Mobile QR Code

1. (College of Information and Communication Technology (CICT), Can Tho University, Vietnam {hxhiep, hbman11, nthai.cit}@ctu.edu.vn )

Rice, Brown plant hopper, Growth and propagation of BPH, GAMA, DIGIPLANT

## 1. Introduction

Rice is an important food crop of agricultural countries, especially in the Mekong Delta, Vietnam. Rice cultivation has been a traditional profession of Vietnamese people since very ancient times. The advances of science and technology in the country and the world in research and rice production have strongly promoted our country’s rice-growing industry to catch up with the world’s advanced level, and daily productivity and quality are now more advanced. But in order to research, cross-breed, and test a rice variety, a new technique, or certain effects (aphids, pests, etc.) on the rice plants according to traditional methods [2], we have to sow seeds, wait for the rice to increase, and check the rice development status. This process takes much time, effort, and money.

Brown plant hoppers (BPHs) are usually very small, and mature ones are just as big as a grain of rice with a brown body. There are two main types: one with short wings and one with long wings. In temperature conditions ranging from 25 to 30 degrees Celsius, the life cycle of a BPH is around 25-30 days. It is divided into three phases: of egg, immature, and adult phases.

Indirect effects of BPH are transmission of bacterial diseases to rice, such as rice grass disease, twisted leaf dwarf diseases, and yellow dwarf diseases. A younger rice plant is more susceptible to infection, and later, it may not flower. When old rice is infected, the yield decreases slightly. Dwarfism does not currently have a cure. Therefore, work is being focused on the investigation of rice plants’ growth when they are infected with yellow dwarf disease. The disease can affect the height of plants. We have attempted to use simulation models to provide images of a sick rice plant to warn farmers.

Tree growth simulation approaches have been proposed in a vast number of studies. One study in [13] proposed a model of overseas movement to show the importance of large-scale systems’ changes and coordination in overseas migration. Another study [14] introduced different responses to abrupt and gradual CO$_{2}$ increases. They calculated the performance of BPH ($\textit{Nilaparvata lugens}$ (St˚al)) reared on rice for 15 successive generations under three CO$_{2}$ levels.

Another study [15] presented a method of sequencing the genomes of BPH populations and illustrated that the Indochinese peninsula is the major source of migration into temperate China. One study [18] introduced a calculation scheme for paddy field water temperature to compute water temperature dynamics in paddy fields. Their proposed method included three elements. The first one was a two-layer heat balance model, which was adopted to calculate water temperature. The solution to the heat balance of paddy water considers the effect of the vegetation layer.

Fig. 1. The 90-day growth stages of rice plants [16].

Fig. 2. Structure of growth units of a rice plant [16].

The second element included a plant growth status parameter calculated from thermal storage variations of paddy water to quantify the effect of vegetation layers on water temperature. The last one consisted of a series of correction methods for meteorological parameters to ensure that the parameters measured far from the paddy field are suitable for the model. Their method exhibited generality under different weather conditions but with the same rice cultivars.

Another study [19] proposed a modified SWAT, which was better than the original model in terms of simulated flow and nutrient load. It used variables such as vegetation coverage and different forest components. This work showed that the simulated flow increased by 7.8%, and simulations of the average, maximum, minimum, and total flow rates were closer to the measured values, thus better reflecting the actual surface water storage.

A different study [21] proposed a temperature response function for wheat plant height based on four cardinal temperatures. They constructed a water stress response function for wheat plant height using relative soil water availability (Aw) as a water stress index. Next, they implemented multipliers to modify the first-order derivative of six distinct growth functions to represent the elongation rate of wheat plant height under soil water stresses. Other work [22] proposed a method to evaluate the plant growth in an Asian country. Their proposed method also proved that it can improve plant growth based on factors such as temperature, air moisture, radiant energy, carbon dioxide levels, and soil pH.

Another study [23] introduced a 3D-simulation of plant growth using a Multi-Layered Perceptron (MLP) machine-learning method combined with a Polynomial-Lindenmayer (Poly-L) system. The model includes parameters such as the trunk/branch growth (L), the leaf width (W), and the number of branching (B) as a function of changes in nitrogen (N), phosphate (P), and potassium (K) in the fertilization process. They exhibited the results of the MLP method and saw that it was quite adaptable to the various changes of N, P, and K values. It was also able to estimate the L, W, and B output.

Farmers and agronomists categorize the influence on rice crops by dwarf diseases, thereby giving appropriate measures to handle the pest situation of BPH. From one study [8], we obtained the infected areas, the number of BPHs, and the corresponding age of rice plants. Through these data in combination with the aphid prevalence and the loss of height of rice plants provided by farmers, we adjusted the input values of the DIGIPLANT program to display rice plants when attacked by BPH.

In this work, we propose a simulation model to reveal the effect of BPH infestation on rice growth. We studied how to convert data between GAMA and DIGIPLANT to illustrate an image of rice plants when attacked by BPH in DIGIPLANT. Yellow dwarf disease currently has no specific efficient treatment. Therefore, the study has focused on investigating the growth of rice plants infected by yellow dwarf disease from the disease-carrying BPH.

The disease can affect rice growth, so we illustrated and simulated the height of the rice by generating an image of rice plants under BHP infestation. The generated images can provide warnings to farmers and agriculturists about the status of rice plants affected by yellow dwarf disease. Thus, they can deliver appropriate treatment measures for the pest situation of BPH.

The remaining sections of the study are as follows. Section 2 illustrates growth stages of a rice plant. We introduce the methods of simulation using GAMA (Section 3) and DIGIPLANT (Section 4). In Section 5.2, we present the approach to combine GAMA and DIGIPLANT to simulate a rice plant under BPH infestation. The experimental results are exhibited in Section 6. Section 7 presents some closing remarks.

## 2. The Growth of a Rice Plant

The rice plant belongs to the genus $\textit{Oryza}$. $\textit{Oryza}$ is the most popular rice for crops nowadays. Rice life begins from the moment a seed germinates until the rice is ripe. It can be divided into three phases: the growth phase (nutritional growth), reproduction phase, and maturation phase (Fig. 1).

The stem (body of a rice plant) consists of a series of nodes (Vietnamese people usually call them eyes'') and internodes, as illustrated in Fig. 2 [20]. The internodes are hollow stems between the eyes and are usually tightly held by the leaf sheath. Normally, the lower internodes are less developed, so the eyes are close together (only about 3-8 at the top) and start to elongate when the rice spikes (2-35 cm). In each eye, there is a leaf, a bud, and two secondary roots. A plant’s growth unit consists of an internode, an eye, two rings of roots, a leaf, and a bud. The first step corresponds to the vegetative stage of creating the internode and the leaves, and the process repeats to create other parts of the plant.

## 3. Spread Simulation of BPH Visualized in GAMA

GAMA [12] is defined as a modeling and simulation development environment to build spatially explicit agent-based simulations. The system can be applied to simulations for one or more locations or provinces. In this study, we evaluated our approach using Dong Thap Province, Vietnam. The main agents were defined in the simulation model, including the commune agent, rice plant agent, BPH egg agent, BPH agent, etc. Each agent has properties and behaviors to interact with the other agents in the environment.

The input data for a simulation includes some excel files: bphdata.csv, temperature.csv, and humidity.csv [8]. The file bphdata.csv contains data on the age of the rice plant and the number of BPHs in each area. The file temperature.csv includes data about the temperature. The file humidity.csv has data about the humidity. General parameters such as the starting date, month of simulation, number of simulation days, and daily aphid death rate can be modified when we perform the simulation.

The simulation result is displayed in Fig. 4 with a view that reveals the hopper density of each area and the growth stage of the rice plant. The levels of infection by BPH are shown in Fig. 4. Another view shows the BPH density statistical results by regions.

Fig. 3. Various levels of infection with BPH illustrated by various colors on a map.

Fig. 4. An illustration of simulation visualization result using GAMA. On the top, the visualization exhibits the status of rice season, while the bottom illustrates infected regions by BPH with colors in Fig. 3.

## 4. Simulation of Plant Structure and Growth using DIGILANT

### 4.1 Simulation of Plant Structure and Growth

Virtual tree planting is increasingly being used in many different fields. The main areas of interest are computer graphics and environmental science (agriculture and forestry). Sievanen et al. [17] divided models into two types for virtual tree planting: geometric models and functional structural models (functional structural models). A geometrical model is mainly based on a tree architecture built on an algorithm and simulated using computer graphics. It focuses mainly on the description of a tree architecture and principles of geometry. This is done by graphic design experts to create images of tree growth on a computer.

A functional structure model relies on mathematics to calculate the parts and related factors for the growth and development of virtual trees. Mathematical formulas have advantages when applied to calculating agronomic, forestry, and ecological factors because they can study system behavior and estimate the parameters for the optimized model and implementation of the system.

### 4.2 Rice Plant Simulation Software using DIGIPLANT

Paul Henry Cournede and his colleagues developed the framework named DIGIPLANT [11] to simulate the growth of rice plants. The framework also supports visualizing rice plants and presents a view of the results of the calculation process and the parameter metrics to create tree structures. We can modify the values of the parameters to change the results of the tree growth simulation. One of these parameters represents the lifetime of the period of the internodes that appears:

Fig. 5. The extracted data from GAMA including the number of BPH and rice plant age, name of region [1].

Fig. 6. The work flow for pre-processing data including BPH and rice plant age extracted by GAMA.

Fig. 7. Data conversion model between GAMA and DIGIPLANT.

- (Internode Expansion Time)

- (Cycle) 1 1

- (Cycle) 10 1

- (Cycle) 11 5

- (Cycle) 16 5

- (Cycle) 17 1

An internode appears in the 1st cycle and can live in 1 cycle. An internode appears in the 11th cycle and then lives for 5 cycles and continues for the next cycle of the internode. The calculate function allows a calculation for creating structures of the tree based on the input parameters and parameters in the function file (Fun-PA32 3D.txt). The results of the calculation process are saved in the file $\textit{biomass.data}$ in the output directory.

## 5. Combining GAMA and DIGIPLANT to Simulate Rice Plant Growth under BPH Infestation

### 5.1 Modeling to Render Data of BPH Transmission in Rice Fields to Assist Experts in Determining the Rate of Reduction of Plant Height.

The speed of the height of a plant decreasing can depend on several factors, including the number of BPHs, the time when rice plants are infected, and the age of the rice plant. Data were extracted during the implementation of a BPH propagation simulation in GAMA, as shown in Fig. 5. The text file contains information about the regions in Dong Thap Province, Vietnam, the number of BPHs, and the age of the rice plant. Fig. 6 exhibits the structure of the file. The number of lines corresponds to the simulated date. Data are stored in the form of the name of the region, density of BPHs, and the age of the rice plant.

### 5.2 Data Conversion Model from GAMA to DIGIPLANT

This section describes the data conversion model from GAMA to DIGIPLANT. A general diagram of the data conversion approach is exhibited in Fig. 7.

$\textbf{Transition input}$

Data on BPHs is a file extracted from the simulation process in GAMA. Based on the data, an agronomist decides to enter an appropriate rate of tree height loss.

### 5.3 Data Transformation from GAMA to DIGIPLANT

The process of creating data files containing a decrease in tree height in ascending order is presented in Fig. 7. In order to perform data transformation, the dataset must first be stored in a comparison table. The data are obtained by changing the internode lifetime values under many scenarios and combined with checking the speed of changes in tree height through the data file $\textit{biomass.data}$. Then, we sort the data in descending order of tree height and save the result in the comparison-file.txt file. Fig. 8 shows the tree-height decrease rate arranged in ascending order in the first column and the other the parameter corresponding to the decrease in tree height.

$\textbf{Data conversion between GAMA and DIGIPLANT}$

The conversion process was performed by comparing the imported tree-height reduction-ratio value from an agricultural specialist with the values in the comparison table. If the input value is less than or equal to the value in the table, we update the function file according to the parameters corresponding to the height in the comparison table.

Fig. 8. The process of creating data files containing a decrease in tree height in ascending order.

Fig. 9. The administrative map of Dong Thap Province, Vietnam.

$\textbf{Data output of data transformation}$

After the data conversion process, the function file has updated values. To display the image of the rice plant, the DIGIPLANT program is executed and references the function file. It performs operations that allow the program to perform calculations and display the image.

$\textbf{Converting input data of DIGIPLANT program according to the reduction ratio of tree height}$

We read data on BPHs in text files (with the *.txt extension) containing the BPH density, the age of the rice, and the area of pest infestation. These parameters are provided from the process of implementing the rice planthopper propagation simulation in GAMA under some scenarios. Each line of each file is read and compared. If the data belong to the considered areas entered by users, we fetch data on BPHs and the age of the rice plant. If not, we skip the line and move to the next line.

Next, we modify the values in the file (Fun-PA32 3D.txt) to display the cropped image as required. The input includes text function data, shoot-height reduction values, and comparison data tables, while the output consists of scaled-shift datasets. The procedure for processing comprises:

- Find a reduction rate value of rice plant buds from a specialist

- Open the comparison table file

- Open the function file

- Start the function that allows us to read each line of the comparison table

- Repeat the following until it reaches the end of the file:

· Perform a read from the beginning of the file to the end of the file

· Check if we are at the end of the file and then terminate the loop

· Check the rate of decrease in shoot height entered and height in the comparison table

· If the input value is less than or equal to the value in the table:

▪ Search for keyword "Internode Expansion Time"

▪ Update values

·Otherwise, read the next line

## 6. Simulation Results

### 6.1 Dataset

First, we present some information on the seasonality and hydrology of Dong Thap Province in Vietnam, where we have collected the data. Dong Thap includes 1 city and 11 districts and towns. It is a province in the Mekong Delta region and is usually considered as one of the largest rice granaries in Vietnam. Fig. 9 illustrates the administrative map of Dong Thap Province. It was processed using OPENJUMP software with the original data from the main map of Dong Thap Province. The communes belonging to the same district are shaded similarly. In a normal year, Dong Thap has two main wind directions: southwest monsoon (from May to November) and northeast monsoon (from December to April next year).

Data used in this model are BPH data provided by Dong Thap Plant Protection Department and include data on temperature, humidity, and average rainfall in months of the year from Dong Thap Statistical Yearbook [8]. The time of the simulation started on July 1, 2009, so we assume that the spreading of BPHs is in the southwest direction (like the wind direction). According to the traditional seasonal calendar, Dong Thap starts the summer and autumn crop. Input data for the simulation include:

- Number of simulation days: 90 days

- Wind direction: southwest

- Wind speed: 12 km/h

- Temperature: 25-28$^{\mathrm{o}}$C (read from the file temperature data.csv)

- Humidity: 84-86% (read from the file humidity data.csv).

- The average death rate of BPH is 3.5%

- The aphid density and rice age in some communes are initialized

### 6.2 Experiment on GAMA Model

The purpose of using this scenario is to find the parameter values for the area, the number of aphids, and the age of the rice. After we initialized the simulation, the status of infected BPH in the areas was illustrated. After simulating 67 steps, BPHs broke out in Tam Nong and Cao Lanh districts in the southwest wind direction. These areas were infected with a very high density (red, purple) and some mild infections (blue), as shown in Fig. 11.

Data on the age of a rice plant and the density of BPH are used to determine the prevalence of BPH and the effect on the reduction rate of shoot height of rice plants. A BPH can transmit rice viruses such as rice grass disease, twisted leaf dwarf diseases, and yellow dwarf diseases. The rate of BPH infected by yellow dwarf disease depends on some factors including the types of outbreaks of disease, year, season, region, etc. Therefore, depending on the time, agricultural experts can provide the corresponding prevalence of yellow dwarf disease. In addition, the speed of decrease in the height of a bud also depends on many factors, including the number of BPHs and the time when the rice plants were infected.

Table 1. BPH density in some communes in Dong Thap, Vietnam.

 District Commune name Ages (days) Adult BPH immature BPH Egg BPH Tan Hong Tan Thanh B 82 8.000 300 Tan Thanh B 5 700 300 100 Hong Ngu Thuong Thoi Hau B 10 200 200

Fig. 10. The initialization screen with the supplied input values.

Fig. 11. The results of the 67th simulation day.

Fig. 12. File storing tree height value when reduced by 1%.

Fig. 13. File storing tree height value showing a decrease in heights by 30%.

Fig. 14. The simulation result of a rice plant after decreasing by 1% (top) and 30% (bottom) in height.

### 6.3 Data Transformation from GAMA to DIGIPLANT

$\textbf{Scenario 1}$

In this situation, we investigated the growth status of plants in normal cases according to the default data values of the DIGIPLANT program. The conversion process is done mainly to change the parameters of the extension time of the internode expansion time. To find the comparison results before and after changing the lifetime value of the log, we performed a check of the height of the tree according to the initial default values of the program. After performing the calculation process to create the tree structure, the program creates a $\textit{biomass.data}$ file containing information from the calculation process, including the height of the tree, which has a value of 34.73 (cm).

$\textbf{Scenario 2}$

The plant height’s decrease value that is entered by experts is equal to the value in the comparison table. We assume that the expert’s height reduction value is 1%, so we change the lifetime value of the corresponding the internodes. The $\textit{biomass.data} file contains information about the calculation of the tree’s structure, and the height of the tree is 34.4 (cm).$\textbf{Scenario 3}$The height reduction value is entered by experts. The entered value is not equal to the values stored in the comparison table. In this case, we assume that the expert’s height reduction value is 30%. The$\textit{biomass.data} file (exhibited in Figs. 12 and 13) contains information about the calculation of the tree’s structure, which is exhibited in the right chart of Fig. 14 with a height of 20.63 (cm).

## 7. Conclusion

This work introduced an approach using simulation methods including GAMA and DIGIPLANT to understand how rice is affected by BPHs’ migration, destruction, and spread of disease. The study proposed an application of a combination of multi-agent modeling and simulation. The study provided an understanding of tree growth and how to model the structure model and the growth stages of plants. However, in this work, we only focused on researching BPHs that transmit yellow dwarf disease and cause dwarf rice plants. Further research should pay attention to aphids that can affect other parts of the various types of plants.

### REFERENCES

1
Hiep Huynh, Man Huynh, Hoa Tran, 2020, , A Data Conversion Approach between GAMA and DIGIPLANT Simulating Rice Plant Growth Under Brown Planthopper Infestation., 9th International Conference on Computational Data and Social Networks, Dec 11-13, 2020, Dallas, USA
2
Ministry of Agriculture and Rural Development , 2006, handbook on prevention and control of brown plant hopper that transmits yellow dwarf and twisted leaves dwarf diseases, National Agricultural Extension Center
3
Phan Huy Cuong , 2009, The rice BPH pest, Master thesis in Information systems, Can Tho University.
4
Nguyen Ngoc De , 2008, Curriculum on Rice text book, Department of Crop Resources, Mekong Delta Development Research Institute, Can Tho University.
5
Nguyen Cong Hien , Nguyen Pham Thuc Anh , 2006, System and simulation Systems text book, Science and Technology Publishing House, Hanoi.
6
Pham Van Kim , 2006, Some important diseases in rice in the Mekong Delta lecture, Ministry of Plant Protection, Can Tho University
7
Nguyen Ngoc Thach , Geographical information system, Vietnam National University of Science, Hanoi National University.
8
Vo Thanh Tung , 2010, Building a simulation model of aphid spread on rice fields, Master thesis of information systems
9
Drogoul Alexis , 2009, Lecture notes on Simulation, Can Tho University.
10
Paul Henry Cournede , 2004, DIGIPLANT, INRIA, Paris, France, 2004
11
Team DIGIPLANTe , 2006, Modeling plants growth and plants architecture, INRIA.
12
Platform GAMA.
13
Yang Shi-Jun, Bao Yun-Xuan, Chen Can, Lu Ming-Hong, Liu Wan-Cai, 2020, ,Analysis of atmospheric circulation situation and source areas for brown planthopper immigration to Korea: a case study., Ecosphere journal
14
Liu J., Zhuang J., Huang W., 2020, Different adaptability of the brown planthopper, Nilaparvata lugens (St˚al), to gradual and abrupt increases in atmospheric CO2., J Pest Sci 93, pp. 979-991
15
ProfileJ , 2020, Migration dynamics of an important rice pest: the brown planthopper (Nilaparvata lugens) across Asia - insights from population genomics., Evolutionary Applications
16
Ngoc De Nguyen , 2008, Rice textbook, Department of Plant Resources, Mekong Delta Development Research Institute, Can Tho University
17
Siev¨anen R., Godin C., DeJong T. M., 2014, Functional-structural plant models: a growing paradigm for plant studies., Annals of botany, Vol. 114, No. 4, pp. 599-603
18
Xie W., Kimura M., Iida T., , Simulation of water temperature in paddy fields by a heat balance model using plant growth status parameter with interpolated weather data from weather stations., 2020
19
Lai Geying, Luo Jingjing, Li Qiyue, Qiu Lin, Pan Ruixin, Zeng Xianggui, Zhang Lingling, June 2020, , Modification and validation of the SWAT model based on multi-plant growth mode, a case study of the Meijiang River Basin, China, Journal of Hydrology, Volume 585,ISSN 0022-1694
20
Morita S., Nemoto K., 1995, Morphology and anatomy of rice roots with special reference to coordination in organo- and histogenesis., In: Baluˇska F., Ciamporov´aˇ M., Gaˇspar´ıkov´a O., Barlow P.W. (eds) Structure and Function of Roots. Developments in Plant and Soil Sciences, Vol. vol 58. Springer Dordrecht
21
Tengcong Jiang, Jian Liu, Yujing Gao, Zhe Sun, Shang Chen, Ning Yao, Haijiao Ma, Hao Feng, Qiang Yu, 2020, Simulation of plant height of winter wheat under soil Water stress using modified growth functions, Agricultural Water Management, Vol. 232
22
Mani G., Chidambaranathan M., Sagi S., 2020, A framework for plant growth analysis and simulation using data analysis techniques, International Journal of Intelligent Unmanned Systems, Vol. 8, No. 4, pp. 281-292
23
Rokhana R., Herulambang W., Indraswari R., 2020, Machine Learning and Polynomial - L System Algorithm for Modeling and Simulation of Glycine Max (L) Merrill Growth, International Electronics Symposium (IES), Surabaya, Indonesia, 2020, pp. 463467

## Author

##### Hiep Xuan Huynh

Hiep Xuan Huynh received the Engineer degree from Can Tho University (CTU), the master’s degree from the l’Institut de la Franco phonie pour l’Informatique (IFI), the Ph.D. degree from the Polytechnics School of Nantes University (Polytech’Nantes), and the Habilitation \`{a} Diriger des Recherches (HDR) degree from the l’University de Bretagne Occidentale (UBO), all in computer science. He is currently an Associate Professor of computer science with the College of Information and Communication Technology, Can Tho University, Vietnam. His research interests include modeling of decisions with interestingness measure and fuzzy integral, recommender systems with statistical implicative analysis, a cyber-physical systems with cellular automata and wireless sensor networks, metrics for computer vision and population dynamics, agriculture, aquaculture, and environmental issues.

##### Man Ba Huynh

Man Ba Huynh was a master student from Can Tho University. He worked as an engineer and a researcher at Can Tho University Software Center, Can Tho, Vietnam.

##### Hai Thanh Nguyen

Hai Thanh Nguyen is a lecturer at the College of Information and Communi-cation Technology, Can Tho University, Vietnam. He received his Engineering degree in Informatics from Can Tho University, the master degree in Computer Science and Engineering from National Chiao Tung University, Taiwan, and obtained the PhD degree in Computer Science from Sorbonne University, France. His current research includes Bioinformatics, Health care systems, Computer Vision, modeling decisions, simulation and recommender systems.