the INS (Internal Navigation System) data. These includes subroutines that allow to extract pitch, roll, heading, altitude and position (lat, lon) of the aerostat for any given picture.

Subroutines to compute Lagrangian statistics are stored into stats.py and include functions for IO and subroutines to compute cloud dispersion.

Example

A typical workflow would be something like the following.

Driftcards identification

First operation would be to indentify the drifcards from the original images. A script to do this would look like this,

import numpy as np
import glob
from driftcardslib.XY import color_processing
from driftcardslib.XY import image_processing
from driftcardslib.XY import IO

exp = 'F8'

files = np.sort(glob.glob('./img/'+exp+'/LASER_'+exp+'_*.jpg'))

a_t = []
b_t = []
t_t = []

for filename in files[:]:
 print filename
 frame = plt.imread(filename)

 # dirty and cheap way to remove Masco 8 from the frame. It tends to be always in the part top of the frame.
 frame = frame[1000:,:,:] # 1000 for most experiments, 500 for F4.


 # the thresholding value for glare is sensitive
 mask, delta_s = color_processing.glare(frame,99.98) #99.99 for F4, 99.98 for F8, 99.999 for F10
 X,Y = image_processing.masking(mask,delta_s)

 path = './csv/XY/LASER_'+exp+'_'+filename[-8:-4]+'_XY.csv'
 print path
 IO.writeXY(path,(X,Y))

Compute INS data for each frame

INS data comes in high frequency data which needs to synchronized to the timestamp of the images. This is quite triky as different experiments were performed with different configurations. Here is an example of how to do it.

import numpy as np
import glob
import driftcardslib.INS.IO as INSIO
import driftcardslib.INS.time as INStime
import datetime

exps = ['F8','F10']

for exp in exps:

 # INS and timestamp are not aligned. Each experiment has a different timeGap.

 Sunday,shift = INStime.timeGap(exp)

 file_INS = '/home/jmensa/LASER/scripts_LASER/P3/INS/'+exp+'_INS_NMEA_LOG.txt'

 INS_all = INSIO.read_INS_all(file_INS)

 files = np.sort(glob.glob('/home/jmensa/LASER/data/P3/img/'+exp+'/LASER_'+exp+'_*.jpg'))

 for filename in files[:]:
  fileout = './csv/INS/LASER_'+exp+'_'+filename[-8:-4]+'_INS.csv'
  print fileout

  time_frame = INStime.timeframe(filename)
  time_INS = INStime.SOW2UTC(INS_all[:,6],Sunday,shift)

  INSIO.write_INS_frame(INS_all,time_INS,time_frame,fileout)

Georectify driftcard positions

Given INS data, we now want to project the XY positions of the driftcards to a UTM grid. A script to do that would look like the following,

import numpy as np
import driftcardslib.XY.rectify as rectify
import driftcardslib.XY.IO as XYIO
import driftcardslib.INS.IO as INSIO
import glob

exps = ['F8','F10']

f_lens = 17 #mm

for exp in exps:

 files = np.sort(glob.glob('./csv/XY/LASER_'+exp+'*_XY.csv'))
 files_INS = np.sort(glob.glob('./csv/INS/LASER_'+exp+'*_INS.csv'))

 ny = 5792
 nx = 8688

 for n in range(len(files)):
  filename = files[n]
  print filename

  X,Y = XYIO.readXY(files[n])

  INS = XYIO.readXY_INS(files_INS[n])
  
  # comp_ang_XY generates a map of the azimuthal angles between lens and driftcard given a lens focal lenght
  [angx,angy] = rectify.comp_ang_XY(f_lens,[nx,ny],X,Y)

  # dan_map_image takes angx, angy, roll, pitch, heading, elevation. Heading was bad during F8 and F10.
  [Xr,Yr] = rectify.dan_map_image(angx,angy,float(INS[3]),float(INS[4]),0,float(INS[2])) 

  fileout = './csv/XY/LASER_'+exp+'_'+files[n][-11:-7]+'_XYr.csv'
  print fileout

  XYIO.writeXY(fileout,(Xr,Yr))

Compute dispersion

Once we have the georectified positions we can compute cloud dispersion.

import numpy as np
import glob
import driftcardslib.XY.IO as IO
import driftcardslib.stats as stats

exps = ['F8','F10']

for exp in exps:

 files = np.sort(glob.glob('./csv/XY/LASER_'+exp+'_*XYr.csv'))

 a_t = []
 b_t = []
 t_t = []

 for filename in files[:]:
  print filename

  X,Y = IO.readXY(filename)
  
  # cloud dispersion is computed with the method of the principal axes
  theta,a,b = stats.princax(X,Y) # angle, a, b

  a_t.append(a)
  b_t.append(b)
  t_t.append(float(filename[-12:-8])*15)

 fileout = './csv/cloud_dispersion_'+exp+'_XY.csv'
 stats.writeDisp(fileout,t_t,a_t,b_t)

jungla / LASER-driftcards