Why we choose the CV, RMSE and r to evaluate accuracy of crop simulation models? What information do they tell us? This would be helpful in the interpretation?
You can use CV to understand how dispersed your data is. You can also use SD for a single data set. But you cannot just depend on SD when doing a comparison between two data sets. CV helps you compare them by eliminating the effect of the different means by taking the ratio of SD to the mean.
RMSE is a measure of how well your model fits to the data you have. It tells you the average error rate of your model.
r gives information about the LINEAR relation of two variables. It is always between -1 and 1. The further you go from 0, the stronger the correlation is, whether it is positive or negative.