Natural Resource Monitoring of Rheum tanguticum by Multilevel Remote Sensing

Remote sensing has been extensively applied in agriculture for its objectiveness and promptness. However, few applications are available for monitoring natural medicinal plants. In the paper, a multilevel monitoring system, which includes satellite and aerial remote sensing, as well as ground investigation, was initially proposed to monitor natural Rheum tanguticum resource in Baihe Pasture, Zoige County, Sichuan Province. The amount of R. tanguticum from images is M = S*ρ and S is vegetation coverage obtained by satellite imaging, whereas ρ is R. tanguticum density obtained by low-altitude imaging. Only the R. tanguticum which coverages exceeded 1 m2 could be recognized from the remote sensing image because of the 0.1 m resolution of the remote sensing image (called effective resource at that moment), and the results of ground investigation represented the amounts of R. tanguticum resource in all sizes (called the future resource). The data in paper showed that the present available amount of R. tanguticum accounted for 4% to 5% of the total quantity. The quantity information and the population structure of R. tanguticum in the Baihe Pasture were initially confirmed by this system. It is feasible to monitor the quantitative distribution for natural medicinal plants with scattered distribution.


Introduction
Remote sensing has been extensively applied in agriculture in recent years. This approach can provide objective, accurate, and timely information on the ecological environment of crops. Many remote sensing applications for medicinal plants are focused on cultivated medicinal plants. By contrast, natural medicinal plants are largely ignored because of their scattering distribution characteristic and small coverage area. A multilevel monitoring system, which includes satellites, aerial remote sensing, and ground investigation was proposed in this paper to monitor natural Rheum tanguticum in the Baihe Pasture, Zoige County, Sichuan Province, to determine the quantitative distribution and population structure of R. tanguticum in this region.
A number of studies indicate that domestic R. tanguticum are mainly distributed in temperate Asian zones, including the Greater Khingan Mountains, Taihang Mountains, Qinling Mountain, Daba Mountains, and Yunnan-Guizhou Plateau [1,2]. R. tanguticum is a peculiar Chinese perennial herb that belongs to the Polygonaceae family. This herb mainly thrives in Southern Gansu Province, Northwest Sichuan Province, and Northeast Tibet. These areas are characterized by high altitudes and short frost-free periods. R. tanguticum grows slowly in these regions and has rich substances with good qualities. The popular "Xining Rheum" and "Quanshui  Rheum" originated from these areas. Recent studies on R. tanguticum are mainly focused on ecological suitability, chemical analysis, and pharmacology [3][4][5][6][7]. The demand for R. tanguticum as an ingredient for Tibetan medicine, health products, and specialty food has gradually increased [8]. R. tanguticum in the Qinghai-Tibet Plateau proliferates slowly because of the cold climate and fragile ecological environment. Moreover, the unreasonable excavation of R. tanguticum leads to a significant reduction in the quantity of this resource and may even lead to desertification. Determining the updated status of the quantitative distribution and population structure of R. tanguticum is necessary for ecological protection and resource development [9]. Thus, the multilevel monitoring system was established to monitor the updated status of R. tanguticum [10][11][12]. Satellite remote sensing and low-altitude remote sensing were initially applied to monitor the quantitative distribution and population structure of natural R. tanguticum in the Baihe Pasture. The viability and effectiveness of the multilevel remote system were evaluated by ground-sampled surveys. This paper presented a new technical method and platform for monitoring wild and scattered medicinal plant resources.  The low-altitude remote sensing system consisted of a flight platform, digital camera, single-axis stabilized platform, and equipment control system ( Figure 2). The flight platform is an unmanned aerial vehicle from the Institute of Medicinal Plant Development. The Canon 5D digital camera with a charge-coupled device array was adopted for data acquisition. The camera also has array pixels and a lens focal length of 4368 pixels × 2912 pixels and 35 mm, respectively. The 0.1 m resolution of the camera for aerial images was appropriate for the requirements of the survey goal based on the ground survey. The high-resolution aerial photography for wild R. tanguticum was initially conducted in the Baihe Pasture of Zoige County by the low-altitude remote sensing system.

Data Analysis
3.1. Image Processing. Bands 4, 3, and 2 in the TM image were fused to distinguish the vegetation and water. The Baihe Pasture boundary in the TM image was determined via the registration function by the feature points. Thus, the Baihe Pasture satellite image was detached from the original TM image. NDVI is an internationally accepted criterion for vegetation coverage level. Consider that NDVI = (NIR − R)/(NIR + R), where NIR denotes the near-infrared band reflectivity and R stands for red band reflectivity. An NDVI value in the range of zero to one generally refers to vegetation regions. The vegetation region of the Baihe Pasture was determined based on the NDVI analysis in Figure 4.

Object-Based Image Classification.
High-spatial resolution images have limited spectral information but contain many spatial information of objects, such as size, shape, and topological information. The object-oriented approach classifies different objects based on spectral signature, shape, and contextual relationship. Homogeneous objects were initially formed by multiscale image segmentation. Thereafter, image classification was used to endow the objects with different semantic information based on size, spectrum, and shape  parameters. These objects were then used as a basis for the fuzzy classification of the image. The use of spectral, textural, and shape properties, as well as fuzzy thinking, might reduce uncertainty in the classification process. The results of the object-based image interpretation showed better integration levels than the pixel-based interpretation method [9,13,14], which is suitable for high-spatial resolution images.

Biological and Community
Characteristics of R. tanguticum. R. tanguticum is a perennial and tall herb that is resistant to cold but has zero endurance to high temperatures. This herb thrives in a cool and moist climate and is mainly scattered in forests, scrubs, and meadows. Natural R. tanguticum plants in the Baihe Pasture can grow for approximately seven or eight years with a height of 1.5 m to 2.0 m and a corolla diameter of 2.4 mm to 3.7 mm. The blooming stage of R. tanguticum is from May to July. The ecology community type of R. tanguticum is alpine meadows with only an herb layer. The vegetation coverage of such alpine meadows is more than 95% and includes a small amount of cultivated barley. Figure 5  others. Cyperaceae and Gramineae assume absolute superiority among these plants. Ranunculaceae, Compositae, and a few fungi compose the weed layer. Figure 6 and Table 1 show the vegetation type and accompanying plants of natural R. tanguticum in the Baihe Pasture, respectively.     (27,600 ha), respectively. The results were consistent with the 43,333 ha provided by the Baihe Pasture. Therefore, remote sensing technology was feasible in monitoring the fields.

Multilevel Remote
As a matter of experience, only R. tanguticum with more than 1 m 2 of coverage (called effective resource) could be interpreted from the image. Thus, the average density of available R. tanguticum could be calculated by the interpretation results of many low-altitude images. The effective resources of R. tanguticum in the entire Baihe Pasture were that the Baihe Pasture vegetation area multiplied by the average R. tanguticum density, which was equal to 350,000. Therefore, the amount of available natural R. tanguticum resources in the Baihe Pasture in 2008 was 350,000.  Table 2).
The ground survey results in 2010 indicated that the number of R. tanguticum in each fence was approximately 1,520, except for approximately 855 in the third fence. Thus, approximately 3.8 R. tanguticum per 100 m 2 existed, and the total amount of R. tanguticum was 1, 049 × 10 4 .
The results in 2011 indicated that the average number of R. tanguticum in each fence was approximately 1,220; the minimum amount was approximately 300 in the second fence, and the amount was approximately 1600 in the first, fourth, and sixth fences. The calculation result was 3.1 R. tanguticum per 100 m 2 . Thus, the total amount of R. tanguticum in 2011 was 856 × 10 4 .
Quantity variance existed in the total amount of R. tanguticum based on the results of different years because of operation errors and other factors. However, the total quantity of R. tanguticum was approximately 750 × 10 4 to 1, 000 × 10 4 . The current available amount and the future total amount of natural R. tanguticum resources in 2008 were, respectively, 35×10 4 and 718×10 4 based on the results of the multilevel remote sensing and ground survey. The population structure of natural R. tanguticum, that is, the available amount of R. tanguticum, comprised 4% to 5% of the total amount. Therefore, the safe digging quantity of R. tanguticum in the Baihe Pasture was 4% to 5% for the protection of natural resources.

Discussion
Remote sensing technology has become an important means of obtaining geographic environment information. This technology has remarkably extended and improved the ability of humans to recognize their surrounding environment. Technology innovation in remote sensing provides a new method and platform for surveying natural Chinese medicine resources (1) Remote sensing has been extensively applied in agriculture in recent years [15][16][17][18][19][20][21][22][23][24][25]. Remote sensing technology can provide objective, accurate, and timely information on the ecological environment of crops and is an important data source for precision agriculture [26]. However, a significant difference exists in the resource distribution between agricultural and natural Chinese medicinal herbs. Artificially cultivated crops are characterized by large areas, centralized distributions, and regular shapes. Thus, these crops have significant image features that are easy to interpret. By contrast, natural Chinese medicinal herbs often exist in wide, inaccessible regions and are scattered in different vegetation communities. Thus, interpreting natural Chinese medicinal herbs from satellite images was difficult because of the lack of distinct image features. Multilevel remote sensing was initially 8 Evidence-Based Complementary and Alternative Medicine proposed to monitor the natural R. tanguticum in the Baihe Pasture. Based on the satellite image results, the Baihe Pasture area was 445 km 2 (44,500 ha), which was consistent with the 43,333 ha provided by the administrative department of the Baihe Pasture. The vegetation coverage in the Baihe Pasture was 276 km 2 . R. tanguticum density was 2.6 to 3.8 per 100 m 2 based on the result of the multilevel remote sensing. Compared with the ground investigation, the multilevel remote sensing technology system was fast and convenient. Results indicated that the available amount of R. tanguticum comprised 4% to 5% of the total amount. The quantity information and population structure of R. tanguticum in the Baihe Pasture of Zoige County were initially confirmed by this multilevel monitoring system. This system was also successfully applied in monitoring Ferula sinkiangensis in the fourth survey of the Chinese medicine resources of Xinjiang.
(2) An object-oriented image-recognition method was initially adopted to extract R. tanguticum in low-altitude aerial images. Only R. tanguticum with coverage greater than 1 m 2 could be extracted from the low-altitude aerial image because of the spatial resolution limitations. Therefore, the amount of R. tanguticum obtained from multilevel remote sensing was the current available resources, whereas the results from the ground investigation were the total resources because all R. tanguticum sizes could be observed through ground surveys. The amount of R. tanguticum obtained from the ground survey represented the total available resource in three years to five years. The results from the two methods (remote sensing, ground survey) represented the present and the future amounts of R. tanguticum. Thus, the corresponding method should be selected to satisfy different demands and purposes.
(3) The optimum time for monitoring R. tanguticum through remote sensing was early July because R. tanguticum was in bloom and taller than the surrounding vegetation during this period. The ground survey results in three years indicated that the total quantity of R. tanguticum was approximately 750 × 10 4 to 1, 000 × 10 4 . The available amount of natural R. tanguticum resources was 35×10 4 in 2008 based on the results of the multilevel remote sensing. However, the total amount in 2008 was 718 × 10 4 based on the ground survey results. So, the safety digging quantity of R. tanguticum in the Baihe Pasture was 4% to 5% for the protection of natural resources.
We conclude that the use of low-altitude remote sensing technology to monitor the scattered distribution of natural medicinal plants was feasible. This study provided a new technical system for monitoring objects with scattered distributions.