R for Photobiology

1.1 Why and how did I develop the packages?

• Because I am lazy, but mostly in the long term
  • … I like efficiency + reliability
• Approach: write code well once and use many times
• Document: for future me and for others
• Test cases: avoid errors now and in the future
• Write clear code: make enhancements and corrections easy

1.2 Did it work out as I expected?

• Expected: In the long run I did save time and effort
• Expected: Became easier to work with large data sets
• Expected: Other researchers are using the packages
• Unexpected: As it made calculations a lot easier
• … I did “what if” calculations more frequently
• … leading to new insights and original research
• Unexpected: Citations!

1.3 Why did I use R?

• I had been using R for > 10 years
  • … as it is the best language for statistics
• The logic of R fits well with how my brain works
  • … it feels natural to me (not necessarily to you)
• R had the best plotting approach available
  • … and I was familiar with it
• The package system of R is very well structured

1.4 Side note on R

1.5 Design aims for my 3rd attempt

I made bad design decisions in my first two attempts at writing an R package for photobiology. My third attempt and current packages have as aims:

• Avoid accidental loss of information
• Automate all that can be automated
• Allow customizations
• Store metadata together with the data
• (Fast computations on large data)

1.6 Design aims in detail

• Automate all that can be automated
• Allow customizations
• Store metadata together with the data
• Avoid accidental loss of information

1.7 Design aims in detail

• Automate all that can be automated
  • Known units and bases of expression
  • Provide good defaults, but allow overriding them
• Allow customizations
• Store metadata together with the data
• Avoid accidental loss of information

1.8 Design aims in detail

• Automate all what can be automated
• Allow customizations
  • Open source code
  • Export “building blocks” to allow extensions
• Store metadata together with the data
• Avoid accidental loss of information

1.9 Design aims in detail

• Automate all that can be automated
• Allow customizations
• Store metadata together with the data
  • What was measured
  • When was it measured
  • Where was it measured
  • How was it measured
  • Store a trace of operations applied
• Avoid accidental loss of information

1.10 Design aims in detail

• Automate all that can be automated
• Allow customizations
• Store metadata together with the data
• Avoid accidental loss of information
  • Use detector pixel wavelengths
  • Keep history of operations on the data
  • (Support undoing of scaling and normalizations)

1.11 Some recent enhancements

• Accurate day and night lengths and position of the sun back in history and into the future
• Support for long time series of spectra
• Fast acquisition of spectra with Ocean Optics spectrometers
• Faster computations: possible to compute summaries for many 1000s of spectra

1.12 Packages

library(photobiology)
library(photobiologyWavebands)
library(photobiologyPlants)
library(photobiologyLEDs)
library(ggspectra)
set_theme(theme_bw(14))

2.1 Light spectra \(f(\lambda)\)

2.2 Spectra: energy vs. photons (= quanta)

• Energy irradiance \(E_\lambda\) in \(W\,m^{-2}\,nm^{-1}\)
• Photon irradiance \(Q_\lambda\) in \(mol\,s^{-1}\,m^{-2}\,nm^{-1}\)
• … conversion based on energy of one mole of photons \[q^\prime = h\cdot N_\mathrm{A} \cdot \frac{c}{\lambda}\] where \(\lambda\) is wavelength, and \(c\), speed of light, \(h\), Planck’s constant and \(N_\mathrm{A}\) Avogadro’s number are constants.
• functions e2q() and q2e() in R package ‘photobiology’

3.1 Data and functions used for plots

• Spectral data from package ‘photobiology’
  • sun.spct spectral irradiance
  • sun_daily.spct spectral daily exposure
• Methods from package ‘ggspectra’ (‘ggplot2’)
  • autoplot()

3.2 Sunlight: energy

autoplot(sun.spct, unit.out = "energy")

3.3 Sunlight: photons

autoplot(sun.spct, unit.out = "photon")

3.5 Sunlight daily: photons

autoplot(sun_daily.spct, unit.out = "photon")

4.1 Data and functions used for plots

• Spectral data from package ‘photobiology’
  • sun.spct spectral irradiance
  • sun_daily.spct spectral daily exposure
• Methods from package ‘photobioloy’
  • e_irrad()
  • q_irrad()
  • waveband()

4.2 Irradiance: energy vs. photons (= quanta)

• Energy irradiance \(E\) in \(W\,m^{-2}\)
• Photon irradiance \(Q\) in \(mol\,s^{-1}\,m^{-2}\)
• … no direct conversion possible!
• \(\to\) compute \(E\) and \(Q\) from spectrum

4.3 (Energy) Irradiance \(E\)

\[E = \int_{\lambda_1}^{\lambda_2} E_\lambda\ d\,\lambda\] where the wavelength range \(\lambda_1\) to \(\lambda_1\) is the waveband to integrate.

4.5 Photon Irradiance (=PFD) \(Q\)

\[Q = \int_{\lambda_1}^{\lambda_2} Q_\lambda\ d\,\lambda\] where the wavelength range \(\lambda_1\) to \(\lambda_1\) is the waveband to integrate.

5.1 Ratios: energy vs. photons (= quanta)

• Energy ratio \(E:E\) \(\neq\) photon ratio \(Q:Q\)
• Unitless
• shape of the spectrum matters
• … no direct conversion possible!
  • compute from spectrum
  • functions e_ratio() and q_ratio()
  • functions eq_ratio() and qe_ratio()

8.1 Sunlight below a yellow filter

filtered_sun.spct <- sun.spct * yellow_gel.spct
autoplot(filtered_sun.spct, unit.out = "photon")

8.2 Two LED spectra

Normalized spectra so that \(\mathrm{max} Q_\lambda = 1\).

autoplot(leds.mspct[c("Nichia_NF2W757GT_F1_sm505_Rfc00",
                      "LedEngin_LZ1_10R302_740nm")],
  unit.out = "photon", norm = "max") +
  theme(legend.position = "none")

8.4 Plot computed spectrum

autoplot(combined.spct, unit.out = "photon")

8.6 Red to far-red photon ratio (R:FR)

R_FR(combined.spct)

R:FR[q:q] 
 1.104595 
attr(,"radiation.unit")
[1] "q:q ratio"

R_FR(white_led_200)

R:FR[q:q] 
  4.62355 
attr(,"radiation.unit")
[1] "q:q ratio"

8.7 Phytochrome photoequilibrium

Pfr_Ptot(combined.spct)

[1] 0.7181451

Pfr_Ptot(white_led_200)

[1] 0.7776205

9.1 Resources

This presentation is a “web page” and the editable ‘.qmd’ markdown source file is at github.