1. Introduction
In recent years, protected horticulture has experienced remarkable growth, substantially increasing the cultivated area [
1,
2]. China boasts over 2.8 million hectares of horticultural facilities, representing over 80% of the world’s total. This area includes 810,000 hectares (29%) of solariums and 1.52 million hectares (53%) of large- and medium-sized greenhouses [
3]. These figures underscore the significant and growing role of facility agriculture in meeting the world’s food needs, and the supply of vegetables is crucial while promoting sustainable development in facility agriculture [
4]. Nevertheless, the expansion of facility agriculture also means a rise in water usage for agricultural purposes. As water scarcity becomes a growing problem worldwide, it challenges the sustainable development of facility agriculture [
5,
6]. By 2050, agricultural water scarcity is expected to spike in over 80% of the world’s countries [
7]. This leaves only a limited scope for increasing the water supply in agriculture, underscoring the need to enhance agricultural water use efficiency. The accurate calculation of the required crop water, also known as crop evapotranspiration (ETc), can help optimize irrigation management and augment water use efficiency during the growing season [
8].
Tomatoes are a vital vegetable crop grown worldwide due to their easy cultivation and high economic efficiency [
9,
10,
11]. To ensure that tomatoes grow and develop optimally, they must have an adequate water supply [
12]. However, in traditional tomato production facilities such as solar and plastic greenhouses, irrigation is mainly based on artificial experience, which often results in significant water wastage [
13]. The leaf area (LA) is the primary organ responsible for transpiration in tomato growth, and the leaf area index (LAI) is the total tomato leaf area per unit of land area, which is directly related to the crop evapotranspiration (ETc) of tomatoes [
14]. Moreover, the plant’s height is an essential indicator for assessing crop evapotranspiration (ETc) and transpiration in tomatoes, which is necessary for a high and stable tomato production [
15,
16]. Therefore, the real-time and accurate determination of the tomato’s LAI and plant height is crucial for estimating crop evapotranspiration (ETc), which, in turn, provides a theoretical basis for evaluating the tomato water demand, making irrigation decisions, and reducing water wastage. Accurate estimates of crop evapotranspiration help greenhouse managers make irrigation decisions and promote the efficient use and conservation of resources [
17].
Many experts and scholars have conducted a lot of research on the accurate determination of leaf area and plant height in tomatoes. Traditional leaf area measurements usually use destructive sampling methods to collect and determine the leaves, such as the punching and weighing method [
18], Image J image processing method [
19], leaf area meter method [
20], etc. Although these type of methods can determine the leaf area more accurately, they destroy the normal growth of tomato plants, and the workload is large and time consuming. In order to determine the leaf area nondestructively, leaf area simulation has been studied by many scholars. L. Bacci et al. [
21] improved the TOMGRO model to increase the simulation accuracy of leaf area; Wang X. et al. [
22] developed a leaf area simulation model for tomato processing based on the physiological development time, using the physiological development time of tomatoes as the time scale; Wang D. et al. [
23] developed a simulation model of pepper leaf area applicable to solar greenhouses using auxiliary heat product as a scale. Traditional plant height measurements are usually conducted directly using a tape measure, to save labor costs. Chang Yibo et al. [
24] used a binary quadratic orthogonal rotated combinatorial design to establish a logistic model for tomato plant height based on radial heat product with an irrigation limit and fertilizer application as determinants; Cheng Chen et al. [
25] constructed a celery plant height simulation model based on the relationship between greenhouse celery plant height and key meteorological factors (air temperature and solar radiation) with the single-plant irradiated heat product as the independent variable; Zhai Zihe et al. [
26] used the XGBoost model to establish a simulation model for the plant height growth of cucumber at five fertility periods in a solar greenhouse, and the accuracy of the model was good. The LAI and plant height can be more accurately simulated through crop growth modeling, but the model requires the input of multiple environmental parameters, and the lack of costly IoT environmental monitoring equipment in most traditional greenhouses makes these types of models less practical.
Therefore, through the real-time determination of tomato leaf area and plant height, a scientific foundation was established to assess tomato crop water requirements and make irrigation decisions in conventional greenhouses. This study’s mathematical simulation model of the tomato leaf area index and plant height was constructed using multiple linear regression based on experimental data from insulated plastic greenhouse fall stubble (the leaf area index was calculated from leaf area simulation results). The model was validated with experimental data from insulated plastic greenhouse spring stubble and solar greenhouse fall stubble. Additionally, a real-time monitoring device for the aboveground weight of tomatoes was designed to analyze the distribution coefficients of aboveground stems and leaves among different stubbles. This information was used to establish a mathematical simulation model of the tomato leaf area index and plant height based on real-time fresh weight, which was further tested with experimental data from overwinter stubble in solar greenhouses and continuous plastic greenhouses. Furthermore, a low-cost and stable operation device for the real-time simulation of the leaf area index and plant height of tomatoes applicable to traditional greenhouses was developed and installed. The aim was to provide a tool for the nondestructive determination of individual or group tomato leaf areas and plant heights in traditional greenhouses while offering scientific references for estimating the evapotranspiration of tomato crops and water-saving irrigation.
4. Discussion
Facility microclimate plays a crucial role in tomato growth, characterized by intricate parameters and rapid fluctuations [
30]. Transpiration is essential for assessing crop responses to microclimate changes caused by water stress, aiming to optimize the growing environment and crop breeding [
31]. The growth and development of tomatoes are influenced by various environmental factors such as air temperature, relative humidity, and solar radiation [
32,
33]. To minimize simulation errors resulting from environmental conditions, this study utilized fresh weight measurements of different morphological organs in tomatoes to simulate plant height and leaf area index. Subsequently, the distribution indices of the stems and leaves were combined with the real-time monitoring of the aboveground fresh weight to establish a simulation model based on the real-time quantity of fresh tomatoes’ plant height and leaf area index. This model was validated in different forks (varieties) and types of greenhouses. The simulated values for the plant height and leaf area index closely matched the measured values, indicating that the constructed simulation models have a high accuracy and applicability to traditional Chinese greenhouses (solar greenhouses and insulated plastic greenhouses). However, further optimization is required for commercially produced continuous plastic greenhouses equipped with greenhouse environmental control equipment.
In past studies on crop growth modeling, experts relied on the destructive sampling method to acquire crop growth index data [
34,
35,
36]. However, this method needed to be improved upon in monitoring the same plant over an extended period, resulting in a sparse data collection and errors during model construction. This study utilized a weighing device to monitor plant growth continuously, while agricultural IoT equipment was used to visualize and analyze the data online. In a study by Liu et al. [
22] on tomato appearance and morphology in insulated plastic greenhouses scaled through an irradiation heat product, it was found that the plant height had an RMSE of 13.66 cm. At the same time, the LAI was 1.03 m
2·m
−2. In this experiment, the average RMSE of the plant height was 10.81 cm, and the LAI was 0.55 m
2·m
−2. Overall, the findings indicate that the model developed in this study closely aligns with the simulation model based on irradiation heat accumulation. Notably, it requires monitoring only the fresh weight of fresh plants without necessitating the installation of environmental monitoring equipment. This feature makes it well suited for adoption in conventional greenhouses that lack sophisticated technology, enhancing its practicality.
The simulation results for the three methods were similar. All of the models simulated the height of tomato plants. The winter stubble in the solar greenhouse was more accurate than the spring stubble in the plastic greenhouse. The latter greenhouse did not have a back wall. This made it less accurate at preserving and storing heat. The second greenhouse’s simulation may not handle cold temperatures well. This can cause slight inaccuracies. This can prevent tomato plants from growing well and make the simulated and real plant heights vary. The simulation accuracy of the leaf area index of overwintering tomatoes in the solar greenhouse is better than that of spring stubble in a thermal insulation plastic greenhouse. The reason for the error may be that the solar greenhouse’s thermal insulation and heat storage capacity are more robust than those of the thermal insulation plastic greenhouse. The growth of tomato leaves is inhibited under low-temperature conditions, and the leaves are prone to disease and aging. When collecting leaf area data, the diseased or aging leaves had been removed by managers through agricultural operations, resulting in an inevitable simulation error. The superposition effect between the two led to the error between the simulated and measured tomato leaf area index values. Nevertheless, it needs to be more accurate in the multispan plastic greenhouse. The accuracy difference might be because the tomato plants were still seedlings in the solar greenhouse. The simulation is more accurate because these plants were mainly growing stems and leaves.
In the actual validation, it was found that the distribution indices of the tomato morphological organs were greatly affected by the environment, and the measured values of daylight greenhouse and continuous plastic greenhouse were generally higher than the simulated values because the environments inside their chambers were more stable and more suitable for tomato growth. When long-term low-temperature, low-light, and other growth adversities occur, the dry matter allocation of tomato stems and leaves will be inhibited to different degrees, which is consistent with the results of Gao et al. [
37]. The distribution coefficient significantly affects the difference between the simulated results and the measured values of the tomato plant height and leaf area index. In addition to time factors, environmental factors, water and fertilizer measures, etc., will specifically impact the tomato distribution index [
38,
39]. This study considered only the linear coupling between the fresh weights of tomato organs and the plant height and leaf area index, and the influence of nonlinear coupling on the final research results remains to be further investigated.
For all experiments conducted in this study, we employed the substrate bag cultivation method. In contrast, growers in traditional Chinese greenhouses mainly use soil cultivation, which can affect tomato growth differently under different cultivation conditions [
40,
41,
42]. In order to improve the applicability of the model in traditional Chinese greenhouses, the model will be validated in the future in different types of greenhouses, using different cultivation methods, and in different climatic zones, and based on this, the model will contribute to the future development of tomato-related research on water demand and irrigation decision-making devices.
5. Conclusions
The study suggests a new way for monitoring the greenhouse tomato plant height and leaf area index. This method is nondestructive and based on real-time simulation. It offers a fresh approach to building greenhouse tomato simulations. It also helps estimate tomato crop transpiration and create water-saving irrigation strategies as follows:
(1) In this study, we developed and installed a long-term, real-time monitoring device for tomato aboveground wire volume, which monitors the aboveground weight of tomato plants in real time without disrupting the average growth of tomatoes, with a coefficient of determination R2 of 0.937, RMSE of 0.05 kg, and MAE of 0.04 kg;
(2) Simulation models of the tomato plant height and leaf area index based on real-time weight were constructed, through which the predicted values of the tomato plant height and leaf area index in different greenhouses were estimated to fit well with the measured values. The average coefficient of determination R2 in the simulation of the plant height was 0.817, the RMSE was 10.81 cm, and the integrated simulation effect of the linear function was good; the average coefficient of determination R2 in the simulation of the leaf area index was 0.854, the RMSE was 0.55 m2·m−2, and the polynomial function simulation was better. The model can be used to estimate the plant height and leaf area index in real time, which provides a new way of thinking for the construction of a simulation model of tomato growth indices in facilities and also establishes a particular foundation for the estimation of tomato evapotranspiration and the formulation of water-saving irrigation strategies.