TOMGRO Model Research Articles

Evaluation of the cooling, heating and water demands for tomato production in a solar greenhouse by integrating the solar energy, evapotranspiration and TOMGRO models

The design, operation and maintenance of solar greenhouses in optimum condition can improve the water and energy consumption and agricultural production. This study has been integrated the energy, humidity and growth models to determine the cooling, heating and water requirements for producing tomato in the even span, uneven span and arc roof type greenhouse structures with north–south and east–west orientations. In the proposed integrated framework, firstly, the incident and transmitted diffuse, direct and ground solar radiations for any side of solar greenhouse, along with different evapotranspiration models and TOMGRO model for tomato growth and production have been assessed comprehensively. Then, by using the set point day and night temperatures and relative humidity the phenomena of the energy and water transfer such as convection, longwave radiation, infiltration, and cover condensation as well as TOMGRO state variables have been calculated. By considering the temperature, humidity, structure and orientation as decision variables, the east–west uneven span structure with day/night temperatures of 20/16 °C and relative humidity of 80 % is recommended as the best case in Pakdasht city of Iran. This solar greenhouse requires 0.92 MW/m2 and 3.83 MW/m2 heating and cooling energy, respectively, to produce 46.68 kg/m2 mature tomatoes during the whole cultivation period. This research has provided a tool to select the most appropriate structure and orientation of a solar greenhouse in any location based on the trade-off between energy and water consumption and tomato production.

CAN ACCURACY ISSUES OF LOW-COST SENSOR MEASUREMENTS BE OVERCOME WITH DATA ASSIMILATION?

The use of mechanistic plant growth models relies on the availability of high-quality inputs to reduce uncertainty in estimates. Measurements of photosynthetically active radiation inside a protected environment are either more expensive to obtain or dependent on assumptions regarding external measurements. This study aimed to reduce the influence of uncertainty in the measurements of low-cost lux meters by using a data assimilation strategy. We first determined, by simulation, the impact of different sensors on the estimates. We then used the Ensemble Kalman Filter to assimilate artificial observations of tomato growth in the Reduced-State Tomgro model, in simulations for which the solar radiation inputs were obtained from a low-cost lux meter. We compared the assimilated estimates to the simulations that used solar radiation obtained with a scientific-grade quantum sensor. For periods of larger radiation intensity, in which the differences in measurements from both instruments are larger, assimilation of observations with low errors lead to estimates that are closer to the ones obtained by scientific grade sensors. These results suggest that low-cost sensors could be used to obtain inputs for growth models in protected environments, provided there are also imperfect observations of the state.

A Novel Model Fusion Approach for Greenhouse Crop Yield Prediction

In this work, we have proposed a novel methodology for greenhouse tomato yield prediction, which is based on a hybrid of an explanatory biophysical model—the Tomgro model, and a machine learning model called CNN-RNN. The Tomgro and CNN-RNN models are calibrated/trained for predicting tomato yields while different fusion approaches (linear, Bayesian, neural network, random forest and gradient boosting) are exploited for fusing the prediction result of individual models for obtaining the final prediction results. The experimental results have shown that the model fusion approach achieves more accurate prediction results than the explanatory biophysical model or the machine learning model. Moreover, out of different model fusion approaches, the neural network one produced the most accurate tomato prediction results, with means and standard deviations of root mean square error (RMSE), r2-coefficient, Nash-Sutcliffe efficiency (NSE) and percent bias (PBIAS) being 17.69 ± 3.47 g/m2, 0.9995 ± 0.0002, 0.9989 ± 0.0004 and 0.1791 ± 0.6837, respectively.

Studies of evolutionary algorithms for the reduced Tomgro model calibration for modelling tomato yields

The reduced Tomgro model is one of the popular biophysical models, which can reflect the actual growth process and model the yields of tomato-based on environmental parameters in a greenhouse. It is commonly integrated with the greenhouse environmental control system for optimally controlling environmental parameters to maximize the tomato growth/yields under acceptable energy consumption. In this work, we compare three mainstream evolutionary algorithms (genetic algorithm (GA), particle swarm optimization (PSO), and differential evolutionary (DE)) for calibrating the reduced Tomgro model, to model the tomato mature fruit dry matter (DM) weights. Different evolutionary algorithms have been applied to calibrate 14 key parameters of the reduced Tomgro model. And the performance of the calibrated Tomgro models based on different evolutionary algorithms has been evaluated based on three datasets obtained from a real tomato grower, with each dataset containing greenhouse environmental parameters (e.g., carbon dioxide concentration, temperature, photosynthetically active radiation (PAR)) and tomato yield information at a particular greenhouse for one year. Multiple metrics (root mean square errors (RMSEs), relative root mean square errors (r-RSMEs), and mean average errors (MAEs)) between actual DM weights and model-simulated ones for all three datasets, are used to validate the performance of calibrated reduced Tomgro model.

Modified TOMGRO outputs as guide factors to estimate evapotranspiration and water use efficiency of three tomato fresh cultivars, grown in a low-tech Italian glasshouse

In Low-Tech Italian greenhouses, the adoption of more rational criteria in resources management would lead to significant economic and environmental advantages for producers. Reliable crop growth and evapotranspiration models, may allow real time assessment of water and nutrient plant requirements and the evaluation of the productive system efficiency, supporting decision-making processes for short, long and medium-term strategies. This paper presents the results of an experimental trial carried out in 2011, in the research facilities of the DISAAA-a at Pisa University, to set a cross-control method for the evaluation of the Water Use Efficiency (WUE) and of the real plant water requirements. A modified version of TOMGRO model was used to simulate Leaf Area Index (LAI) and total dry biomass (shoot and fruits) of three different cultivars of tomato (Solanum lycopersicum L.), grown in the same environmental conditions. For each cultivar, the hourly values of LAI provided by the TOMGRO model were used as input in the Baille equation for the estimation of the evapotranspiration (ET) rate, finding a significant agreement between simulated and measured ET values (R2=0.8 on average). A good prediction of the WUE of each thesis was also obtained (R2>0.8),, offering interesting elements for plant performance evaluation. Considering these evidences, a synthetic index (estimated biomass/estimated ET) may be used for a direct comparison between different scenarios.

Global sensitivity analysis by means of EFAST and Sobol' methods and calibration of reduced state-variable TOMGRO model using genetic algorithms

Graphical abstractDisplay Omitted FAST and Sobol' sensitivity analysis methods were compared.Genetic algorithms were performed to calibrate reduced TOMGRO model.Reduced TOMGRO model calibration all...

Global sensitivity analysis by means of EFAST and Sobol’ methods and calibration of reduced state-variable TOMGRO model using genetic algorithms

One common constraint for using crop models for decision making in precise greenhouse crop management is the need for accurate values of model parameters depending on climate conditions, crop varieties, and management. Estimating these parameters from observed data on the crop, using a crop model, is an interesting possibility. Nevertheless, the accuracy of estimations depends on the sensitivity of the model output variables to the parameters. Therefore, this paper proposes the use of the reduced state variable TOMGRO model which describes nodes, leaf area index, total plant weight, total fruit weight, and mature fruit weight as states variables. The objective of this work was to compare EFAST and Sobol’ sensitivity analysis methods to determine the most sensitive parameters for TOMGRO model outputs. A former sensitivity analysis showed that 8 parameters were the most sensitive and they were calibrated using genetic algorithms (GAs) to adapt the model to semi-arid weather conditions of Central Mexico. Genetic algorithms are important heuristic search algorithms for optimization problems and have been used to calibrate non-linear models related to control of greenhouse climate conditions. Simulation and analysis of the TOMGRO model showed that the estimations for the state variables are close to the measured data. The model could be adapted for simulating other greenhouse crops by means of sensitivity analysis and calibration.

Evaluation and adaptation of TOMGRO model to Italian tomato protected crops

A simplified version of TOMGRO, a well-known model for tomato growth simulation, was delivered by its authors in 1999, with the aim to adapt the program to operational exigencies. This model version was chosen to integrate the functions of a decision support tool, oriented to support Italian tomato growers in soilless crop fertigation management. During the preliminary evaluation phase, its application on data collected in three greenhouse experiments conducted on tomato cv. Jama in Pisa (central Italy) in 2004 and 2005, showed an insufficient accuracy in the Leaf Area Index (LAI) estimation of tomato crops. Consequently, node number and LAI computational algorithms were modified. This paper discusses model modifications, analysing their effects on tomato growth simulation. In comparison with the observed data, the modified model showed a good enhancement of estimation accuracy of node and LAI, with significantly positive effects also on plant and fruit biomass estimation.

Mathematical modeling on tomato plants: A review

Mathematical models allow for predictions of behavior under specific handling and environmental conditions, and are particularly useful in expensive studies or in studies where long term effects may be difficult to monitor. In mathematical modeling there are two main types of models: descriptive models and mechanistic models; the first are relationships between response and predictor which are not ruled by biological processes; the latter takes into account the basic processes in plants by means of differential equations to account for the development of plants. This requires a deeper knowledge of the physiological development of plants. This work reviews mathematical modeling on tomato plant. The TOMGRO model is modular and has been widely studied and calibrated under several climatic conditions which demonstrates that it is a robust model. As a future research the TOMGRO model is proposed to be adapted to other crops. Key words: Differential equations, descriptive models, mechanistic models.

Simulation of transpiration, drainage, N uptake, nitrate leaching, and N uptake concentration in tomato grown in open substrate

Free-drainage or “open” substrate system used for vegetable production in greenhouses is associated with appreciable NO3− leaching losses and drainage volumes. Simulation models of crop N uptake, N leaching, water use and drainage of crops in these systems will be useful for crop and water resource management, and environmental assessment. This work (i) modified the TOMGRO model to simulate N uptake for tomato grown in greenhouses in SE Spain, (ii) modified the PrHo model to simulate transpiration of tomato grown in substrate and (iii) developed an aggregated model combining TOMGRO and PrHo to calculate N uptake concentrations and drainage NO3− concentration. The component models simulate NO3−–N leached by subtracting simulated N uptake from measured applied N, and drainage by subtracting simulated transpiration from measured irrigation. Three tomato crops grown sequentially in free-draining rock wool in a plastic greenhouse were used for calibration and validation. Measured daily transpiration was determined by the water balance method from daily measurements of irrigation and drainage. Measured N uptake was determined by N balance, using data of volumes and of concentrations of NO3− and NH4+ in applied nutrient solution and drainage. Accuracy of the two modified component models and aggregated model was assessed by comparing simulated to measured values using linear regression analysis, comparison of slope and intercept values of regression equations, and root mean squared error (RMSE) values. For the three crops, the modified TOMGRO provided accurate simulations of cumulative crop N uptake, (RMSE = 6.4, 1.9 and 2.6% of total N uptake) and NO3−–N leached (RMSE = 11.0, 10.3, and 6.1% of total NO3−–N leached). The modified PrHo provided accurate simulation of cumulative transpiration (RMSE = 4.3, 1.7 and 2.4% of total transpiration) and cumulative drainage (RMSE = 13.8, 6.9, 7.4% of total drainage). For the four cumulative parameters, slopes and intercepts of the linear regressions were mostly not statistically significant (P < 0.05) from one and zero, respectively, and coefficient of determination (r2) values were 0.96–0.98. Simulated values of total drainage volumes for the three crops were +21, +1 and −13% of measured total drainage volumes. The aggregated TOMGRO–PrHo model generally provided accurate simulation of crop N uptake concentration after 30–40 days of transplanting, with an average RMSE of approximately 2 mmol L−1. Simulated values of average NO3− concentration in drainage, obtained with the aggregated model, were −7, +18 and +31% of measured values.

ENVIRONMENTAL IMPACT OF GREENHOUSE TOMATO PRODUCTION STRATEGIES USING LIFE CYCLE ASSESSMENT APPROACH

In Colombia, greenhouse tomato (Solanum lycopersicum L.) production has been increasing during the last years. Greenhouse production has caused landscape changes, ecosystem flux variations and risings on energy inputs and residue generation. The aim of this work was to quantify the relative intensity of energy demand, resources consumption and pollutants emissions to air, soil and water as a consequence of the implementation of technological alternatives for greenhouse tomato production. Life cycle assessment (LCA) approach was applied to determine the environmental sustainability of tomato production chain. Three strategies were evaluated: soil production system under single layer plastic greenhouse (SSP), substrate production system under double layer plastic greenhouse (HDP) and HDP including heating (HDPH). The Tomgro model was used to determine biomass production for each crop cycle under the three strategies. Potential impact of infrastructure and its contribution to environmental loads were determined especially for categories: climate change, eutrophication and marine aquatic ecotoxicity. LCA results showed the lowest energy consumption for SSP strategy, followed by HDP, while HDPH registered the highest energy demands. The major impact of greenhouse tomato production on the aforementioned categories is related to leaching for SSP and HDP, while, for HDPH, fuel consumption causes the highest environmental impact. Biodegradable residues are of major importance since they lower the impacts for all categories. Water pollution and eutrophication are affected by high nutrient emissions. LCA approach and the use of simulation models allowed assessing the environmental impact of production strategies and possible effects of technological improvements on commercial production systems.

Use of geostatistical and crop growth modelling to assess the variability of greenhouse tomato yield caused by spatial temperature variations

Users of potential crop growth models have generally assumed homogenous greenhouse climate conditions for simulation purposes. Geostatistics offers the possibility to represent the spatial dependence of climate variables such as temperature distribution in greenhouses. In order to obtain insight in the relevance of temperature distribution on the performance of a crop, geostatistical and crop growth modeling tools were combined in the present work. A plastic greenhouse with the size used in commercial practices, was used to assess temperature distribution by installing a 25-sensor grid, during a 28-day measurement period. Geostatistical analyses were performed on the temperature data, previously divided in five ranges that were established as a function of solar radiation intensity. Estimated semivariograms were fitted to theoretical spherical model for posterior estimation at unsampled points by ordinary kriging method. Results of this analysis indicated an increasing temperature spatial dependence as global radiation augmented. Crop growth simulations with the Tomgro model applied to the estimated temperature distribution, quantified the faster plant development rates in the central zone of the greenhouse, resulting in plants with one additional truss and a significant higher yield, when compared to plants next to the side walls. The results of this work suggest that the location of the climate sensor station is a sensible factor when using crop simulation models for greenhouse conditions. Final predictions done by crop growth simulation models may be biased and the results cannot be generalized for the entire greenhouse area when microclimate patterns are not considered. Application of geostatistics enabled to assess temperature patterns inside greenhouse and to analyze the relationship with global outside radiation.

Sensitivity of the Tomgro Model to Solar Radiation Intensity, Air Temperature and Carbon Dioxide Concentration

In this study, variations on the output of the Tomgro model are analysed, as a function of variations in solar radiation intensity S, greenhouse air temperature TA and CO2 concentration CA, common for the production of greenhouse tomato in the Bogota Plateau, Colombia. The variation on the output of the model is apportioned to the sources of variation to obtain an analysis of sensitivity. Among the results of this test, a similar degree of variability on the prediction of fruit dry weight WF is detected for the three climate variables, while S and TA were the most sensitive climate variables. Fruit dry weight increased with S and CA, while for TA an optimal range is detected. Next, a factorial experiment is performed where the same climate variables are varied in a small range. Variance decomposition showed that S is the most sensitive climate variable for WF, followed by TA and CA. The development of the vegetative plant parts is more sensitive to TA than to S and CA. The two innovative techniques that are presented for assessing the sensitivity to climate conditions can be used for the selection of production locations or for optimising greenhouse design.

SENSITIVITY ANALYSES OF TOMGRO OUTPUT VARIABLES TO VARIATIONS IN CLIMATE CONDITIONS

The sensitivity of the second version of the modified TOMGRO model to climate is analyzed using a database that was compiled for the Bogota Plateau. The variability of an output variable is defined as the variation caused on that variable when the climate factor is varied in its estimated distribution space. On the other hand, the sensitivity is defined here as the variation caused on the output variable, per unitary variation on the climate factor. In a first test, the radiation intensity, air temperature and CO2 concentration are tested one-by-one within the possible range for different situations in the Bogota Plateau. The relative ranges of the slopes of the output variables are computed to quantify variability. The relation of the relative range of the output variable with the relative range of the respective climate factor is calculated to quantify the sensitivity of each climate factor. Among the results of this test, a similar degree of variability on the prediction of fruit dry weight is detected for the three climate factors, while the radiation intensity and temperature are the most sensitive climate factors. In a second test, the climate factors are varied in a small and similar range to ensure a linear response. A general linear model is applied to describe the slopes of each output variable as a function of the level of each climate factor. The contribution to the total variance apportions the variance in the output to the variance in the input and is a measure of how much each climate factor contributes to the total sensitivity of the model. This test shows that the solar radiation is the most sensitive climate factor for fruit weight, followed by temperature and CO2 concentration.

Validation of a Photosynthesis Model through the Use of the CO2Balance of a Greenhouse Tomato Canopy

Several leaf photosynthesis models were developed from well controlled experiments in growth chambers. However, only a few have been validated under greenhouse conditions for their quantitative and qualitative adequacy. In this paper, rates of net photosynthesis for a tomato crop (Lycopersicon esculentum Mill) were measured in a semi-commercial greenhouse (615 m3) for a significant time period. Concomitant measurements of climatic conditions and LAI were used for simulation of net photosynthesis using the TOMGRO model which integrates Acock's model for photosynthesis calculations. From simulations and from sensitivity analysis, the prediction of net photosynthesis appeared to be very sensitive to the quantum use efficiency. The Acock model with original parameters underestimated the net photosynthesis rate, but an increase in the quantum use efficiency by 10% gave a good fit. In an effort to generalize the validity of the model, a residual analysis was performed and showed a systematic bias related to light intensity intercepted by the canopy. The Marquardt algorithm was used to adjust our data to the model but did not eliminate residual heterogeneity of variance with new parameter values. On the basis of collected data, the criteria of goodness of fit used showed that the photosynthesis model is inadequate in describing the CO2-balance of the greenhouse agrosystem. However, it was determined that it could be used as a submodel within a more complex model for predicting growth and development.

Integration of a Crop Growth Model in a Greenhouse Management Software System

In greenhouses, computerized climate control systems can be used to dynamically and automatically regulate climatic and environmental parameters. To support such a mode of operation, a crop growth model that was developed for greenhouse tomato plants (TOMGRO) was chosen. The model describes the phenological development and increase in dry weight of various organs from planting till maturity under variable environmental conditions. The assimilate partitioning is based on sink strength. A FORTRAN version of TOMGRO was converted to the Smalltalk object-oriented programming environment. This model was integrated into GX, a dynamic climate management software system that was developed at Laval Univ. GX defines a general architecture that may accommodate different decision-making modules based on mathematical models and rule bases. The TOMGRO model is being used to evaluate different production scenarios and will be used to calculate and predict crop growth rates, development, and yields. The model can be used to perform real-time and seasonal cost–benefit analysis and for the dynamic optimization of greenhouse climatic parameters.