Skip to content

Rescale curves to survey points

Response curves are often measured on a few leaves, while a larger survey measures A on many leaves under one fixed condition. Rescale mode puts the curves on the survey's level of photosynthesis. It keeps each curve's shape and scales its amplitude, then refits.

How it works

  1. Fit the curves as usual.
  2. Take the survey's mean light and leaf temperature as the reference condition. The reference Ci is the highest Ci in the curve data, or the value you set with --ref-ci.
  3. Predict A at that condition from the fit.
  4. Compute the factor: the survey A statistic divided by the predicted A, using --stat mean (default), median or max.
  5. Multiply every measured A by the factor and fit again.

Steps

examples/survey.csv holds eight survey points with columns A, Ci, Qabs and Tleaf:

pk fit photosynthesis examples/aci.csv --model fvcb --set gm=1000000 --fix gm,alphaG --survey examples/survey.csv

The result has a rescale section:

"rescale": {
  "factor": 1.296502,
  "refCi": 1602.84,
  "refQ": 1500,
  "refT": 301.15,
  "stat": "mean",
  "surveyPoints": 8,
  "firstPass": { "r2": ..., "rmse": ..., "params": { ... } }
}

factor is the multiplier applied to A. firstPass holds the fit before rescaling. params, r2 and rmse describe the rescaled fit.

Give several survey files by repeating --survey. Their points are pooled. With several curve files, each curve is rescaled by its own factor.