In recent years, UAV remote sensing has rapidly emerged in grassland resource monitoring and grassland ecology due to its advantages of high resolution, high timeliness, strong maneuverability, and low-altitude flight under clouds. This paper first introduces the composition of the UAV remote sensing system and the application of different sensors in grassland monitoring. The application research in rodent monitoring is reviewed, and the problems and limitations of this technology are discussed, so as to solve the objects and problems that need to be monitored, which is helpful for constructing accurate data in the survey of grassland animal and plant resources. Obtaining and realizing real-time dynamic monitoring has important research significance and practical application value.

Unmanned aerial vehicle (UAV), referred to as unmanned aerial vehicle (UAV), is an unmanned aerial vehicle that can carry a variety of equipment, perform multi-domain tasks and fly autonomously through remote control equipment. The combination of UAV and remote sensing technology, that is, UAV remote sensing (unmanned aerial vehicle remote sensing, UAVRS). UAV is used as a carrier to carry cameras (including visible light cameras, multi-spectral cameras, thermal infrared cameras, etc.), Lidar and other sensors are used to obtain low-altitude high-resolution remote sensing data. Compared with the traditional satellite-based aerospace remote sensing, UAV remote sensing has the advantages of low-altitude flight under clouds, high mobility, etc. The characteristics of temporal and spatial resolution are also different from traditional satellite remote sensing with a long revisit period and hundreds of kilometers above the ground.
feel possessed. Compared with traditional ground field surveys, UAV remote sensing does not require a lot of manpower and material resources, has fast response, low cost, strong timeliness, and has a wide range of applications. It is the third generation of remote sensing technology after traditional aviation and aerospace remote sensing platforms. With the development of related technologies, UAVs have developed rapidly in the fields of ecological monitoring, environmental monitoring, disaster investigation, precision agriculture, grassland ecological monitoring, etc., and have become a hot topic for scholars at home and abroad. Based on the application of UAV technology in grassland resource monitoring, this paper explores the research content and methods of this technology in this field, and summarizes the current problems, limitations and development prospects of UAV technology.

1 System composition of UAV remote sensing

1.1 System composition and workflow

The UAV remote sensing system mainly uses UAV as the flight and carrying platform. By carrying various sensors, combined with the ground control and data transmission system, it can obtain ground or aerial real-time images and various remote sensing data. The system composition mainly includes ground system, mission load and aircraft system. Based on its system composition and working principle, the comprehensive summary of the UAV image data acquisition process is shown in Figure 1.

1.2 Photo processing technology
The complexity of the data processing process is further exacerbated by the differences in the results required to process the photos. The Complexity of Data Processing of Grassland Animal and Plant Resources

The nature is not only reflected in the data samples themselves, but also in the heterogeneity, multi-source and multi-space interaction of grassland ecosystems. In the research of grassland ecology, the demand for different types of data in different fields is gradually increasing, which also makes the processing of experimental data further refined and precise. With the rapid development of UAV technology and computer software processing technology, more UAV image processing research has begun to use the automatic interpretation function of the software, and through comparative analysis, the best data processing renderings have been obtained. In the process of UAV video data processing, PIX4D and Photoscan are the most commonly used data processing software. The processing flow of Pix 4D software is: import the original photo – fill in various parameters (photo GPS location information, shooting height, overlap, etc.) – obtain a digital orthograph model (DOM) with geographic coordinates, a digital elevation model ( digital elevation model, DEM), digital surface model (digital surface model.DSM) and 3D model map—the software automatically mosaics and equalizes color—completes the image stitching. The process of PhotoScan processing software is: actively creating new items, importing photos, stitching multiple photos, generating dense point cloud, outputting digital elevation model and digital orthophoto image, exporting stitching results, generating thematic map. PhotoScan software completes the stitching of pictures according to the overlapping degree and coordinate elevation information of the pictures. It is mainly used in the generation of some thematic maps, such as soil erosion area maps, forest and grass coverage maps, soil erosion intensity maps, etc. Among the above two types of image processing software, Pix 4D has the advantages of specialization, simplification, and one-key A-movement, etc., but it cannot complete the automatic selection of mounds and bald spots in photos. PhotoScan mainly has the advantages of simple operation and support for oblique images. , supports the advantages of multi-aircraft and multi-resolution images, but lacks orthophoto editing and modification functions. Therefore, in the extraction of photo information, a variety of software can be used for measurement processing, so as to derive understandable data information. For the utilization of animal and plant data and information of complex grassland resources, it is not only necessary to generate and analyze data, but also to enable a large amount of data to be reused and recycled, so as to realize the sharing of existing resources and lay the foundation for subsequent research.

1.3 Common Sensors and Application Fields

High-resolution digital cameras such as visible light, multispectral, hyperspectral, thermal imager, and lidar are common UAV sensors. The output results, advantages and disadvantages, and main application fields of each sensor are summarized and classified in Table 1.

With the continuous deepening of related research, the application of sensors is becoming more and more extensive. The target data obtained by different sensors are different. But in general, visible light cameras, hyperspectral cameras, and multispectral cameras are widely used in the acquisition of various plant data. Thermal infrared cameras are closely related to temperature maps, and are used in animal research. It is widely used, and lidar is more used to obtain plant canopy structure, so it is more used to obtain forest plant biomass and vegetation height.

2 Application of UAV in monitoring and management of grassland animal and plant resources

Grassland ecosystem is one of the important terrestrial ecosystems and plays a pivotal role in the ecological environment. As an indispensable part of the grassland ecosystem, grassland resources play an important role in the cycle of the grassland ecosystem. In order to conduct fast, convenient, accurate and effective monitoring of grassland resources, scholars at home and abroad have begun to use satellite-based space remote sensing to carry out dynamic monitoring research on grassland vegetation coverage and biomass, grassland rodent monitoring, grassland ungulate wild animals. . However, the traditional medium and low resolution satellite remote sensing images have a long acquisition cycle and are easily affected by climate, and cannot obtain effective ground information in local areas. Compared with satellite remote sensing, UAV remote sensing has the characteristics of high resolution and image acquisition under the cloud, which can significantly reduce the impact of mixing effects on monitoring accuracy, effectively making up for the low surface resolution and long revisit cycle of satellite space air remote sensing systems. , and is greatly affected by water vapor, which provides a new method for the application and research of remote sensing monitoring of grassland resources at medium and small scales.

2 .1The application of unmanned aerial vehicle in the monitoring of grassland plant resources
2.1.1 Monitoring of grassland vegetation coverage The grassland ecosystem needs vertical vegetation structure to evaluate the health status of grassland. Vegetation Cover Index

The percentage of the vertical projection area of ​​vegetation in the observed K domain to the total surface area of ​​the observed K is a direct quantitative indicator reflecting the growth status of vegetation, and is also a key parameter for evaluating and monitoring ecosystems and their functions. Vegetation cover and its changes over time are also directly used as indicators of grassland degradation, soil erosion and desertification.
At present, many scholars at home and abroad have used UAV remote sensing to monitor grassland vegetation coverage. Ge Jing et al. used unmanned aerial vehicle (UAV), ordinary digital camera (Canon), agricultural multispectral camera (Agricultural digital camera, ADC) and other equipment to obtain a large number of photos of alpine grassland in the eastern region of the Yellow River source, combined with the corresponding MODIS NDVI (normalized The inversion model between vegetation coverage and MODIS vegetation index based on UAV, Canon and ADC photos was constructed, and the leave-one-out cross-validation method was used to evaluate the performance of various models. Therefore, it is established that the grassland coverage inversion model constructed by the data obtained by the UAV is the optimal model for remote sensing monitoring in the source area of ​​the Yellow River. Song Qingjian et al. established a regression model between the two vegetation indices in the alpine grassland of Gannan Prefecture, using EVI and NDVI as independent variables, and the grassland vegetation coverage data obtained by UAV as the dependent variable, and obtained the results with Canon digital cameras. The data of grassland vegetation coverage is the real value, and the established regression model is evaluated for accuracy, and the logarithmic model based on EVI is selected as the optimal inversion model of grassland vegetation coverage in the study area.

Considering the influence of model accuracy and stability on grassland coverage monitoring, Meng et al. used UAV technology to conduct a comparative study of inversion models for alpine grassland coverage in southern Gansu, based on 14 factors related to grassland coverage, respectively. Single-factor and multi-factor parametric inversion models and multi-factor non-parametric inversion models are established. Through cross-validation, it is concluded that the multi-factor non-parametric model has higher accuracy and stability than the single-factor and multi-factor parametric models. Based on the good fitting between the vegetation coverage obtained by UAV aerial photography and the vegetation coverage measured on the ground, Yi Shuhua et al. The correlation coefficient between vegetation cover and remote sensing index shows that the former has a larger correlation coefficient, indicating that the use of UAVs can provide high-precision ground cover information. Chen et al. obtained the vegetation coverage at two spatial scales (satellite image pixel scale and ground quadrat scale) using a combination of UAV aerial survey and ground quadratic survey, and evaluated the vegetation coverage at the satellite image pixel scale. The accuracy of the partial vegetation coverage (FVCground) estimated by the ground sampling method and the partial vegetation coverage (FVCUAV) estimated by the UAV sampling method, and NDVI, EVT, RVI (ratio vegetation index), and MSAVI (modified soil adjustment) were selected. The relationship between vegetation index and FVC was analyzed, and it was found that FVCUAV was the most accurate at the pixel scale of satellite images. These studies have led to the further application and development of UAV technology in grassland vegetation coverage monitoring.

2.1.2 Estimation of Aboveground Biomass of Grassland Plants Aboveground biomass of grassland is an important measure for the dynamic research of natural grassland ecosystem, and an important basis for rational utilization of grassland resources and monitoring of stocking balance. Aboveground biomass can evaluate grassland ecosystem productivity, assess grassland growth and yield. Under actual production conditions, changes in aboveground biomass can reflect the degree of grassland growth and utilization, and can provide early warning and reference thresholds for grassland protection and management. Therefore, accurate estimation of above-ground biomass in large-scale grassland is of great significance for evaluating the application status and management of grassland resources.
Lusscm applied height features and vegetation index features in red, green and blue (RGB) bands derived from low-cost UAV image data as predictors for estimating grassland biomass, and compared them with the narrow-band vegetation index measured by spectrometer. It is feasible to estimate grassland biomass based on the turf height derived from UAV image SFM-MVS (Structure from Motion-Multi View Stereo). Therefore, the application of UAV-based imaging sensors is a fast and non-destructive data acquisition method with high temporal and spatial resolution. Zhang Zhengjian et al. obtained the aboveground biomass distribution of grassland in the study area based on the ground-measured sample tree data and UAV visible light images, and established biomass to green-red ratio index (GRRI), green-blue ratio index (GBRI), normalized green-red ratio index (GBRI). Exponential regression models such as difference index (NGRDI), normalized green-blue difference index (NGBDI), etc.; by comparing vegetation index models established in different bands, it is determined that the GRRI and NGRDI vegetation index regression models based on red and green bands affect biomass. The simulation and prediction accuracy is good, and it can be used for the estimation of regional grassland biomass. In order to quickly, effectively and accurately estimate the above-ground biomass of natural grasslands, Sun Shize et al. used a multi-rotor DJI UAV to obtain high-resolution multispectral images containing near-infrared bands according to different grassland types and vegetation types on yin and yang slopes, combined with ground measurements. The correlation analysis was carried out on the above-ground biomass of grassland and NDVI, RVI, VDVI, MSAVI, and DVI (difference vegetation index) 5 vegetation indices, and an estimation model was established. Good, highest precision. The macroscopic, dynamic and comprehensive nature of earth observation by UAV remote sensing technology makes it superior to the traditional point-to-surface field survey method, and has become an important means for humans to obtain information on large-scale macroscopic grassland resources.

2.2Application of UAV in monitoring grassland animal resources
2.2.1 Monitoring of wild ungulates in grasslands Monitoring of wild ungulates in grasslands is an important and challenging task.

Requires a significant investment of time and resources. The resolution or scale of traditional field survey data often does not match the data obtained by remote sensing methods.

Testing species and poinsettias at different spatial scales often results in inconsistent patterns of relationships between the two

In the monitoring of ungulate wildlife. UAV remote sensing technology is a feasible and efficient monitoring tool. Torney and others deploy drone systems to collect reindeer

Aerial footage from Victoria Island to Canada

, replicating the fine interaction rules of neighbor direction selection. Luo Wei et al. took the Sanjiangyuan area of ​​Qinghai as the research area and the Tibetan wild donkey as the research object, and discussed the

The method of automatically obtaining wild herbivore information and counting the number of wild herbivores from UAV images has the characteristics of high speed and high precision. on the basis of,

Shao Quanqin and others used drone aerial photography to investigate the image interpretation signs of ungulates such as Tibetan wild ass, Tibetan antelope, Tibetan sheep, and yak in Maduo County, the source of the Yellow River

The library can be interpreted through human-computer interaction to obtain the number of populations in the survey transect. Guo Xingjian and others used drones to survey rocks in Maduo County, the source of the Yellow River.

The sheep were taken aerial photos, and the software Pix4Dmapper and LiMapper were used to stitch the photos, and the population of the blue sheep in the study area was estimated through visual interpretation.

The number and density of populations, combined with software ArcGIS and 3S technology to study and analyze their habitats, so as to provide information for the investigation and research of large wild animals in the plateau area.

provides new ideas. In the monitoring of zebra habitats, Xu et al.

One is divided into virtual grids, each grid contains a cluster of sensor nodes, and one of the nodes is selected as the cluster head to receive wireless sensor data.

data packets and send them to the drone, which acts as a mobile receiver to collect the data. The network model utilizes the real moving tracks of zebras

Traces to explore the regularity of its activities to achieve maximum visual perception of the target. The simulation results show that the path planning method outperforms the random sum

TSP-based path planning method (Figure 2). Luis et al. use a system that combines UAVs with thermal imaging capabilities and human intelligent image processing to locate

The location of wildlife in its natural habitat addresses the challenge of automatic detection of wildlife in drone imagery. In recent years, due to aerial imaging

With the advancement of technology, the possibility of the application of drones in professional fields has increased. Aerial imaging provides detailed grassland images and dynamic

The rapid detection of biological communities is a very effective tool to solve the monitoring of large wild animals in grasslands.

2.2.2 Monitoring of small rodents in grasslands

The grassland has been degraded to varying degrees and has become more serious, and rodent damage is frequent. Therefore, the temporal and spatial dynamics and distribution of grassland rodent damage need to be scientifically recognized.

Awareness and quantitative assessment. It is a new research idea to apply drones to the survey of rodent damage area, rodent damage distribution monitoring and assessment in alpine grassland.
Since the 1980s, scholars have successively carried out research on the application of large-area grassland remote sensing technology and research on the prediction and prediction of grassland rodent damage.

, but still in the exploratory stage. Li Bo et al. applied the "3S" technology to establish a dynamic monitoring system for the balance of grass and livestock in temperate grasslands in China.

Condition monitoring provides an important foundation. On this basis, some studies have used the method of combining "3S" technology and ground investigation to determine the impact of grass

The biological and abiotic factors of the original rodent infestation were used to build a rodent density monitoring model based on 3S technology; other scholars analyzed the effect of "3S" technology on grass

The principle and application of primary rodent monitoring, and pointed out that TM remote sensing images can be used as the main data source for grassland rodent research; Li Peixian et al.

Combined with the GPS data collected on the spot, monitor and interpret the grassland rat wasteland and rat infestation areas in Altun Mountains, and obtain the rat infestation area.

According to the habitat characteristics such as vegetation coverage and altitude in the rat infestation area, the spatial distribution characteristics of the rat wasteland and the rat infestation area were analyzed.

Some scholars use a low-altitude remote sensing platform constructed with powered delta wings and digital cameras to take aerial photos of rat wasteland to obtain high-resolution prairie rat infestation figures

Based on the visual interpretation of remote sensing, the spatial distribution and damage degree of rodents in the experimental area were obtained by using GIS spatial interpolation and statistical methods.

; In addition, He Yongqi and others used "3S" technology to determine the elevation, slope, slope aspect, grassland type, soil type, EVI 6 rodent monitoring models

The main factors were used to establish a grassland rat pest monitoring model based on "3S" technology, and the thresholds for different rat pest occurrence areas and damage areas were extracted.

UAV remote sensing can flexibly obtain multi-scale and multi-temporal ground observation data, which can be better used in rodent research. Taking DJI's Elf series drones as an example, it can efficiently and accurately provide information such as vegetation coverage, patches, and mouse holes on the ground (Figure 3), and manually mark the pictures obtained by the drone for identification. The bald spot formed by plateau pika or zokor; Dong Guang took the area with severe rodent infestation in Xieman Township, Ruoergai County as the experimental research site, and used the Elf 4Pro to obtain aerial video of the zokor and marmot rodent-infested areas in spring and summer. The software Pix4Dmapper is used to generate digital orthophoto images, and the rodent damage information of the pictures is obtained by four methods: gray threshold segmentation, object-oriented method, BP neural network and optimal color texture. The best method, thus laying the foundation for the follow-up monitoring research of prairie rodents. In the monitoring and identification of mouse holes in the typical area of ​​Maduo County, Sanjiangyuan, Zhou Xiaolin et al. used the UAV images in the visible light band as the data source, and established the support vector machine method and the object-oriented template matching method in the study area. Research. The results show that the object-oriented template matching method has higher recognition accuracy for rat holes in grassland with different vegetation coverage. Ma Tao et al. used the UAV low-altitude remote sensing monitoring platform to take two aerial photos of the study area to discuss the effectiveness of low-altitude remote sensing in monitoring the spatial density distribution of giant gerbils, and to evaluate the degree of rodent damage in the area, which is the most important for desert forests. Provide the basis for scientific prevention and control of gerbils. However, the research on grassland rodent monitoring based on UAV remote sensing is still in its infancy. In order to give full play to the obvious advantages of UAVs in quickly obtaining high-resolution images, a model can be established based on the geographical features of grasslands, combined with ground survey data, and established models. Fast and accurate monitoring of grassland ecological rodent damage.

3 Problems, limitations, prospects and evaluation of UAV remote sensing technology in grassland resource survey and monitoring
3.1 Problems and limitations
Judging from relevant research at home and abroad, there are still many problems and limitations in the extraction of UAV information, which are mainly summarized as follows:
1) Complex terrain effect: The large spatial variability across regions, the heterogeneity of the ground, and the incomplete ground sampling lead to the lack of data integrity.

In addition, the resolution of field surveys is extremely high, and dead leaves and vegetation can be distinguished, while aerial images can only distinguish vegetation patches, which leads to the detection of unmanned aerial vehicles.

Vegetation coverage is generally higher than field surveys.
2) There are many models to establish: due to the different conditions of water and heat conditions, vegetation characteristics, soil factors and other conditions in different study areas, UAV data is finally used.

The determined models are different, resulting in low monitoring and estimation accuracy and poor stability of grassland remote sensing vegetation.
3) Photo recognition software development: In the grass picture information obtained by the drone, some target information is often not obtained due to the occlusion of obstacles

, which requires researchers to observe with the naked eye, which greatly reduces the work efficiency and also limits the acquisition of other information in the image. Therefore, developing

Simpler and more efficient picture automatic recognition processing software is particularly important.
4) The matching limitation of sensor equipment: UAV has little research on grassland resource monitoring, although a small number of scholars have started research in this area

However, due to the high cost of advanced sensor equipment and the lack of ground survey data that is highly matched with UAV remote sensing, it is greatly limited.

Application of man-machine in grassland resource monitoring.
3.2 Evaluation and Outlook
With the wide application of UAV remote sensing technology, the research on grassland animal and plant resources has entered a new stage.

Satellite remote sensing, real-time, comprehensive, effective and accurate monitoring of grassland resources.
1) Multi-scale and multi-level monitoring: aiming at the horizontal and vertical zonal characteristics of grassland distribution, local microenvironment differences and grassland vegetation

Long-term seasonal changes can be combined with ground surveys, UAV remote sensing, and satellite space remote sensing to achieve small, medium and large scale grasses.

Multi-level monitoring of land resources, improve data matching in different spatial regions, and reduce the gap between data obtained by UAV remote sensing and field survey data.

difference.
2) Time-sensitive fusion of multivariate data: Most studies using drones to monitor grassland resources still rely on traditional ground surveys or "3S"

Data obtained by satellite remote sensing technology. As an important means of grassland ecology research, ground survey has an irreplaceable role.

It improves the timeliness of data acquisition to a certain extent. Therefore, through the method of multivariate data fusion, the data obtained from the ground survey can be combined with the

The data obtained by man and machine are fused to make up for the insufficiency of a single data source, to realize the complementary advantages of various data sources, and to achieve spatial resolution and

The complementary temporal resolution provides more systematic, scientific and effective support for grassland resource monitoring.
3) Popularization of the application of different sensors: With the development of integrated technology, the sensors that can be carried by UAVs are more diverse. Variety of grass types

In this way, suitable sensors can be selected according to the characteristics of different grassland resources, and the objects and problems to be monitored can be solved in a targeted manner. This pair

It has important research significance and practical application value in constructing accurate data acquisition and realizing real-time dynamic monitoring.
In general, the UAV grassland ecological monitoring mechanism based on high temporal and spatial resolution will become a hot spot and trend in future development. drone remote sensing

Its many advantages over satellite remote sensing and ground-based manual surveys provide a new technical platform for the research on grassland resource monitoring, especially in grasslands.

The seasonal phase of ground plants, grassland plant coverage, biomass yield, grassland livestock, grassland rodent populations, etc., can be targeted for large-scale

The accumulated aerial monitoring and small-scale fixed-point monitoring are of great practical value for the rational use of grassland and health management.

Reviewing Editor: Fu Ganjiang

Leave a Reply

Your email address will not be published.