Next Article in Journal
Joint Transcriptome and Metabolome-Based Analysis Reveals Key Modules and Candidate Genes for Drought Tolerance in Wheat (Triticum aestivum L.) Seedlings
Previous Article in Journal
A Multimodal Data-Driven Framework for Enhanced Wheat Carbon Flux Monitoring
Previous Article in Special Issue
Transcriptome Insights into Resistance Mechanisms Against Soybean Mosaic Virus Strain SC4 in Soybean
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Agronomy
Volume 15
Issue 4
10.3390/agronomy15040921
agronomy-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Monitoring of Soybean Bacterial Blight Disease Using Drone-Mounted Multispectral Imaging: A Case Study in Northeast China

by
Weishi Meng
1,
Xiaoshuang Li
1,
Jing Zhang
2,
Tianhao Pei
1 and
Jiahuan Zhang
1,3,*
1
College of Plant Protection, Jilin Agricultural University, Changchun 130118, China
2
Changbaishan Customs, People’s Republic of China, Antu 133613, China
3
Key Laboratory of Soybean Disease and Pest Control, Ministry of Agriculture and Rural Affairs, Changchun 130118, China
*
Author to whom correspondence should be addressed.
Agronomy 2025, 15(4), 921; https://doi.org/10.3390/agronomy15040921
Submission received: 4 March 2025 / Revised: 1 April 2025 / Accepted: 8 April 2025 / Published: 10 April 2025
(This article belongs to the Special Issue Recent Advances in Legume Crop Protection)
Browse Figures
Versions Notes

Abstract

:
Soybean bacterial blight disease is a threat to soybean production. Multispectral technology has shown good potential in detecting this disease and can overcome the limitations of traditional methods. The aim of this study was to perform field monitoring of the dynamics of this disease in Northeast China in 2022. The correlation between the soybean chlorophyll content index (CCI) and disease grade was obtained using artificial inoculation of the pathogen. The correlation between the soybean CCI, disease grade, green normalized difference vegetation index (GNDVI), and soybean yield was analyzed using a drone-mounted spectrometer platform for image acquisition and preprocessing. The soybean CCI was negatively correlated with the disease grade. The GNDVI declined with disease progression, which allowed for an indirect determination of the disease grade. The soybean yield loss was significant at disease grade 4 for soybean bacterial blight disease. The random forest regression model was more accurate than the regression model in estimating the yield based on the GNDVI. Therefore, the GNDVI could be used to survey the disease class and estimate the yield using the random forest model. This study provides support for field trials of drone-mounted multispectral equipment. This surveillance approach holds the potential to bring about precision plant protection in the future.

1. Introduction

Soybeans are a crucial crop globally, playing a significant role in food production and the agricultural economy. However, they are constantly threatened by various diseases [1], including soybean bacterial blight disease, which is caused by infestation of Pseudomonas savastanoi pv. glycinea [2,3,4]. This disease occurs in all soybean production areas in both northern and southern China, with more severe cases in the north. The disease used to occur mainly in the Huang-Hua-Hai region in China and was one of the major diseases of the soybean [5]. In recent years, due to climate warming, it has become more and more common in the main soybean-producing areas in Northeast China, and the degree of damage has gradually increased [6]. When severe, it can reduce soybean yields [7]. In Northeast China, the onset of the disease generally occurs during the soybean flowering stage up to the full grain stage (in July–August), especially under cool and humid climatic conditions [8,9]. The disease mainly affects leaves, but also petioles, stems, and bean pods (Figure 1). Leaf spots in the early stage are green, polygonal, and water-soaked, which then turn yellow to light brown and then reddish brown to blackish brown. The edge of the spot has a clear yellow-green halo and white bacterial pus overflows from the back of the spots. The spots often converge to form a large spot, causing some or all the leaves to turn yellow and die, resulting in early defoliation [10,11]. As soybean canopy leaves are the key part of the plant for photosynthesis, their growth, and death due to disease directly affect the final soybean quality and yield [12].
Conventional detection methods usually involve manual visual inspection; however, using these methods, the disease can only be observed after a period of crop infestation and therefore, they cannot be used for prevention [13]. To overcome these limitations and achieve early disease and pest detection, new technologies are needed. Unmanned aerial vehicles (UAVs) are aircraft that can be operated from miles away without the presence of a pilot [14]. UAVs can be equipped with various types of auxiliary sensors to obtain and quantify environmental parameters in the vicinity of the drone, which makes them useful in many fields [15]. In recent years, UAVs have been widely used in the field of low-altitude and low-speed remote sensing of farmland [16]. UAVs are used for remote sensing because of their high flexibility, low operating costs and ability to capture high-resolution images [17]. Moreover, multispectroscopy has been widely used in land-use planning, agricultural production, environmental protection, and other research fields because of its advantages such as a fast detection speed and the ability of on-site detection [18,19,20,21]. The potential of using UAVs and spectroscopic techniques to monitor soybean diseases and pests has been demonstrated in a few studies [22]. For example, near-infrared reflectance, detected from UAV-based multispectral imagery, decreased with increasing soybean aphid (Aphis glycines) populations in open-field trials when soybean aphid populations were above the economic threshold [23]. UAVs and cell phones were also used to capture images of caterpillars and Diabrotica speciosa (cucurbit beetle) [24]. In another study, UAV-acquired images were used to visualize the spatial and time series variation in an area damaged by red crown rot (a soil-borne disease of soybean) [25]. Tetila et al. (2017) proposed a computer vision system to track soybean foliar diseases in the field using images captured by an unmanned aerial vehicle (model DJI Phantom 3), which can aid experts and farmers in monitoring diseases in soybean fields [26]. Nagasubramanian et al. (2019) deployed a novel 3D deep convolutional neural network to identify charcoal rot disease in soybean stems [27]. Liu et al. (2023) discussed the use of hyperspectral analysis for classifying soybean diseases; they found that the PCA-SI combination method had a significantly better classification accuracy and could effectively distinguish between healthy and diseased soybean leaves [28]. Therefore, multispectroscopy image surveys of canopy leaves could be used to monitor soybean growth and disease for targeted field management.
The features of soybean diseases in fields are highly complex and multiple similar symptoms are usually mixed together [29,30]. The resolution of monitoring tools is also not sufficient for detecting the actual occurrence of diseases in the field. There is a need to continue developing and improving monitoring methods for soybean bacterial blight disease. Thus, the aim of this study was to explore a method for real-time monitoring of the occurrence of soybean bacterial blight disease using drone-based multispectral images and to provide a theoretical basis for the prediction and forecasting of large-scale occurrences of soybean bacterial spot disease for disease control in Northeast China.

2. Materials and Methods

The workflow of this study is presented in Figure 2, and includes three main parts: (1) UAV image and field dataset collection and preprocessing; (2) polynomial regression and random forest regression (RFR) analyses of the disease grades of soybean bacterial blight, the green normalized difference vegetation index (GNDVI), the chlorophyll content index (CCI) and soybean yield; and (3) screening methods for soybean yield estimation.

2.1. Overview of the Test Area

The test area is located in the Teaching and Research Experimental Base of Jilin Agricultural University (43°49′14″ N, 125°23′52″ E), Changchun City, Jilin Province, China, at an altitude of about 222.0 m (Figure 3). The geographic coordinate is GCS_WGS_ 1984 (EPSG: 4326). It is in the temperate continental semi-humid monsoon climate zone, with an average annual temperature of 4.6 °C, annual precipitation of 400–600 mm, and average annual sunshine time of about 2600 h. The terrain is flat, easily irrigated, and drained, and the soil is mainly black soil that is rich in organic substances and suitable for soybean growth. The experimental area was 946.4 m2, which was divided into 25 plots. Each plot consisted of 6 rows; each row was 20 m long and 0.65 m wide. The soybean variety ‘Jiyu47’ that was planted is susceptible to soybean bacterial blight disease [9,10].

2.2. Establishment of Sampling Points

The sampling points were located in two adjacent rows that were 0.65 m long and 1.5 m wide along the furrow, and contained a total area of about 2 m2 of plants; the soybean seedling densities were approximately the same between sampling points. Two adjacent sampling points were marked with a tag inserted into the soil between the two points (Table S1). In order to achieve real-time monitoring, manual periodic tracking was conducted and the information on disease grade was supplemented and updated in a timely manner based on disease markers in the field prior to each UAV flight.

2.3. Pathogen Culture and Inoculation

2.3.1. Preparation of Bacterial Suspensions

A sample of the bacterial strain that causes soybean bacterial blight disease was taken from the test area of this study and then isolated, purified, and cultured in the lab in NBY medium (liquid), which consists of 3 g of beef extract, 5 g of peptone, 2.5 g of sucrose (fructose), 2 g of yeast extract, 2 g of K2HPO4, 0.5 g of KH2PO4, 7.5 g of MgSO4, 0.5 g of KH2PO4, 1.5 g of MgSO4·7H2O, and 1000 mL of distilled water. A 20 μL sample of the bacterial suspension was transferred to 30 mL of NBY liquid medium using a pipette gun in an ultra-clean bench. The bacterial suspension (concentration: 1 × 108 CFU/mL) was shaken at 177 rpm for 24 h at 28 °C in a gas bath incubator. Then, 160 mL of the bacterial suspension was added to 4.5 L of sterile water and used immediately to inoculate in the field.

2.3.2. Inoculations of Pathogens in Field

The field inoculation was carried out twice on 1 August and 15 August 2022. The inoculation area was the 25 plots, which had fewer than three replicates of the different canopy leaf disease grades to ensure that the disease occurrence in the field showed all the disease grades. To ensure the effectiveness of the inoculation, we avoided the midday hours, when the sun is strong, and periods before rainfall. The prepared bacterial suspension was placed in a Drexel hand sprayer and sprayed on the soybean canopy leaves in a uniform manner so that each plant was evenly wetted. In order to achieve better inoculation efficiency, the bacterial suspension was sprayed onto the back of the soybean leaves from the bottom to the top to increase the contact area between the bacterial solution and the leaf surface. This method effectively reduces the risk of burns to the leaves caused by volatilization of the bacterial solution and the concentration of light, and it helps the bacteria invade the plant through the stomata on the back of the leaves during respiration, thus improving the inoculation effect.

2.4. Ground Data Acquisition

The ground data mainly included the CCI of the soybean leaves and the incidence of soybean bacterial blight disease [31]. The CCI values were measured 24–36 h prior to each UAV spectral acquisition using a CCM-200 handheld chlorophyll meter (Opti-Sciences, Hudson, NH, USA). At each sampling point, onset leaves were identified based on the presence of a large area of spots that did not contain cavities or other damage. Since only the metal probe part of the meter was used, the measurement of the affected leaves was performed at leaf spots, at spots without disease, and at the intersection between disease spots and health tissues. Several groups (10–20 samples) of samples were selected for averaging, and the equipment was calibrated after every 3–5 measurements to avoid measurement errors. The data obtained from the drone image acquisition were the soybean canopy leaf physical condition and disease level (Table S2).

2.5. UAV Image Acquisition and Preprocessing

2.5.1. UAV Multispectral Remote Sensing Platform

The UAV multispectral remote sensing platform mainly consisted of a UAV flight platform and an airborne imaging spectroscopy system [32]. The platform uses DJJ Wizard 4 (Shenzhen DJI Innovation Technology Co., Ltd., Shenzhen, China), a multispectral version (maximum takeoff weight: 1.482 kg; maximum single sortie duration: 27 min), and a multispectral camera (Figure 4). The multispectral parameters of Genie 4 have a spectral range of 460–950 nm, 6 sensors, 5 bands, 208 × 104 pixels, and a spectral imaging speed of 10−2–10−4 s (website: https://www.dji.com/cn/support/product/p4-multispectral, accessed on 7 April 2025).

2.5.2. Monitoring Method

Monitoring Time

In order to obtain high-quality multispectral images, the time period in which the observation data were collected was from 10:00 to 14:00 to avoid the influence of factors such as uneven light and unstable wind speeds. The remote sensing image acquisition by the UAV was performed under windy and sunny weather conditions, with wind speeds less than 5.4 m/s. The spectral camera was whiteboard corrected before the UAV took off, and the flight was completed within 40–50 min in order to avoid errors caused by equipment factors. The flight altitudes of the UAV were 30 m and 50 m to adapt to different flight requirements. After considering the above factors, the UAV remote sensing images were taken on 17 August (podding stage to the beginning of the grain stage), 20 August (beginning of the grain stage to the full grain stage), 27 August (the full grain stage to the first maturity stage), 1 September (the first maturity stage to the full grain stage) and 7 September (the full grain stage), 2022 [6,33]. The output of each photograph produced a high resolution TIFF image, which was used to extract spectral information and calculate the GNDVI by using ArcGIS 10.2.2, which was used in the next step of constructing a relational model of the impact on yield.

Image Spectral Bands

The parameters for each band in the UAV multispectral images are shown in Table 1. Red light is strongly absorbed by plant leaves, which reflect and transmit very low amounts of red light. Red edge (RE) is invisible light with a wavelength range of 700–780 nm and can indicate vegetation nutrition, growth, moisture, leaf area, etc. Near-infrared (NIR) is also invisible light, with a wavelength range of 780–1350 nm. When soybean vegetation is lush, with vigorous growth and high pigmentation, the red edge of the reflected light spectrum will shift towards near-infrared, and when the vegetation is affected by a variety of insect pests and grasses, environmental pollution, leaf aging, or other factors, the red edge will shift towards the red light [34].

Vegetation Index

The GNDVI (green normalized difference vegetation index) is a modification of the NDVI and is used to monitor the degree of denseness of the canopy at the maturity stage of soybean plants. It is calculated as the near-infrared band minus the green light band, divided by the near-infrared band plus the green light band [35]. Because soybean leaf photosynthetic pigments, especially chlorophyll, absorb red and blue light and reflect green light, the GNDVI replaces the red light band in the NDVI with the green light band, which is more suitable for disease development monitoring. The vegetation index formulas used in this study are shown in Table 2.

2.6. Soybean Yield Estimation

Before the soybeans reached maturity, from 18 to 25 September, the location of each sampling point and the occurrence of disease were recorded. All soybean samples within the delineated area were harvested, and the harvested samples were fully threshed and counted in the laboratory to calculate the soybean yield data for each sampling point. The soybean variety used (Jiyu 47) has a theoretical weight of 20 g per 100 grains. The yield was calculated using the following formula: Yield = (Total number of grains from sampling point/100) × 20. The yield per 2 m2 was calculated. The final yield per acre was then determined by scaling up the data using the conversion factor of 666 m2 = 1 acre. To ensure accuracy, three sets of replicate sampling point yield data were collected for all disease classes.

2.7. Statistical Analysis

The linear regression between CCI and disease grade, the calculation of the GNDVI range for each disease grade, and the construction of a regression model for GNDVI and CCI and for GNDVI and yield using the polynomial regression equation and random forest method were performed using Python 3.10.11 (Python software, Python Software Foundation, Wilmington, DE, USA). One-way ANOVA of the CCI at different disease grades and at each soybean vegetative stage and one-way ANOVA of soybean yield was performed using SPSS 26.0 (IBM Inc., Chicago, IL, USA).

3. Results

3.1. Correlation Analysis Between Soybean CCI and Disease Grades of Soybean Bacterial Blight

The soybean CCI was significantly and negatively correlated with the disease grade of soybean bacterial blight. From 17 August to 7 September, the soybean plants developed from the pod stage to the filling and maturity stages, and the leaves gradually changed from green to yellow. The CCI at the different observation dates and disease grades are shown in Table 3, and the results of the regression fitting are shown in Figure 5.
With an increase in disease grade, the CCI gradually decreased. This indicates that the higher the incidence, the more pronounced the decrease in the soybean CCI (Figure 6). The coefficients of determination (R2) of the regression models on different observation dates were all greater than 0.98, the F-values of the dates were high, and the p-values were much less than 0.05, indicating that the models had high significance (Table 4). There was a tendency for the CCI to decrease over time for the same disease grade. The coefficient of determination (R2) of the regression model for the different observation dates was greater than 0.98, the F-value for each date was high, and the p-value was much less than 0.05. This indicates that the model is highly significant (Table 4).

3.2. Correlation Analysis of Soybean CCI and GNDVI

The multispectral camera carried by the UAV was used to capture images of the soybean test field five times, and the GNDVI subplot of the multispectral images was calculated according to the disease grade in the plot and the selected sampling points (Figure 7). The GNDVI in all the sampling points showed a clear decreasing trend over time and with increasing disease severity (Table 5). When the CCI exceeded 30, the GNDVI no longer showed a linear growth trend and showed a flat trend (Figure 8), suggesting that soybeans tend to mature over time and that the CCI trend tends to flatten under disease stress. A regression model was established with GNDVI as the dependent variable and CCI as the independent variable (Table 6), and the fitted equation was y = −0.000250086570x2 + 0.0240266388x + 0.387753418, with a coefficient of determination R2 of 0.849, an F-value of 298.264, and a p-value (1.110 × 10−16) less than 0.05, indicating that the variables are correlated (Figure 6). The random forest regression models had a higher explanatory accuracy than the regression models (Figure 9). Therefore, the disease grade of soybean plants can be indirectly determined using the GNDVI, which can be used to monitor soybean bacterial blight disease.

3.3. Estimation of Soybean Yields

The one-way analysis of variance (ANOVA) between soybean bacterial blight disease grade and yield showed that the most significant yield loss of 8.6% was observed when the disease reached grade 4, which showed significant differences (Figure 10). Since the results of the above analyses concluded that there was a significant positive correlation between the GNDVI and disease grade, a model for estimating the GNDVI and soybean yield was constructed using polynomial regression (Figure 11) and random forest regression (Figure 12). The results showed that both models had the highest accuracy on 7 September (polynomial regression: R2 = 0.983; random forest regression: R2 = 0.978). On the other observation dates, the random forest model had a higher accuracy than the polynomial regression model. Thus, the random forest model can better explain the performance on 17 August, the models are comparable for 27 August, and the polynomial regression model was more suitable for the performance on 7 September (Table 7). Therefore, the GNDVI and random forest model could be used to estimate soybean yields. By optimizing the yield estimation model, we can obtain a more reliable method for soybean yield prediction based on multispectral technology, which can aid in decision-making in preventative disease control.

4. Discussion

In this study, UAV-borne multispectral imaging technology and ground-based data collection were combined to monitor soybean bacterial blight disease. This is a new method for monitoring soybean bacterial blight disease, and this study provides a more feasible solution for the application of this technology in actual soybean production. While previous studies have used spectral technology or ground surveys alone [27,28], by combining multispectral images of soybean canopy leaves using a UAV and ground surveys collecting data such as the CCI values of soybean leaves and the extent of disease incidence, we can obtain a comprehensive understanding of soybean growth and disease at both the macro- and microscopic levels. This multi-technology integration provides richer data support for accurate disease monitoring and analysis and enables more timely and precise disease detection. In selecting the research area and sampling points, the actual planting environment and disease occurrence pattern were fully considered, and the disease occurrence in the field, manipulated through artificial inoculation of pathogenic bacteria, reflected similar conditions to those observed in actual production.
The number of sampling points was insufficient to cover each disease level adequately and as a result, in-depth causal analysis and related vegetation index modeling could not be conducted. Further experiments are required in the future to explore these aspects. To reach the desired disease incidence levels (disease grades), especially during the less satisfactory mid stage of soybean spot development, leaves with no or a low disease incidence were intensively and manually inoculated with a specific dose of the bacterial solution. This was performed to increase the area of soybean canopy leaf spots. During the subsequent field recording and remote sensing monitoring, this treatment yielded favorable experimental results and achieved the expected goals. After a thorough and comprehensive analysis, it was found that this method has a certain degree of feasibility, and the experimental results provide a certain theoretical foundation and reference value. To ensure that the experimental process was close to that of real soybean production, the natural progression of disease development was utilized to allow the test plots to develop soybean bacterial blight disease.
UAV technology has been widely used in crop phenotypic research because of its flexibility and efficiency [32]. In this study, to initially establish the monitoring and early warning model, image data collected at a height of 30 m were used. However, the cost of acquiring UAV images at a height of 30 m is high, making it unsuitable for large-scale monitoring. For large areas of land, remote sensing images taken at higher altitudes can be used for broad-area exploration. Suspected disease areas can be segmented, and then more accurate image acquisition can be carried out at 30 m or other appropriate heights. Based on the results of this study, to obtain more accurate data for guiding agricultural production and developing disease prediction models, we need to continuously address the deficiencies in the experiment and conduct multiple repetitions for verification to obtain a more accurate and efficient prediction model.
The GNDVI is an improvement of the traditional NDVI, which replaces the red light band of the NDVI with a green light band for the absorption and reflection of light by photosynthetic pigments in soybean leaves and is therefore more effective in monitoring the density of the soybean canopy and the development of diseases at the maturity stage. This targeted application of the GNDVI provides new perspectives and methods for soybean disease monitoring [35,36]. The GNDVI was found to be significantly correlated with wheat yield; the GNDVI had better discriminating efficiency, allowing for better predictions of yield when recorded at early vegetative stages, and showed better results compared to the NDVI [37]. The GNDVI was also found to predict the aboveground biomass yield of maize better than the NDVI [38]. This was corroborated by the results of this study. After the disease outbreak, there were relatively few completely healthy, disease grade 0 sampling points. Thus, it was challenging to establish a large disease-free area and select a sufficient number of grade 0 sampling points for the control group. Due to the lack of field inoculation for soybean bacterial mottle disease and suboptimal field conditions for disease development, there were fewer samples with higher disease grades (7 to 9). This made it difficult to select a control for severely diseased samples. In contrast, the distribution of disease samples at grades 1-6 was more extensive, with grade 1 disease samples being the most prevalent. For the less severe grades 1 to 4, during image color rendering and considering the gradient differences among the different disease grades, choosing the green to brownish-red color band gradient was more effective. However, most of the images of disease grades 1 to 4 consisted of light green and yellow, making image recognition more difficult compared to the other disease grades. The images of disease grades 5 to 6 had a medium level of image recognition difficulty. Although a small number of sampling points were inaccurately recognized, they could be accurately identified after calibration. Future research can optimize the monitoring and synchronous calibration technology, which will significantly improve the monitoring and UAV accuracy. The final image was presented in a vertical perspective, facilitating the data processing by the researchers.

5. Conclusions

This study provides a basis for field trials using drone-mounted spectrometers to estimate soybean disease grades and yields using random forest regression and polynomial regression. The most significant yield loss of 8.6% was observed when the disease reached grade 4. The random forest regression model was found to be more accurate in explaining the yield on the different observation dates, especially on 7 September. The monitoring method applied in this study is effective in visualizing the damage due to soybean bacterial blight disease, but further improvements are required in the evaluation of intermediate damage and the generalization of the evaluation procedure.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/agronomy15040921/s1; Table S1: Field markers for disease locations; Table S2: Grading criteria for foliar conditions of soybean bacterial blight disease.

Author Contributions

Conceptualization, J.Z. (Jiahuan Zhang); methodology, J.Z. (Jiahuan Zhang); software, W.M. and X.L.; investigation, X.L., W.M. and J.Z. (Jing Zhang); data curation, T.P. and X.L.; writing—original draft preparation, X.L., W.M., J.Z. (Jing Zhang) and T.P.; writing—review and editing, W.M. and J.Z. (Jiahuan Zhang); funding acquisition, J.Z. (Jiahuan Zhang). All authors have read and agreed to the published version of the manuscript.

Funding

This research was supported by the Capital Construction Funds within the Provincial Budget in 2020 (Innovation Capacity Building) of Jilin Province, Jilin Provincial Development and Reform Commission (Grant No. 2020C019-5) and the Earmarked Fund for the China Agriculture Research System of MOF and MARA (Grant No. CARS04).

Data Availability Statement

The original contributions presented in the study are included in the article. Further inquiries can be directed to the corresponding author.

Acknowledgments

We thank Shaozhong Song from Jilin College of Engineering and Technology, Huanjun Liu and Yion Wang from the Northeast Institute of Geography and Agroecology, Chinese Academy of Sciences, for their guidance and assistance in performing the remote sensing. We thank the anonymous reviewers for reviewing this paper and providing constructive comments.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Bandara, A.Y.; Weerasooriya, D.K.; Bradley, C.A.; Allen, T.W.; Esker, P.D. Dissecting the economic impact of soybean diseases in the United States over two decades. PLoS ONE 2020, 15, e0231141. [Google Scholar] [CrossRef] [PubMed]
  2. Tarakanov, R.; Shagdarova, B.; Lyalina, T.; Zhuikova, Y.; Il’ina, A.; Dzhalilov, F.; Varlamov, V. Protective properties of copper-loaded chitosan nanoparticles against soybean pathogens Pseudomonas savastanoi pv. glycinea and Curtobacterium flaccumfaciens pv. flaccumfaciens. Polymers 2023, 15, 1100. [Google Scholar] [CrossRef]
  3. El-Esawi, M.A.; Ali, H.M.; Hatamleh, A.A.; Al-Dosary, M.A.; El-Ballat, E.M. Multi-functional PGPR Serratia liquefaciens confers enhanced resistance to lead stress and bacterial blight in soybean (Glycine max L.). Curr. Plant Biol. 2024, 40, 100403. [Google Scholar] [CrossRef]
  4. Feng, J. Recent advances in taxonomy of plant pathogenic bacteria. J. Integr. Agric. 2017, 50, 2305–2314, (In Chinese with English Abstract). [Google Scholar] [CrossRef]
  5. Wang, C.; Linderholm, H.W.; Song, Y.; Wang, F.; Liu, Y.; Tian, J.; Xu, J.; Song, Y.; Ren, G. Impacts of drought on maize and soybean production in Northeast China during the past five decades. Int. J. Environ. Res. Public Health 2020, 17, 2459. [Google Scholar] [CrossRef]
  6. Ye, W.W.; Liu, W.C.; Wang, Y.C. Occurrence status and whole-process green control technologies for soybean diseases and pests in China. J. Plant Prot. 2023, 50, 265–273, (In Chinese with English Abstract). [Google Scholar] [CrossRef]
  7. Surbhi, K.; Singh, K.; Aravind, T.; Bhatt, P.; Jeena, H.; Rakhonde, G. GIS-based survey and molecular detection of bacterial blight of soybean in Sub-Himalayan Ranges of Uttarakhand, India. Trop. Plant Pathol. 2023, 48, 332–346. [Google Scholar] [CrossRef]
  8. Zhao, F.Z.; Wang, Y.A.; Cheng, W.; Antwi-Boasiako, A.; Yan, W.K.; Zhang, C.T.; Gao, X.W.; Kong, J.J.; Liu, W.S.; Zhao, T.J. Genome-wide association study of bacterial blight resistance in soybean. Plant Dis. 2025, 109, 341–351. [Google Scholar] [CrossRef]
  9. Zhang, J.H.; Gao, J.; Yuan, M.L.; Li, Y. Studies on the distribution of physiological races of Pseudomonas syringae pv. glycinea. J. Jilin Agric. Univ. 2003, 25, 24–26, (In Chinese with English Abstract). [Google Scholar] [CrossRef]
  10. Zhang, J.H.; Gao, J.; Yuan, M.L. Identification of physiological races of Pseudomonas syringae pv. glycinea in Jilin province. J. Jilin Agric. Univ. 1993, 15, 24–27+105, (In Chinese with English Abstract). [Google Scholar] [CrossRef]
  11. Jagtap, P.G.; Dhopte, S.B.; Dey, U. Evaluation of different antibacterial antibiotics against bacterial blight of soybean caused by Pseudomonas syringae pv. glycinea under field conditions. Ind. Phytopathol. 2013, 66, 411–412. [Google Scholar]
  12. Wang, X.Y.; Li, Y.; Pan, T.; Zhang, Z.G.; Qu, J.Q.; Sun, Y.Z.; Zhao, H.J.; Miao, L.X.; Hu, Z.B.; Zhao, Z.M.; et al. Diagnosis of soybean bacterial blight progress stage based on deep learning in the context of data-deficient. Comput. Electron. Agric. 2023, 212, 108170. [Google Scholar] [CrossRef]
  13. Sotelo, J.P.; Rovey, M.F.P.; María, E.C.; Moliva, M.V.; Oliva, M.D.L.M. Characterization of Pseudomonas syringae strains associated with soybean bacterial blight and in vitro inhibitory effect of oregano and thyme essential oils. Physiol. Mol. Plant Pathol. 2023, 128, 102133. [Google Scholar] [CrossRef]
  14. Rahman, M.H.; Sejan, M.A.S.; Aziz, M.A.; Tabassum, R.; Baik, J.-I.; Song, H.-K. A comprehensive survey of unmanned aerial vehicles detection and classification using machine learning approach: Challenges, solutions, and future directions. Remote Sens. 2024, 16, 879. [Google Scholar] [CrossRef]
  15. Mukherjee, A.; Misra, S.; Raghuwanshi, N.S. A survey of unmanned aerial sensing solutions in precision agriculture. J. Netw. Comput. Appl. 2019, 148, 102461. [Google Scholar] [CrossRef]
  16. Huang, Y.B.; Thomson, S.J.; Brand, H.J.; Reddy, K.N. Development and evaluation of low-altitude remote sensing systems for crop production management. Int. J. Agric. Biol. Eng. 2016, 9, 1–11. [Google Scholar] [CrossRef]
  17. Yuan, J.; Zhang, Y.; Zheng, Z.; Yao, W.; Wang, W.; Guo, L. Grain crop yield prediction using machine learning based on UAV remote sensing: A systematic literature review. Drones 2024, 8, 559. [Google Scholar] [CrossRef]
  18. Zahra, A.; Qureshi, R.; Sajjad, M.; Sadak, F.; Nawaz, M.; Khan, H.A.; Uzair, M. Current advances in imaging spectroscopy and its state-of-the-art applications. Expert Syst. Appl. 2024, 238, 122172. [Google Scholar] [CrossRef]
  19. Hou, Y.; Bao, H.; Rimi, T.I.; Zhang, S.; Han, B.; Wang, Y.; Yu, Z.; Chen, J.; Gao, H.; Zhao, Z.; et al. Rice quality and yield prediction based on multi-source indicators at different periods. Plants 2025, 14, 424. [Google Scholar] [CrossRef]
  20. Yu, Y.; Li, C.; Shen, W.; Yan, L.; Zheng, X.; Yao, Z.; Cui, S.; Cui, C.; Hu, Y.; Yang, M. Correlation study between canopy temperature (CT) and wheat yield and quality based on infrared imaging camera. Plants 2025, 14, 411. [Google Scholar] [CrossRef]
  21. Sukhova, E.; Zolin, Y.; Popova, A.; Grebneva, K.; Yudina, L.; Sukhov, V. Broadband normalized difference reflectance indices and the normalized red–green index as a measure of drought in wheat and pea plants. Plants 2025, 14, 71. [Google Scholar] [CrossRef] [PubMed]
  22. Shrivastava, S.; Singh, K.S.; Hooda, S.D. Color sensing and image processing-based automatic soybean plant foliar disease severity detection and estimation. Multimed. Tools Appl. 2015, 74, 11467–11484. [Google Scholar] [CrossRef]
  23. Marston, Z.P.D.; Cira, T.M.; Hodgson, E.W.; Knight, J.F.; MacRae, I.V.; Koch, R.L. Detection of stress induced by soybean aphid (Hemiptera: Aphididae) using multispectral imagery from unmanned aerial vehicles. J. Econ. Entomol. 2020, 113, 779–786. [Google Scholar] [CrossRef]
  24. Mignoni, M.E.; Honorato, A.; Kunst, R.; Righi, R.; Massuquetti, A. Soybean images dataset for caterpillar and Diabrotica speciosa pest detection and classification. Data Brief 2022, 40, 107756. [Google Scholar] [CrossRef] [PubMed]
  25. Yamamoto, S.; Nomoto, S.; Hashimoto, N.; Maki, M.; Hongo, C.; Shiraiwa, T.; Homma, K. Monitoring spatial and time-series variations in red crown rot damage of soybean in farmer fields based on UAV remote sensing. Plant. Prod. Sci. 2023, 26, 36–47. [Google Scholar] [CrossRef]
  26. Tetila, E.C.; Machado, B.B.; de Souza Belete, N.A.; Guimaraes, D.A.; Pistori, H. Identification of soybean foliar diseases using unmanned aerial vehicle images. IEEE Geosci. Remote Sens. Lett. 2017, 14, 2190–2194. [Google Scholar] [CrossRef]
  27. Nagasubramanian, K.; Jones, S.; Singh, A.K.; Sarkar, S.; Singh, A.; Ganapathysubramanian, B. Plant disease identification using explainable 3D deep learning on hyperspectral images. Plant Methods 2019, 15, 98. [Google Scholar] [CrossRef]
  28. Liu, S.; Yu, H.-Y.; Sui, Y.-Y.; Kong, L.-J.; Yu, Z.-D.; Guo, J.-J.; Qiao, J. Hyperspectral data analysis for classification of soybean leaf diseases. Spectrosc. Spectr. Anal. 2023, 43, 1550–1555. [Google Scholar] [CrossRef]
  29. Afolabi, C.G.; Salihu, S.; Shokalu, O. Screening of soybean [Glycine max (L.) merrill] lines for reaction to natural field infection and resistant against bacteria foliar diseases. Arch. Phytopathol. Plant Protect. 2023, 56, 269–283. [Google Scholar] [CrossRef]
  30. Brown, M.T.; Mueller, D.S.; Kandel, Y.R.; Telenko, D.E.P. Influence of integrated management strategies on soybean sudden death syndrome (SDS) root infection, foliar symptoms, yield and net returns. Pathogens 2023, 12, 913. [Google Scholar] [CrossRef]
  31. Ravelombola, W.S.; Qin, J.; Shi, A.; Nice, L.; Bao, Y.; Lorenz, A.; Orf, J.H.; Young, N.D.; Chen, S. Genome-wide association study and genomic selection for soybean chlorophyll content associated with soybean cyst nematode tolerance. BMC Genom. 2019, 20, 904. [Google Scholar] [CrossRef] [PubMed]
  32. Yang, C.; Yang, G.; Wang, H.; Li, S.; Zhang, J.; Pan, D.; Ren, P.; Feng, H.; Li, H. Identifying key traits for screening high-yield soybean varieties by combining UAV-based and field phenotyping. Remote Sens. 2025, 17, 690. [Google Scholar] [CrossRef]
  33. Maimaitijiang, M.; Sagan, V.; Sidike, P.; Hartling, S.; Esposito, F.; Fritschi, F.B. Soybean yield prediction from UAV using multimodal data fusion and deep learning. Remote Sens. Environ. 2020, 237, 111599. [Google Scholar] [CrossRef]
  34. Huang, W.J.; Huang, M.Y.; Liu, L.Y.; Wang, J.H.; Zhao, C.J.; Wang, J.D. Inversion of the severity of winter wheat yellow rust using proper hyper spectral index. Trans. Chin. Soc. Agric. Eng. 2005, 21, 97–103. [Google Scholar]
  35. Shi, W.; Li, Y.; Zhang, W.; Yu, C.; Zhao, C.; Qiu, J. Monitoring and zoning soybean maturity using UAV remote sensing. Ind. Crops Prod. 2024, 222, 119470. [Google Scholar] [CrossRef]
  36. Jia, S.; Cui, M.; Chen, L.; Guo, S.; Zhang, H.; Bai, Z.; Li, Y.; Deng, L.; Li, F.; Zhang, W. Soybean water monitoring and water demand prediction in arid region based on UAV multispectral data. Agronomy 2025, 15, 88. [Google Scholar] [CrossRef]
  37. Kyratzis, A.C.; Skarlatos, D.P.; Menexes, G.C.; Vamvakousis, V.F.; Katsiotis, A. Assessment of vegetation indices derived by UAV imagery for durum wheat phenotyping under a water limited and heat stressed mediterranean environment. Front. Plant Sci. 2017, 8, 1114. [Google Scholar] [CrossRef]
  38. Tiruneh, G.A.; Meshesha, D.T.; Adgo, E.; Tsunekawa, A.; Haregeweyn, N.; Fenta, A.A.; Reichert, J.M. A leaf reflectance-based crop yield modeling in Northwest Ethiopia. PLoS ONE 2022, 17, e0269791. [Google Scholar] [CrossRef]
Agronomy 15 00921 g001
Figure 1. Soybean bacterial blight disease in a field. (A) Early stage. (B) Mid stage. (C) Late stage. Red circles indicate disease spots (Photographed by Jiahuan Zhang).
Figure 1. Soybean bacterial blight disease in a field. (A) Early stage. (B) Mid stage. (C) Late stage. Red circles indicate disease spots (Photographed by Jiahuan Zhang).
Agronomy 15 00921 g001
Agronomy 15 00921 g002
Figure 2. Workflow diagram for this study.
Figure 2. Workflow diagram for this study.
Agronomy 15 00921 g002
Agronomy 15 00921 g003
Figure 3. Study area diagram.
Figure 3. Study area diagram.
Agronomy 15 00921 g003
Agronomy 15 00921 g004
Figure 4. UAV equipment diagram. (A) Flying UAV in the air. (B) UAV in the lab. (Photographed by Xiaoshuang Li and Weishi Meng).
Figure 4. UAV equipment diagram. (A) Flying UAV in the air. (B) UAV in the lab. (Photographed by Xiaoshuang Li and Weishi Meng).
Agronomy 15 00921 g004
Agronomy 15 00921 g005
Figure 5. Soybean CCI at different disease grades. Different colors indicate different disease grades.
Figure 5. Soybean CCI at different disease grades. Different colors indicate different disease grades.
Agronomy 15 00921 g005
Agronomy 15 00921 g006
Figure 6. Regression fitting of soybean CCI and disease grade. (A) 17 August; (B) 20 August; (C) 27 August; (D) 1 September; (E) 1 September; (F) Regression fitting of five observations.
Figure 6. Regression fitting of soybean CCI and disease grade. (A) 17 August; (B) 20 August; (C) 27 August; (D) 1 September; (E) 1 September; (F) Regression fitting of five observations.
Agronomy 15 00921 g006
Agronomy 15 00921 g007
Figure 7. GNDVI at different disease grades.
Figure 7. GNDVI at different disease grades.
Agronomy 15 00921 g007
Agronomy 15 00921 g008
Figure 8. Polynomial regression analysis between GNDVI and soybean CCI on different observation dates. (A) 17 August; (B) 20 August; (C) 27 August; (D) 1 September; (E) 1 September; (F) Polynomial regression of five observations.
Figure 8. Polynomial regression analysis between GNDVI and soybean CCI on different observation dates. (A) 17 August; (B) 20 August; (C) 27 August; (D) 1 September; (E) 1 September; (F) Polynomial regression of five observations.
Agronomy 15 00921 g008
Agronomy 15 00921 g009
Figure 9. Random forest regression analysis between GNDVI and soybean CCI on different observation dates. (A) 17 August; (B) 20 August; (C) 27 August; (D) 1 September; (E) 1 September; (F) Random forest regression of five observations. Blue triangles represent the actual values.
Figure 9. Random forest regression analysis between GNDVI and soybean CCI on different observation dates. (A) 17 August; (B) 20 August; (C) 27 August; (D) 1 September; (E) 1 September; (F) Random forest regression of five observations. Blue triangles represent the actual values.
Agronomy 15 00921 g009
Agronomy 15 00921 g010
Figure 10. Soybean yields at different disease grades. Lowercase letters indicate differences among disease grades.
Figure 10. Soybean yields at different disease grades. Lowercase letters indicate differences among disease grades.
Agronomy 15 00921 g010
Agronomy 15 00921 g011
Figure 11. Polynomial regression analysis between GNDVI and soybean yield on different observation dates. (A) 17 August; (B) 20 August; (C) 27 August; (D) 1 September; (E) 1 September; (F) Polynomial regression of five observations.
Figure 11. Polynomial regression analysis between GNDVI and soybean yield on different observation dates. (A) 17 August; (B) 20 August; (C) 27 August; (D) 1 September; (E) 1 September; (F) Polynomial regression of five observations.
Agronomy 15 00921 g011
Agronomy 15 00921 g012
Figure 12. Random forest regression analysis between GNDVI and soybean yield on different observation dates. (A) 17 August; (B) 20 August; (C) 27 August; (D) 1 September; (E) 1 September; (F) Random forest regression of five observations. Blue triangles represent the actual values.
Figure 12. Random forest regression analysis between GNDVI and soybean yield on different observation dates. (A) 17 August; (B) 20 August; (C) 27 August; (D) 1 September; (E) 1 September; (F) Random forest regression of five observations. Blue triangles represent the actual values.
Agronomy 15 00921 g012
Table 1. Band parameters for the multispectral images.
Table 1. Band parameters for the multispectral images.
BandWavelength Range (nm)
Blue450 ± 16
Green560 ± 16
Red650 ± 16
Red Edge730 ± 16
Near-Infrared840 ± 26
Table 2. Vegetation index equations.
Table 2. Vegetation index equations.
Vegetation IndexEquation
Normalized Red Light (R)R/(R + G + B)
Normalized Green Light (G)G/(R + G + B)
Normalized Blue Light (B)B/(R + G + B)
Green Normalized Vegetation Index (GNDVI)(NIR − G)/(NIR + G)
Table 3. Soybean CCI at different disease grades on various observation dates.
Table 3. Soybean CCI at different disease grades on various observation dates.
Disease Grade17 August
(Podding Stage to Beginning of Grain Stage)		20 August
(Beginning of Grain Stage to Full Grain Stage)		27 August
(Full Grain Stage to First Maturity Stage)		1 September
(First Maturity Stage to Full Grain Stage)		7 September
(Full Grain Stage)
043.28 ± 0.72 a42.18 ± 0.82 a35.74 ± 2.06 a30.57 ± 1.04 a29.80 ± 1.10 a
139.04 ± 0.45 b37.84 ± 0.59 b31.12 ± 1.32 b28.57 ± 0.93 b25.55 ± 1.15 b
233.94 ± 0.53 c33.26 ± 0.29 c27.88 ± 1.26 c25.90 ± 0.36 c23.00 ± 1.40 c
330.04 ± 0.43 d30.02 ± 0.36 d24.50 ± 0.80 d23.97 ± 0.47 d20.65 ± 0.85 d
427.18 ± 0.24 e27.80 ± 0.23 e21.70 ± 0.74 e21.00 ± 0.80 e18.30 ± 0.60 e
523.80 ± 0.34 f23.74 ± 0.48 f19.28 ± 0.73 f18.90 ± 0.78 f15.20 ± 0.90 f
619.84 ± 0.18 g19.52 ± 0.32 g17.18 ± 0.88 g16.73 ± 0.58 g12.70 ± 0.60 g
715.76 ± 0.43 h15.90 ± 0.45 h14.70 ± 1.17 h13.77 ± 1.07 h10.55 ± 0.35 h
812.10 ± 0.45 i12.06 ± 0.43 i12.64 ± 1.15 i12.13 ± 0.72 i9.00 ± 0.10 i
97.74 ± 0.70 j7.94 ± 0.76 j10.36 ± 1.17 j8.77 ± 0.29 j8.10 ± 0.20 j
103.38 ± 0.68 k3.00 ± 0.42 k8.52 ± 1.43 k3.17 ± 0.35 k6.10 ± 0.80 k
Note: Data are shown as the mean ± S.E. Lower case letters indicate differences between disease grades on the same survey date.
Table 4. Regression fitting of soybean CCI and disease grade.
Table 4. Regression fitting of soybean CCI and disease grade.
Observation DateFitting ModelR2Fpmaemsermse
17 August
(podding stage to beginning of grain stage)		y = −3.75x + 45.710.9983801.1373.918 × 10−130.4600.3050.552
20 August
(beginning of grain stage to full grain stage)		y = −3.78x + 41.920.9972901.1791.318 × 10−120.5610.4430.665
27 August
(full grain stage to first maturity stage)		y = −2.68x + 35.260.988716.6056.854 × 10−100.7410.8520.923
1 September
(first maturity stage to full grain stage)		y = −2.87x + 37.510.982484.3793.898 × 10−90.6910.9160.957
7 September
(full grain stage)		y = −2.57x + 31.650.980435.1876.258 × 10−90.8191.0151.000
Note: mae—mean absolute error; mse—mean square error; rmse—root mean square error. These abbreviations are also used below.
Table 5. GNDVI at different disease grades on different observation dates.
Table 5. GNDVI at different disease grades on different observation dates.
Disease GradeGNDVIGNDVI_SdGNDVI_Range
17 August
(Podding Stage to Beginning of Grain Stage)		20 August
(Beginning of Grain Stage to Full Grain Stage)		27 August
(Full Grain Stage to First Maturity Stage)		1 September
(First Maturity Stage to Full Grain Stage)		7 September
(Full Grain Stage)		17th August to 7th September
(Podding Stage to Full Grain Stage)
00.92870.96210.94080.94200.91600.938 ± 0.0170.9621~0.9160
10.90670.95250.93050.92430.88430.920 ± 0.0260.9525~0.8843
20.82220.85960.84710.78740.79160.822 ± 0.0320.8596~0.7874
30.77860.84180.81280.75780.75230.789 ± 0.0380.8418~0.7523
40.73110.82560.79040.74800.71630.762 ± 0.0450.8256~0.7163
50.63900.79530.77640.71460.67260.720 ± 0.0660.7953~0.6390
60.55600.75300.74650.67880.63820.675 ± 0.0820.7465~0.5560
70.50570.67540.68060.59650.57210.606 ± 0.0740.6806~0.5057
80.46890.65350.65860.55920.47170.562 ± 0.0930.6586~0.4689
90.40410.59610.59030.52990.42780.510 ± 0.0900.5961~0.4041
100.36210.52700.53350.45690.34590.445 ± 0.0890.5335~0.3459
Table 6. Polynomial regression and random forest regression between GNDVI and soybean CCI on different observation dates.
Table 6. Polynomial regression and random forest regression between GNDVI and soybean CCI on different observation dates.
Observation DatePolynomial RegressionRandom Forest Regression
Fitting ModelR2FpmaemsermseR2
17 August
(podding stage to beginning of grain stage)		y = −0.00011x2 + 0.02x + 0.3130.981465.967<0.0010.0110.0010.0130.995
20 August
(beginning of grain stage to full grain stage)		y = −0.00007x2 + 0.013x + 0.5530.985587.210<0.0010.0090.0010.0110.991
27 August
(full grain stage to first maturity stage)		y = −0.00033x2 + 0.027x + 0.4030.987672.916<0.0010.0090.0010.0120.989
1 September
(first maturity stage to full grain stage)		y = 0.016x + 0.4390.965249.450<0.0010.0120.0010.0150.988
7 September
(full grain stage)		y = −0.00100x2 + 0.044x + 0.2050.972316.937<0.0010.0150.0010.0160.989
17 August to 7 September (podding stage to full grain stage)y = −0.00025x2 + 0.02403x + 0.387750.849298.264<0.0010.0240.0010.0310.957
Table 7. Polynomial regression and random forest regression between GNDVI and soybean yield on different observation dates.
Table 7. Polynomial regression and random forest regression between GNDVI and soybean yield on different observation dates.
Observation DatePolynomial RegressionRandom Forest Regression
Fitting ModelR2FpmaemsermseR2
17 August
(podding stage to beginning of grain stage)		y = 145.981x2 − 112.79x + 77.0390.953182.185<0.0011.6816.1002.4700.976
20 August
(beginning of grain stage to full grain stage)		y = 315.913x2 − 366.992x + 160.3480.961223.983<0.0011.8686.7992.6070.974
27 August
(full grain stage to first maturity stage)		y = 303.712x2 − 330.855x + 142.1210.967260.446<0.0011.8596.6662.5820.974
1 September
(first maturity stage to full grain stage)		y = 146.966x2 − 100.863x + 65.5660.964242.715<0.0012.0196.9622.6380.973
7 September
(full grain stage)		y = 195.218x2 − 164.304x + 88.5130.983523.781<0.0011.6025.7082.3890.978
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Meng, W.; Li, X.; Zhang, J.; Pei, T.; Zhang, J. Monitoring of Soybean Bacterial Blight Disease Using Drone-Mounted Multispectral Imaging: A Case Study in Northeast China. Agronomy 2025, 15, 921. https://doi.org/10.3390/agronomy15040921

AMA Style

Meng W, Li X, Zhang J, Pei T, Zhang J. Monitoring of Soybean Bacterial Blight Disease Using Drone-Mounted Multispectral Imaging: A Case Study in Northeast China. Agronomy. 2025; 15(4):921. https://doi.org/10.3390/agronomy15040921

Chicago/Turabian Style

Meng, Weishi, Xiaoshuang Li, Jing Zhang, Tianhao Pei, and Jiahuan Zhang. 2025. "Monitoring of Soybean Bacterial Blight Disease Using Drone-Mounted Multispectral Imaging: A Case Study in Northeast China" Agronomy 15, no. 4: 921. https://doi.org/10.3390/agronomy15040921

APA Style

Meng, W., Li, X., Zhang, J., Pei, T., & Zhang, J. (2025). Monitoring of Soybean Bacterial Blight Disease Using Drone-Mounted Multispectral Imaging: A Case Study in Northeast China. Agronomy, 15(4), 921. https://doi.org/10.3390/agronomy15040921

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop