Models Calibration and Evaluation

Mukhtar Ahmed, Shakeel Ahmed, Muhammad Ali Raza, Uttam Kumar, Muhammad Ansar, Ghulam Abbas Shah, David Parsons, Gerrit Hoogenboom, Taru Palosuo, Sabine J. Seidel

Publikation: Bidrag til bog/antologi/rapport/proceedingBidrag til bog/antologiForskning

39 Citationer (Scopus)

Abstract

Calibration of crop model is standard practice, and it involves estimation of crop parameters based upon observed field data. It is the process of estimation of unknown parameters using practical observations. It is generally carried out manually by adjusting the parameters of the model. It consists of choosing the accurate numbers of coefficients that play a significant role in the adjustment of soil nitrogen, soil organic carbon, soil phosphorus, crop growth, phenological development, biomass accumulation, dry-matter partitioning, nutrients uptake, grain dry weight, grain numbers, grain yield, grain nitrogen (N) at maturity and protein content. A minimum data set (MDS) is required for the calibration of the model. A number of different steps could be used to calibrate the crop model. The initial step involves running of the model with default crop parameters and comparison of simulation outcomes with the observed data set. Afterwards, crop parameters are adjusted to have good agreement with observed and simulated data. It starts with phenology, then vegetative growth and biomass, afterwards yield components and finally yield. Optimization tools such as generalized likelihood uncertainty estimation (GLUE) and Markov chain Monte Carlo (MCMC) can be used for the calibration. Finally, evaluation of models needs to be performed with independent data set. The quantification of calibration and evaluation goodness should be evaluated by different skill scores such as root-mean-squared error (RMSE)/root-mean-square deviation (RMSD), relative RMSE (RRMSE) or normalized objective function (NOF), root-mean-square deviation-systematic error (RMSDse), root-mean-square deviation-non-systematic error (RMSDnse), mean absolute error (MAE), mean bias error (ME), coefficient of determination (R2), Nash-Sutcliffe modelling efficiency (EF) test, maximum difference (MD) and D index (index of agreement). These skill scores confirm the calibrated model performances under different sets of scenarios.

OriginalsprogEngelsk
TitelSystems Modeling
Antal sider28
ForlagSpringer
Publikationsdato2020
Sider151-178
StatusUdgivet - 2020
Udgivet eksterntJa

Fingeraftryk

Dyk ned i forskningsemnerne om 'Models Calibration and Evaluation'. Sammen danner de et unikt fingeraftryk.

Citationsformater