Testing the AWWEsmo scale accuracy

During the review of our paper, one of the reviewer mentioned that the scale that we are using (HX711) is not reliable across the temperatures that are typically used within the growth chambers. So we decided to test it.

First - we have tested the accuracy of the AWWEsmo device within the same growth chamber that it was calibrated in (GC1, temp 20C):

data <- read.csv("AWWEsmo.scale.calibration.csv")
head(data)
library(ggpubr)
## Loading required package: ggplot2
library(ggplot2)

ggscatter(data, x = "Lab.scale", y = "AWWEsmo.GC1", add = "reg.line")+
  stat_cor(method = "pearson", label.x = 3, label.y = 400)+ 
  stat_regline_equation(label.x = 3, label.y = 350)

As the regression line suggests, the correction factor within the same temperature is only 0.3 grams, which is within the acceptable range for watering accuracy.

Now - let’s see how the AWWEsmo accuracy is behaving across the environmental chambers with different temperatures:

library(reshape2)

mdata <- melt(data, id = c("Object", "Lab.scale"))
mdata
AWWEsmo_Fig <- ggscatter(mdata, x = "Lab.scale", y = "value", add = "reg.line", color = "variable", 
                         palette = "jco", fullrange = TRUE, ylab = "AWWEsmo weight (g)", 
                         xlab = "Lab scale weight (g)") 
AWWEsmo_Fig <- AWWEsmo_Fig + stat_cor(aes(color = variable), label.x = 3) 
AWWEsmo_Fig <- AWWEsmo_Fig + stat_regline_equation(aes(color = variable), label.x = 150) 
AWWEsmo_Fig

Although the overall correlation between the weights measured in GC2 (temp 18C) and GC3 (temp 26C) is still high with the weight recorded with the laboratory scale (R > 0.99), you can see that the regression lines that we are drawing are now having a correction factor of -16 to 32 g - which is quite substantial, and thus would not be a reliable to record evapotranspiration.

And thus - we include this as a supplemental figure - to show the device low reliability across the growth chamber temperatures, and advice to calibrate the AWWEsmo within the growth chamber conditions for reliable watering and weighing.

pdf("Fig.S1.AWESMO.scale.precision.pdf")
plot(AWWEsmo_Fig)
dev.off()
## quartz_off_screen 
##                 2