Smart Agriculture

Recent Advances and Future Outlook for Artificial Intelligence in Aquaculture | Open Access

LI Daoliang, LIU Chang

2020, 2(3): 1-20. doi:10.12133/j.smartag.2020.2.3.202004-SA007

Asbtract ( 5474 )

HTML ( 7183)

PDF (2843KB) ( 6655 )

References | Related Articles | Metrics

The production of China's aquaculture has changed from extensive model to intensive model, the production structure is continuously adjusting and upgrading, and the production level has been continuously improved. However, as an important part of China's agricultural production, aquaculture plays an important role in promoting the development of China's agricultural economy. Low labor productivity, production efficiency and resource utilization, low-quality aquatic products, and the lack of safety guarantees have severely limited the rapid development of China's aquaculture industry. Using modern information technology and intelligent devices to realize precise, automated, and intelligent aquaculture, improving fishery productivity and resource utilization is the main way to solve the above contradictions. Artificial intelligence technology in aquaculture is to use the computer technology to realize the production process of aquaculture, monitor the growth of underwater organisms, judge, discuss and analyze problems, and then perform feeding, disease treatment, and breeding. In order to understand the development status and technical characteristics of artificial intelligence technology in aquaculture, in this article, five main aspects of aquaculture, i.e., life information acquisition, aquatic product growth regulation and decision-making, fish disease prediction and diagnosis, aquaculture environment perception and regulation, and aquaculture underwater robots, combined with the practical problems in aquaculture, were mainly focused on. The application principles and necessity of artificial intelligence technology in each aspect were explained. Commonly used technical methods were point out and the classic application cases were deeply analyzed. The main problems, bottlenecks and challenges in the current development of artificial intelligence technology in aquaculture were analyzed, including turbid water, multiple interference factors, corrosion of equipment, and movement of underwater animals, etc., and reasonable research directions for these potential challenges were pointed out. In addition, the main strategic strategies to promote the transformation of aquaculture were also proposed. The development of aquaculture is inseparable from artificial intelligence technology, this review can provide references to accelerate the advancement of digitalization, precision and intelligent aquaculture.

Current State and Challenges of Automatic Lameness Detection in Dairy Cattle | Open Access

HAN Shuqing, ZHANG Jing, CHENG Guodong, PENG Yingqi, ZHANG Jianhua, WU Jianzhai

2020, 2(3): 21-36. doi:10.12133/j.smartag.2020.2.3.202006-SA003

Asbtract ( 1730 )

HTML ( 725)

PDF (1720KB) ( 910 )

References | Related Articles | Metrics

Lameness in dairy cattle could cause significant economic losses to the dairy industry. Detection of lameness in a timely manner is critical to the high-quality development of dairy industry. The traditional method is visual locomotion scoring by dairy farmers, which is low efficiency, high cost and subjective. The demand for automated lameness detection is increasing. The review was conducted to find out the current state and challenges of automatic lameness detection technology development and to learn from the latest findings. The current automatic lameness detection systems were reviewed in this paper mainly rely on five technologies or combinations thereof, including machine vision, pressure distribution measuring system, wearable sensor system, behavior analysis and classification; the principle, function and features of these technologies were analyzed. Machine vision technique is to extract feature variables (e.g. back arch, head bob, abduction, stride length, walking speed, temperature, etc.) from video recordings of cattle movement by image processing. Pressure distribution measuring system contains an array of load cells to sense gait variables, when dairy cattle are walking by. By using accelerometer with high frequency data collection, the gait cycle parameters can be extracted and used for lameness detection. By using wearable devices, the number of lying/standing bouts and their duration, the total time spent lying, standing and ruminating per day can be recorded for individual cattle. The lameness can also be detected by behavior analysis. Currently, most of these studies were in the stage of sensor development or validation of algorithm. A few studies were in the stage of validation of performance and decision support with early warning system. The challenges to apply automatic lameness detection system in dairy farm includes the difficulties of acquiring high quality data of lameness features, lack of techniques to detect early lameness, identification errors caused by individual gait differences among dairy cattle, difficulties to function well in unstructured environment and difficulties to evaluate the benefits. To accelerate the development of automatic lameness detection systems, recommendations are proposed as follows: ①promoting lameness data sharing and data exchange among dairy farms; ②developing individual-based lameness classification model; ③developing multifunctional smart station which can detect lameness, measure body condition score, weighing, etc; ④evaluating the significance of automatic lameness detection to the dairy industry from the perspective of animal welfare, environment and food safety.

Construction Method and Performance Test of Prediction Model for Laying Hen Breeding Environmental Quality Evaluation | Open Access

LI Hualong, LI Miao, ZHAN Kai, LIU Xianwang, YANG Xuanjiang, HU Zelin, GUO Panpan

2020, 2(3): 37-47. doi:10.12133/j.smartag.2020.2.3.202003-SA010

Asbtract ( 960 )

HTML ( 502)

PDF (2140KB) ( 637 )

References | Related Articles | Metrics

Environmental quality of facilities affects the healthy growth and production of laying hens. The breeding environment of laying hens is a complex and non-linear system in which multiple environmental factors interact and restrict each other. It is difficult to make an accurate and effective evaluation on the suitability of laying hens with a single breeding environment parameter. In order to solve the above problem, an improved cuckoo search algorithm optimization neural network (CS-BP) model for the evaluation and prediction of the environmental suitability of laying hen facility was proposed in this research. In this model, the effects of environmental factors such as temperature, humidity, light intensity and ammonia concentration were comprehensively analyzed, and the problem of high prediction accuracy caused by BP neural network easily falling into local minimum value was overcome. In the experiment, the model was compared with BP neural network, genetic algorithm optimized BP neural network (GA-BP) and particle swarm optimization BP neural network (PSO-BP). The results showed that the mean absolute error (MAE), mean relative error (MAPE) and the coefficient of determination (R²) of the prediction model based on the improved CS-BP were 0.0865, 0.0159 and 0.8569, respectively. The prediction model based on the improved CS-BP had a strong generalization ability and a high testing precision, and its index performance was better than the above three comparison models. The classification accuracy of the improved CS-BP model was tested, and the result was 0.9333. The model constructed in this research can provide more comprehensive and effective scientific evaluation for the environmental quality of laying hens facility, which is of great significance to realize the optimal control of the production environment and promote the production performance of layers.

Design and Application of Facility Greenhouse Image Collecting and Environmental Data Monitoring Robot System | Open Access

GUO Wei, WU Huarui, ZHU Huaji

2020, 2(3): 48-60. doi:10.12133/j.smartag.2020.2.3.202007-SA006

Asbtract ( 1202 )

HTML ( 1984)

PDF (2668KB) ( 1048 )

References | Related Articles | Metrics

China's facility horticulture has developed rapidly in the past 30 years and now comes to the first in the world in terms of area. However, the number of farmers is decreasing. "Machine replaces labor" has become the current research hotspot. In order to realize the fine collection of crop images and environmental monitoring data, a three-dimensional environmental robot monitoring system for crops was designed. The robot consists of three parts: perception center, decision center and execution center, which carry out environmental perception from machine perspective, data analysis, decision instruction generation and action execution respectively. In perception layer, the system realized real-time videos, images, data monitoring such as air temperature, air humidity, illumination intensity and concentrations of carbon dioxide in grid scale from multi-angle with high accuracy. At the system level, automatic speech recognition was integrated to make the system easier to use, especially for farmers who usually work in the fields. In transport layer, monitoring data and control instructions were converged to local data center through wireless bridges. Concretely, transmission mode was chosen according to different characteristics of data, wire transmission is available for big size data, such as images and videos, while wireless transmission is mainly applied to small size data, such as environmental monitoring parameters. In data processing layer, feedbacks and control instructions were made by multi-source heterogeneous data of crop model analysis, in terms of commands, independent inspection mode and real-time remote-control mode were available for users. Plant type, user information, historical data and management data were taken into consideration. Finally, in application layer, the system provided web and mobile intelligence services that could be used for the whole growth periods in terms of images, real-time videos, monitoring data collection and analysis of cucumbers, tomatoes, greenhouse peaches, etc. The system has been demonstrated and applied in solar greenhouse No. 7 of Beijing Xiaotangshan National Precision Agriculture Base and No. 5 of Shijiazhuang Agricultural and Forestry Science Research Institute with good achievements. Farmers and researchers have realized real-time monitoring, remote control and management. On one hand, the system can used to avoid working in extreme environment, such as high temperature and pesticide environment. On the other hand, with the help of the robot, independent inspection and data collection could achieve instead of people, and it is very intuitive in time-saving and indirect costs saving for productions and researchers. The results showed that the system could be widely applied in greenhouse facilities production and research.

Distinguishing Volunteer Corn from Soybean at Seedling Stage Using Images and Machine Learning | Open Access

FLORES Paulo, ZHANG Zhao, MATHEW Jithin, JAHAN Nusrat, STENGER John

2020, 2(3): 61-74. doi:10.12133/j.smartag.2020.2.3.202007-SA002

Asbtract ( 1819 )

HTML ( 2685)

PDF (1967KB) ( 1596 )

References | Related Articles | Metrics

Volunteer corn in soybean fields are harmful as they disrupt the benefits of corn-soybean rotation. Volunteer corn does not only reduce soybean yield by competing for water, nutrition and sunlight, but also interferes with pest control (e.g., corn rootworm). It is therefore critical to monitor the volunteer corn in soybean at the crop seedling stage for better management. The current visual monitoring method is subjective and inefficient. Technology progress in sensing and automation provides a potential solution towards the automatic detection of volunteer corn from soybean. In this study, corn and soybean were planted in pots in greenhouse to mimic field conditions. Color images were collected by using a low-cost Intel RealSense camera for five successive days after the germination. Individual crops from images were manually cropped and subjected to image segmentation based on color threshold coupled with noise removal to create a dataset. Shape (i.e., area, aspect ratio, rectangularity, circularity, and eccentricity), color (i.e., R, G, B, H, S, V, L, a, b, Y, Cb, and Cr) and texture (coarseness, contrast, linelikeness, and directionality) features of individual crops were extracted. Individual feature's weights were ranked with the top 12 relevant features selected for this study. The 12 features were fed into three feature-based machine learning algorithms: support vector machine (SVM), neural network (NN) and random forest (RF) for model training. Prediction precision values on the test dataset for SVM, NN and RF were 85.3%, 81.6%, and 82.0%, respectively. The dataset (without feature extraction) was fed into two deep learning algorithms—GoogLeNet and VGG-16, resulting into 96.0% and 96.2% accuracies, respectively. The more satisfactory models from feature-based machine learning and deep learning were compared. VGG-16 was recommended for the purpose of distinguishing volunteer corn from soybean due to its higher detection accuracy, as well as smaller standard deviation (STD). This research demonstrated RGB images, coupled with VGG-16 algorithm could be used as a novel, reliable (accuracy >96%), and simple tool to detect volunteer corn from soybean. The research outcome helps provide critical information for farmers, agronomists, and plant scientists in monitoring volunteer corn infestation conditions in soybean for better decision making and management.

Identification and Morphological Analysis of Adult Spodoptera Frugiperda and Its Close Related Species Using Deep Learning | Open Access

WEI Jing, WANG Yuting, YUAN Huizhu, ZHANG Menglei, WANG Zhenying

2020, 2(3): 75-85. doi:10.12133/j.smartag.2020.2.3.202008-SA001

Asbtract ( 1775 )

HTML ( 851)

PDF (1962KB) ( 1569 )

References | Related Articles | Metrics

Invasive pest fall armyworm (FAW) Spodoptera frugiperda is one of the serious threats to the food safety. Early warning and control plays a key role in FAW management. Nowadays, deep learning technology has been applied to recognize the image of FAW. However, there is a serious lack of training dataset in the current researches, which may mislead the model to learn features unrelated to the key visual characteristics (ring pattern, reniform pattern, etc.) of FAW adults and its close related species. Therefore, this research established a database of 10,177 images belonging to 7 species of noctuid adults, including FAW and 6 FAW close related species. Based on the small-scale dataset, transfer learning was used to build the recognition model of FAW adults by employing three deep learning models (VGG-16, ResNet-50 and DenseNet-121) pretrained on ImageNet. All of the models got more than 98% recognition accuracy on the same testing dataset. Moreover, by using feature visualization techniques, this research visualized the features learned by deep learning models and compared them to the related key visual characteristics recognized by human experts. The results showed that there was a high consistency between the two counterparts, i.e., the average feature recognition rate of ResNet-50 and DenseNet-121 was around 85%, which further demonstrated that it was possible to use the deep learning technology for the real-time monitoring of FAW adults. In addition, this study also found that the learning abilities of key visual characteristics among different models were different even though they have similar recognition accuracy. Herein, we suggest that when evaluating the model capacity, we should not only focus on the recognition rate, the ability of learning individual visual characteristics should be allocated importance for evaluating the model performance. For those important taxonomical traits, if the visualization results indicated that the model didn't learnt them, we should then modify our datasets or adjusting the training strategies to increase the learning ability. In conclusion, this study verified that visualizing the features learnt by the model is a good way to evaluate the learning ability of deep learning models, and to provide a possible way for other researchers in the field who want to understand the features learnt by deep learning models.

Rapid Recognition Model of Tomato Leaf Diseases based on Kernel Mutual Subspace Method | Open Access

ZHANG Yan, LI Qingxue, WU Huarui

2020, 2(3): 86-97. doi:10.12133/j.smartag.2020.2.3.202009-SA001

Asbtract ( 828 )

HTML ( 532)

PDF (1950KB) ( 653 )

References | Related Articles | Metrics

Research on tomato disease recognition based on leaf images has been widely concerned in recent years, and with the development of machine learning and deep learning, researchers from various countries have proposed a variety of methods and models to solve this problem. In this research, a new approach by fusion color and texture features, and kernal mutual subspace method (KMSM) were introduced and a rapid recognition model of tomato leaf disease was established. The color and texture features introduced in this research including color moment (CM), color coherence vector (CCV) and histogram of oriented gradient (HOG) features. The CCHKMSM (CM+CCV+HOG+KMSM) model firstly mapped the extracted color and texture features from different classes of leaf disease data sets to high-dimensional space using gauss kernal function. Then the principal component of the mapped high-dimensional space was analysed, and the nonlinear disease characteristic space was generated. Finally, the diseases based on the minimum cosine angle of nonlinear feature space were identified. Validation experiment was conducted based on public agricultural disease data sets of PlantVillage, which providing 9 kinds of most commonly tomato leaf disease and 1 kind of healthy leaf image, and the filed took image include 3 kinds of tomato leaf diseases images. For experiment based on PlantVillage data set, the results showed that the CCHKMSM realized the most high recognize accuracy rate of 100% when the number of each class was 350. The training time cost and recognition time cost was 0.1540 s and 0.013 s, respectively. Meanwhile, experiments were conducted in the range of sample image numbers from 150 to 1000 images for each class, with step length of 50, and the obtained results showed that the average recognition rate was 99.14%. For experiment based on field took tomato diseases data sets, after segment original image into sub-size image, average recognize accuracy for the kind of diseases arrived at 96.21%, which was higher than other typical machine learning models such as SVM and KMSM, and at the same level by comparing with deep learning-based recognition methods. On the other hand, as an significant adventure of the proposed CCHKMSM model, the computing cost was low, both the training time and testing time were much lower than deep learning methods, and requirement is loss the system to run. As a conclusion, the proposed CCHKMSM model, has high potential to be applied in low-configuration equipment such as hand-held devices and edge computing terminals.

Hybrid Multi-Hop Routing Algorithm for Farmland IoT based on Particle Swarm and Simulated Annealing Collaborative Optimization Method | Open Access

SUN Haoran, SUN Lin, BI Chunguang, YU Helong

2020, 2(3): 98-107. doi:10.12133/j.smartag.2020.2.3.202007-SA001

Asbtract ( 623 )

HTML ( 172)

PDF (1518KB) ( 471 )

References | Related Articles | Metrics

Agricultural wireless sensor networks plays a key role in obtaining multi-source heterogeneous big data of farmland soil, environment and crop growth. The increasing network scale brings challenges to the application of agricultural Internet of Things. In order to solve the problem that sensors are not uniformly distributed in farmland and constrained by energy, a collaborative optimization hybrid multi hop routing algorithm, particle simulated multipath routing (PSMR) based on particle swarm optimization and simulated annealing was proposed. Firstly, cluster heads were selected by node residual energy and node degree weighting, and cluster structure was used to realize efficient dynamic networking of heterogeneous networks. Then, the multi hop data structure between cluster heads was used to solve the problem of high energy consumption in long-distance transmission of cluster heads. Particle swarm optimization and simulated annealing were used to improve the convergence speed, and sink nodes could accelerate the collection of aggregated data in cluster heads. The simulation results showed that compared with the energy-efficient load balancing multipath routing scheme (EMR), the network lifetime of PSMR algorithm was increased by 57%. EMR selected the data transmission link with low energy consumption and small delay by calculating the weight of total link hops and transmission energy consumption. Compared with greedy perimeter stateless routing-algorithm (GPSR-A algorithm), which could ensure the shortest data transmission distance and lower network transmission delay, the first dead sensor node was delayed for two rounds in the same network life cycle, and the residual energy standard deviation was reduced by 0.04 J, which had good network energy consumption balance. PSMR algorithm could reduce the extra energy consumption of remote cluster heads by multi hop between cluster heads, and improved the energy balance performance of cluster heads with different distances. It can provide technical basis for long-term, efficient and stable data acquisition and monitoring of large-scale farmland complex environment, and improve the resource utilization efficiency of agricultural Internet of Things.

Short-Term Price Forecast of Vegetables Based on Combination Model of Lasso Regression Method and BP Neural Network | Open Access

YU Weige, WU Huarui, PENG Cheng

2020, 2(3): 108-117. doi:10.12133/j.smartag.2020.2.3.202008-SA003

Asbtract ( 1253 )

HTML ( 447)

PDF (1295KB) ( 1218 )

References | Related Articles | Metrics

Vegetables are an important part of residents' diet. The abnormal fluctuation of vegetable prices has caused losses to the economic interests of vegetable farmers and also affected the daily diet and quality of life of residents. However, there are some difficulties in vegetable price prediction, such as large price fluctuation and complicated influencing factors. Cucumber is the main category of vegetables and a common food on the daily table of residents and its recent price fluctuations have aroused widespread concern. In this research, taking cucumber as the research object, a combination model (L-BPNN) combining Lasso regression method and BP neural network was constructed to forecast the short-term price of cucumber. Firstly, the factors affecting the price of cucumber, such as supply, demand and circulation were analyzed. Then the price fluctuation characteristics of cucumber in China from 2010 to 2018 were analyzed and 24 factors were selected as the influencing factors of cucumber price. In the case of complex factors, Lasso regression was used to compress the 24 input influencing factors and the 12 remaining influencing factors with large correlation degree after compression were used as the input influencing factors of BP neural network. Among the 12 related factors , the positive effects included: land cost, per capita disposable income of urban residents, urban vegetable consumption price index, fuel surcharge, booth fee, packaging and processing fee, inflation rate, affected area and temperature deviation from normal value; negative effects included sown area, industrial support amount and average temperature. On this basis, a combination model combining Lasso regression method with BP neural network (L-BPNN) was constructed to forecast the short-term price of cucumber. The neural network was used to train and adjust the model between the input influencing factors and the output price. Compared with the regression analysis and intelligent analysis methods, the results show that the average relative error of L-BPNN combination model was the smallest, only 0.66%, which was 64.52%, 82.11% and 86.2% lower than Lasso regression model, BP neural network model and RBF neural network model respectively, and had higher prediction accuracy. The results of this study realizes the short-term price forecast of cucumber, and can also be extended to other vegetable varieties, which is of great significance for guaranteeing the income of vegetable farmers and stabilizing the market price of vegetables.

Estimation Method of Leaf Area Index for Summer Maize Using UAV-Based Multispectral Remote Sensing | Open Access

SHAO Guomin, WANG Yajie, HAN Wenting

2020, 2(3): 118-128. doi:10.12133/j.smartag.2020.2.3.202006-SA001

Asbtract ( 1285 )

HTML ( 792)

PDF (3246KB) ( 1018 )

References | Related Articles | Metrics

Maize is an important food crop in China. In order to quickly and non-destructively estimate summer maize leaf area index (LAI) under different water stress conditions, in this study, maize samples with multiple irrigation treatments throughout the growth period were used for modeling analysis. Then, based on the unmanned aerial vehicle (UAV) multi-spectral remote sensing technology, combined with the summer maize LAI collected in the field during the same period, five kinds of vegetation indices, including the normalized difference vegetation index (NDVI), soil adjusted vegetation index (SAVI), enhanced vegetation index (EVI), green normalized difference vegetation index (GNDVI) and visible atmospherically resistant index (VARI) were selected in this research as model input parameters, and random forest regression algorithm was used to establish the relationship between the field maize canopy vegetation indices and LAI under different irrigation conditions during the entire growth period. The accuracies of the model were compared with that of the model established by the university linear regression and multiple linear regression algorithms. The results showed that under sufficient irrigation condition, the vegetation index using multiple linear regression model could well (R² = 0.83, RMSE = 0.05) estimate LAI; under water stress conditions, the vegetation index using random forest regression model could well estimate LAI (R² = 0.74~0.87, RMSE = 0.02~0.10), water stress factors had little effect on the random forest regression model, and NDVI and VARI contributed the LAI estimation model better. The spatial distribution map of LAI was generated based on the random forest regression algorithm. The above results showed that it was feasible to use the random forest regression algorithm to estimate the summer maize LAI under various irrigation conditions based on the UAV multi-spectral remote sensing technology. The results indicates that the model established has a good applicability. This research can provide technical and method support for the rapid and accurate monitoring of field summer maize LAI under different irrigation conditions during the entire growth period.

Hyperspectral Estimation Model Construction and Accuracy Comparison of Soil Organic Matter Content | Open Access

LIU Tianlin, ZHU Xicun, BAI Xueyuan, PENG Yufeng, LI Meixuan, TIAN Zhongyu, JIANG Yuanmao, YANG Guijun

2020, 2(3): 129-138. doi:10.12133/j.smartag.2020.2.3.201912-SA004

Asbtract ( 1096 )

HTML ( 390)

PDF (1642KB) ( 822 )

References | Related Articles | Metrics

Soil organic matter (SOM) is an important source of crop growth, its content can reflect soil fertility status. In order to realize the fast and real-time estimation of the SOM, based on hyperspectral data, a rapid estimation model of SOM content in orchards was established. A total of 100 brown soil samples were collected from the apple orchard of Qixia county, Yantai city, Shandong province. After drying and grinding, the hyper-spectrum of the soil was measured in the laboratory using ASD FieldSpec. The spectral data was preprocessed by the method of moving average, and the spectral reflectance features of orchard soil were analyzed to study the correlation between spectral reflectance and its soil organic matter content. In order to enhance the correlation between relevant spectral parameters and soil indexes, the original data were processed by using the multivariate scattering correction, the first derivative and the first derivative of MSC. After the sensitive wavelengths of soil organic matter content were selected and the spectral indexes were constructed. Multiple linear regression models (MLR), support vector machines (SVM) and random forest (RF) models were respectively established. The estimation accuracy of the orchard soil organic matter estimation model was measured by the determination coefficient (R²), root mean square error (RMSE) and relative analysis error (RPD). The sensitive wavelengths of soil organic matter content selected were 678, 709, 1931, 1939, 1996 and 2201 nm. The spectral parameters were constructed using the selected wavelengths, which were NDSI_{(678, 709)}, NDSI_{(678, 1931)}, NDSI_{(678, 2201)}, NDSI_{(709, 1939)}, and NDSI_{(1939, 2201)}. These models established include MLR, SVM and RF model. The RF model had the best precision. The calibration sample R² was 0.8804, the RMSE was 0.1423 and RPD reached 2.25; the R²of the verification model was 0.7466, the RMSE was 0.1266, and the RPD was 1.79. The results showed that the fitting effect of the hyperspectral inversion model based on RF regression analysis was better than that based on MLR analysis and SVM regression analysis. As a promising and effective method, RF can play a vital role in predicting soil organic matter. The results can help understanding the distribution of soil nutrients, guiding farmers to apply fertilizer reasonably and improving the efficiency of orchard production and management.

Application of Satellite Remote Sensing Yield Estimation Technology in Regional Revenue Protection Crop Insurance: A Case of Soybean | Open Access

CHEN Ailian, LI Jiayu, ZHANG Shengjun, ZHU Yuxia, ZHAO Sijian, SUN Wei, ZHANG Qiao

2020, 2(3): 139-152. doi:10.12133/j.smartag.2020.2.3.202006-SA002

Asbtract ( 1159 )

HTML ( 805)

PDF (4211KB) ( 1092 )

References | Related Articles | Metrics

In recent years, revenue protection crop insurance is an innovative insurance that has been prioritized in China. But it still lacks the support of the third-party yield data around crop harvest time. Aiming to provide objective yield data for revenue protection crop insurance, satellite remote sensing production estimation technology was employed to discuss its application mode and applicability. Taking the soybean revenue protection insurance in Jiaxiang county, Shandong province as an example, we first extracted soybean planting plots, calculated vegetation index and crop physiological parameters based on Sentinel-2 satellite images in 2018 . Combining to TRMM precipitation data from TRMM precipitation-monitoring radar satellite and MODIS land surface temperature data from Terra/Aqua satellite and site yield data, we established a multi-parameter linear regression model, and estimated soybean yield per unit area. The crop extraction results showed that the soybean planting area in the study area was 1.24 km², which was in good agreement with the 1.27 km² reported by the local agricultural bureau; and with using the actual measurement plots, the remote sensing identification accuracy of the planting distribution plots reached 90%. The yield estimation results showed that the NDVI of the soybean pod stage on August 23 and the leaf area index of the soybean seedling stage on September 7 explained the soybean yield per hectare the best, and the average estimated yield of the whole area was 244,500 kg/m², which reflects the severely affected agricultural conditions, comparing to 299,800 kg/km² in previous years.The regression coefficient between the estimated yield data and the measured data reached 0.92, which meet the application needs.With this results, the estimated yield of different towns can be summarized, and the regional yield was present, and was used as the real yield in 2018, multiplying with the average soybean price around October 11 to December 10 from the local price bureau, the real revenue was obtained. Compared the real revenue to the expected revenue in the contract of insurance, the claims work was decided. The results indicated that the Sentinel-2 satellite data could be used to identify the soybean planting distribution in the study area accurately, and to complete the yield estimation as soon as one week after the soybean harvest, which could guide the insurance company's claims work. The whole methodology is capable of aiding the claims work in revenue protection crop insurance.

Table of Content