-
The legume module, much like other crop modules within APSIM emulates the growth, development, yield, and nitrogen accumulation of legumes in response to various environmental factors such as temperature, radiation, photoperiod, soil water, and nitrogen availability. The principles employed in APSIM-legume have been expanded upon from the original modeling approaches introduced by Ritchie[15] and subsequently summarized by Ritchie[16]. The model operates on a daily time-step and has been specifically designed to simulate a homogeneous field, predicting grain yield, crop biomass, crop nitrogen absorption and fixation, as well as the distribution of resources within the plant on a spatial basis. The methodologies employed in modeling crop processes aim to strike a balance between providing a comprehensive depiction of the observed variations in crop performance across diverse production environments and minimizing the number of parameters that are difficult to quantify[17].
When parameterizing a generic model for legume species, particularly the common bean, it proves beneficial to introduce the concept of essential, desirable, and optional parameters. Essential parameters are those that exhibit the greatest sensitivity within the model and often differ significantly across various species. Consequently, these parameters must be determined through experimentation for each species[18]. Examples of essential parameters include the phyllochron, radiation extinction coefficient, and radiation-use efficiency. While it is sometimes possible to define essential parameters based on published sources, due to their critical nature, it is often essential to verify the value of such parameters through local experimentation using commonly cultivated varieties. The development of the common bean module can be attributed to Peter Carberry and Michael Robertson, and their work is elaborated upon in the paper authored by Robertson et al.[19].
Calibration and validation of the common bean-legume model
-
The calibration of the model was performed by utilizing measured data obtained from a two-year field experiment that was carried out during the growing seasons of 2020 and 2021. The experimental design employed was a split plot, based on a randomized complete block design with four replications, and it was conducted at the Agricultural Research Station of Shahid Bahonar University of Kerman in Iran, located at the coordinates 32.38° N and 51.40° E. The primary factor considered in the experiment was irrigation, which consisted of three levels: 80%, 60%, and 40% of the field capacity. Additionally, the sub-factor taken into account was the maturity stage, with early and late-maturity being the respective categories.
The date of planting was the 10th of May for both years. Each experimental plot consisted of a length of 6 m and four rows with a spacing of 50 cm between rows, and a plant density of 30 plants per square meter. The soil texture was classified as silty loam. Irrigation was carried out using a drip system. The amount of water applied to each plot was measured using a counter. All plots were irrigated uniformly every four days until the four-leaf stage. After this stage, irrigation was adjusted based on the field capacity using the weight method. The common bean crop was harvested on the 10th of July and the 1st of July in the years 2020 and 2021, respectively. To determine the grain yield, the final harvest involved the two middle rows (10 m2) at physiological maturity (with a humidity level of 12%). The field data were used to estimate genotype-specific parameters for the model (as shown in Table 1). The model was calibrated using a trial and error approach, aiming to minimize the difference between observed and simulated values. Parameters that had a significant impact on dry matter and leaf area index (LAI) were adjusted accordingly. This process was done until the model's simulated values closely matched the observed values for all treatments[20].
Table 1. Cultivar-specific parameters for a early- and late-maturity common bean cultivars.
Parameter Units Parameter description Late-maturity
cultivarEarly-maturity
cultivarCardinal temperatures for thermal time calculation °C Base temperature 7.5 7.5 Optimum temperature 30 30 Maximum temperature 40 40 Shoot_rate Degree-days/mm Thermal time required per mm elongation by young shoot before emergence 0.6 0.56 node_app_rate Degree-days Thermal time required for node appearance on main stem 100 92 Leaves_per_node lf/node No. of leaves per plant per main stem node 2 2 Extinction_coef Extinction coefficient (at default row spacing) 0.40 0.40 Rue g/MJ Radiation-use efficiency 0.94 0.94 frac_leaf_pre_flower Fraction allocated to leaves pre-flowering 0.55 0.60 frac_leaf_grain_fill Fraction allocated to leaves in grain fill 0.30 0.35 frac_stem2 pod Fraction allocated to pod before grain fill 0.46 0.49 frac_pod2 grain Fraction allocated to pod relative to grain during grain fill 0.28 0.28 n_conc_crit g/g Critical nitrogen concentration of grain 0.045 0.045 Specific_root_length mm/g Specific root length 65000 65000 Trans_eff_coef Pa Transpiration efficiency coefficient 0.0055 0.0055 To calibrate and validate the water content, soil moisture content data collected from two field experiments conducted in 2020 and 2021 were utilized. To validate the crop model, the dry matter, LAI, and grain yield traits acquired from these field experiments were taken into account. To assess the variances between the observed and simulated data, various metrics were employed[21].
These metrics included the coefficient of determination (R2), the 1:1 line, the normalized root mean square error (nRMSE) as proposed by Wallach & Goffinet[22], and the model efficiency (EF) as defined by Willmott[23]. These metrics were utilized to compare the observed and simulated values of key parameters such as grain yield, dry matter, leaf area index (LAI), and soil moisture content.
$ nRMS E\;\left({\text{%}}\right)=\sqrt{\dfrac{\sum _{i=1}^{n}{\left({S}_{i}-{O}_{i}\right)}^{2}}{n}}\times \dfrac{100}{\overline{O}} $ (1) $ \mathrm{E}\mathrm{F}=1-\dfrac{\sum _{\mathrm{i}=1}^{\mathrm{n}}{\left(\mathrm{S}\mathrm{i}-\mathrm{O}\mathrm{i}\right)}^{2}}{\sum _{\mathrm{i}=1}^{\mathrm{n}}{\left(\overline{o}-\mathrm{O}\mathrm{i}\right)}^{2}} $ (2) Where,
, O, S, and n represent the average observed data, observed data, simulated data, and number of observations, respectively. The accuracy of the model improves as the nRMSE value approaches zero. Additionally, a value of one for EF (model efficiency) indicates that the simulated values better depict the trend in the measured data compared to the average of the observations[24]. Furthermore, a higher R2 value nearing unity demonstrates that the model accurately replicates reality.$\overline O $ -
We are grateful to the Department of Agriculture and Plant Breeding of Shahid Bahonar University of Kerman, as well as the General Directorate of Meteorology Organization, Agricultural Research Center, and Agricultural Jihad Organization of Kerman Province for their assistance in conducting the present research.
-
About this article
Cite this article
Amiri S, Zakeri N, Yousefi T. 2024. Evaluation of the APSIM common bean model using different cultivars and water-management scenarios. Technology in Agronomy 4: e017 doi: 10.48130/tia-0024-0015
Evaluation of the APSIM common bean model using different cultivars and water-management scenarios
- Received: 15 January 2024
- Accepted: 15 May 2024
- Published online: 04 July 2024
Abstract: The common bean (Phaseolus vulgaris) is a widely consumed legume worldwide and holds significant value in terms of direct human consumption, surpassing all other legume crops combined. This study aimed to assess the applicability of the APSIM common bean model under different water-management conditions and cultivars in southeastern Iran. A two-year field experiment was conducted at the Agricultural Research Station of Shahid Bahonar University of Kerman, Iran, spanning from 2020 to 2021. The experiment followed a split-plot design with a randomized complete block structure and included four replications. The primary irrigation factor consisted of three levels, representing 80%, 60%, and 40% of crop capacity, while the secondary factor encompassed early and late-maturity cultivars. To evaluate the model's performance, simulated and measured grain yield, total dry matter, leaf area index (LAI), and soil water content were compared using the adjusted coefficient of correlation, normalized root mean square errors (nRMSE), and model efficiency (EF). The results indicated a satisfactory agreement between predicted and observed grain yield (nRMSE = 12% and EF = 0.92). Similarly, the agreement between simulated and observed total dry matter was reasonable (nRMSE = 13% and EF = 0.83). The observed and predicted soil water content also exhibited good agreement. Consequently, the APSIM common bean model proves suitable for research purposes, particularly in the areas of irrigation and cultivar selection.
-
Key words:
- Grain yield /
- Irrigation /
- LAI /
- Soil water content