Skip to contents

Fitting a model to concentration data and estimating the slope

Source: R/flux_fitting.R
flux_fitting.Rd

fits gas concentration over time data with a model (exponential, quadratic or linear) and provides the slope later used to calculate gas fluxes with flux_calc

Usage

flux_fitting(
  conc_df,
  conc_col,
  datetime_col,
  f_start = f_start,
  f_end = f_end,
  f_fluxid = f_fluxid,
  start_cut = 0,
  end_cut = 0,
  t_window = 20,
  cz_window = 15,
  b_window = 10,
  a_window = 10,
  roll_width = 15,
  t_zero = 0,
  fit_type
)

Arguments

conc_df

dataframe of gas concentration over time

conc_col

column with gas concentration data

datetime_col

column with datetime of each concentration measurement Note that if there are duplicated datetime in the same f_fluxid only the first row will be kept

f_start

column with datetime when the measurement started

f_end

column with datetime when the measurement ended

f_fluxid

column with ID of each flux

start_cut

time to discard at the start of the measurements (in seconds)

end_cut

time to discard at the end of the measurements (in seconds)

t_window

enlarge focus window before and after tmin and tmax (exponential fit)

cz_window

window used to calculate Cz, at the beginning of cut window (exponential fit)

b_window

window to estimate b. It is an interval after tz where it is assumed that the model fits the data perfectly (exponential fit)

a_window

window at the end of the flux to estimate a (exponential fit)

roll_width

width of the rolling mean for CO2 when looking for tz, ideally same as cz_window (exponential fit)

t_zero

time at which the slope should be calculated (for quadratic fit)

fit_type

exponential, quadratic or linear. Exponential is using the exponential model from Zhao et al (2018)

Value

a dataframe with the slope at t zero (f_slope), a datetime column of t zero (f_start_z), a factor column indicating the cuts (f_cut), the time in seconds since the start of the measurement (f_time), the modeled fit (f_fit), the modeled slope (f_fit_slope), the parameters of the fit depending on the model used, and any columns present in the input. The type of fit is added as an attribute for use by the other functions.

References

Zhao, P., Hammerle, A., Zeeman, M., Wohlfahrt, G., 2018. On the calculation of daytime CO2 fluxes measured by automated closed transparent chambers. Agricultural and Forest Meteorology 263, 267–275. https://doi.org/10.1016/j.agrformet.2018.08.022

Examples

data(co2_conc)
flux_fitting(co2_conc, conc, datetime, fit_type = "exp")
#> Cutting measurements...
#> Estimating starting parameters for optimization...
#> Optimizing fitting parameters...
#> Calculating fits and slopes...
#> Done.
#> Warning: 
#>  fluxID 5 : slope was estimated on 205 points out of 210 seconds
#>  fluxID 6 : slope was estimated on 206 points out of 210 seconds
#> # A tibble: 1,251 × 23
#>    datetime            temp_air temp_soil  conc   PAR turfID       type 
#>    <dttm>                 <dbl>     <dbl> <dbl> <dbl> <fct>        <fct>
#>  1 2022-07-28 23:43:35    NA         NA    447. NA    156 AN2C 156 ER   
#>  2 2022-07-28 23:43:36     7.22      10.9  447.  1.68 156 AN2C 156 ER   
#>  3 2022-07-28 23:43:37    NA         NA    448. NA    156 AN2C 156 ER   
#>  4 2022-07-28 23:43:38    NA         NA    449. NA    156 AN2C 156 ER   
#>  5 2022-07-28 23:43:39    NA         NA    449. NA    156 AN2C 156 ER   
#>  6 2022-07-28 23:43:40    NA         NA    450. NA    156 AN2C 156 ER   
#>  7 2022-07-28 23:43:41    NA         NA    451. NA    156 AN2C 156 ER   
#>  8 2022-07-28 23:43:42    NA         NA    451. NA    156 AN2C 156 ER   
#>  9 2022-07-28 23:43:43    NA         NA    453. NA    156 AN2C 156 ER   
#> 10 2022-07-28 23:43:44    NA         NA    453. NA    156 AN2C 156 ER   
#> # ℹ 1,241 more rows
#> # ℹ 16 more variables: f_start <dttm>, f_end <dttm>, f_fluxid <fct>,
#> #   f_ratio <dbl>, f_flag_match <chr>, f_time <dbl>, f_cut <fct>, f_Cz <dbl>,
#> #   f_Cm <dbl>, f_a <dbl>, f_b <dbl>, f_tz <dbl>, f_slope <dbl>, f_fit <dbl>,
#> #   f_fit_slope <dbl>, f_start_z <dttm>
flux_fitting(co2_conc, conc, datetime,  fit_type = "quadratic",
t_zero = 10, end_cut = 30)
#> # A tibble: 1,251 × 24
#>    datetime            temp_air temp_soil  conc   PAR turfID       type 
#>    <dttm>                 <dbl>     <dbl> <dbl> <dbl> <fct>        <fct>
#>  1 2022-07-28 23:43:35    NA         NA    447. NA    156 AN2C 156 ER   
#>  2 2022-07-28 23:43:36     7.22      10.9  447.  1.68 156 AN2C 156 ER   
#>  3 2022-07-28 23:43:37    NA         NA    448. NA    156 AN2C 156 ER   
#>  4 2022-07-28 23:43:38    NA         NA    449. NA    156 AN2C 156 ER   
#>  5 2022-07-28 23:43:39    NA         NA    449. NA    156 AN2C 156 ER   
#>  6 2022-07-28 23:43:40    NA         NA    450. NA    156 AN2C 156 ER   
#>  7 2022-07-28 23:43:41    NA         NA    451. NA    156 AN2C 156 ER   
#>  8 2022-07-28 23:43:42    NA         NA    451. NA    156 AN2C 156 ER   
#>  9 2022-07-28 23:43:43    NA         NA    453. NA    156 AN2C 156 ER   
#> 10 2022-07-28 23:43:44    NA         NA    453. NA    156 AN2C 156 ER   
#> # ℹ 1,241 more rows
#> # ℹ 17 more variables: f_start <dttm>, f_end <dttm>, f_fluxid <fct>,
#> #   f_ratio <dbl>, f_flag_match <chr>, f_time <dbl>, f_cut <fct>,
#> #   f_param1 <dbl>, f_param2 <dbl>, f_rsquared <dbl>, f_adj_rsquared <dbl>,
#> #   f_intercept <dbl>, f_pvalue <dbl>, f_slope <dbl>, f_fit <dbl>,
#> #   f_fit_slope <dbl>, f_start_z <dttm>