ZHANG Dong-yan, LAN Yu-bin, WANG Xiu, ZHOU Xin-gen,CHEN Li-ping*, LI Bin, MA Wei

1. Beijing Research Center for Information Technology in Agriculture, Beijing 100097, China 2. Anhui Engineering Laboratory of Agro-Ecological Big Data, Anhui University, Hefei 230601, China 3. Key Laboratory of Agri-informatics, Ministry of Agriculture, Beijing 100097, China 4. College of Engineering, South China Agricultural University, Guangzhou 510642, China 5. Texas A&M AgriLife Research and Extension Center, Beaumont, TX 77713, USA

Assessment of Aerial Agrichemical Spraying Effect Using Moderate-Resolution Satellite Imagery

ZHANG Dong-yan1, 2, 3, LAN Yu-bin4, 5, WANG Xiu1, 3, ZHOU Xin-gen5,CHEN Li-ping1, 3*, LI Bin1, 3, MA Wei1, 3

1. Beijing Research Center for Information Technology in Agriculture, Beijing 100097, China 2. Anhui Engineering Laboratory of Agro-Ecological Big Data, Anhui University, Hefei 230601, China 3. Key Laboratory of Agri-informatics, Ministry of Agriculture, Beijing 100097, China 4. College of Engineering, South China Agricultural University, Guangzhou 510642, China 5. Texas A&M AgriLife Research and Extension Center, Beaumont, TX 77713, USA

Remote sensing technique can be used to examine the effects of agrichemical application on the performance of field crops at a large scale in an effort to develop precision agricultural aerial spraying technology. In this study, an airplane M-18B at the 4-m flight height was used to spray a mix of agrichemicals (a fungicide and a plant growth regulator) to control rice leaf blast disease and improve the growth vigor of rice plants in the field. After the aerial spraying, satellite imagery of tested area was acquired and processed to calculate vegetation indices (VIs). Ground agrichemical concentration data were also collected. The relationships between droplets deposition and VIs were analyzed. The results indicated that the highest correlation coefficient between single phase spectral feature (NDVI) and droplets deposition points density (DDPD, points·cm-2) was 0.315 withP-value of 0.035 while the highest correlation coefficient between temporal change characteristic (MSAVI) and droplets deposition volume density (DDVD, μL·cm-2) was 0.312 withP-value of 0.038). Rice plants with the greatest growth vigor were all detected within the spraying swath, with a gradual decrease in the vigor of rice plants with the increase of droplets drift distance. There were similar trend patterns in the changes of the spraying effects based on the spatial interpolation maps of droplets deposition data and spectral characteristics. Therefore, vegetation indexes, NDVI and MSAVI calculated from satellite imagery can be used to determine the aerial spraying effects in the field on a large scale.

Satellite imagery; Vegetation index; Aerial spraying; Droplet deposition; Drift

Introduction

Recent global climate changes have increased worldwide crop diseases, insect pests and natural disasters, leading to a significant reduction in crop production[1-2]. It has been a tremendous challenge to reduce the losses caused by disease and insect pests to ensure adequate and safe food supply for the increasing population of the world[3-4].

The development and implementation of agricultural aerial application technology have the high potential of effectively managing disease and insect pests while reducing environmental and economic costs by reducing the use of pesticides. This technology can be operated on a large scale with a low cost[5-6]. The technology has been used in some of major countries in the world. Up until 2013, the proportion of agricultural aviation services to be used in agricultural production is 65%, 38%, 35% 32%, 32%, 30%, and 20% for the United States, Japan, Russia, Australia, Brazil, and South Korea respectively. However, aviation services in China only accounts for 6.2% (2012)[7-8]. It is very important for China to increase the agricultural aviation services to ensure adequate and safe food supply for its almost 1.4 billion people, accounting for 20 percentage of the world’s population. Beidahuang General Aviation Company (BGAC), the largest civil aviation company in China, has been the leader toward that direction. BGAC has successively bought more than 30 airplanes for the agricultural aviation services from 2010 to 2014. Such efforts of BGAC have reduced the losses caused by disease and insect pests in the Northeast, a major grain crop production area in China[9]. BGAC has also provided research instruments that have significantly improved the aerial application technology.

Remote sensing technology, an important part of precision agriculture, can quickly detect the areas or crop infested with disease and insect pests[10-12], assess their severity[13-14], guide aerial spraying[15-16]and evaluate the operation effect in the field[17-18]. Lan et al.[15]developed an integrated system through the collection and analysis of aerial images in the operation area to generate a severity prescription map of disease and insect pests to implement the variable aerial spraying. Huang et al.[19]designed a data collection platform based on an unmanned aerial vehicle (UAV) for agricultural pesticide spraying. These studies have adequately shown the advantages of flexibility and easy operation to improve agricultural production management.

Currently, the assessment of aerial spraying effect is primarily implemented based on a low-altitude platform. Huang et al.[17]and Ortiz et al.[18]used aerial multispectral images to evaluate the damage severity caused by glyphosate drift on cotton and soybean, and found there was a good identification effect of the damage area using vegetation indexes calculated from the multispectral image. This proved that remote sensing technology is feasible for estimating spray effect in the large field. Until now, this technique has been widely applied in developed countries, which is determined by the characteristics of agricultural production, national government support, and related technology[8]. However, low-altitude agricultural aerial remote sensing and operation assessment of aerial spraying are still restricted by China’s national policy and planting pattern[9]. Therefore, satellite-based remote sensing has been a good choice to detect the presence of disease and insect pests and evaluate the efficacy of pesticide applications on a large scale. Zhang et al.[20]and Yuan et al.[21]utilized moderate- and high-resolution satellite images to detect crop diseases, their results can provide a previous basis for the assessment of the efficacy of aerially-applying pesticides. Meanwhile, along with the environmental resource satellites, HJ-1a/1b (2008), ZY-02c (2011), ZY-3 (2011) and GF-1 (2013), fast-boat No.1 (2013)/No.2 (2014) etc., there are more than ten satellites that have been launched. All these satellites provide enough data resources of spatial resolution of meter-level (1.2～2.36 m), revisit period of 3～5 days, and multi/hyperspectral images[22], which can be used for large-scale detection of plant disease and insect pests and the assessment of the efficient of aerial spraying applications.

The objective of this study is to evaluate aerial spraying effect of a low-altitude flight using satellite remote sensing technology, in order to provide scientific guidance for large-scale aerial spraying.

1 Materials and methods

1.1 Study Site

The study site (133°09′28.0″E, 47°31′42.5″N) was located at Qianjin farm, Jiamusi city, Heilongjiang province, China (Fig.1). It is within an important rice crop production area in China. The site has the climate conditions of long winter, short summer, 130 frost-free days, 2 521 ℃ of annual average effective accumulated temperature , and 510 mm of annual average rainfall.

Fig.1 Location map of the study site

The experimental field was 400 m long (east-west) and 180 m wide (north-south). The rice variety Longjing 26, susceptible to rice blast caused by the fungusPyriculariagrisea, was used in the experiments. Rice blast causes symptoms or damage on leaves and panicles. Leaf blast usually develops in the middle of July at the tested site. Under favorable environmental conditions, the disease spreads quickly in the field and cause severe damage to rice production. On July 6, 2013, a mix of fungicide and plant growth regulator was sprayed at the tillering stage using an airplane M-18B at 4 m flight height.

1.2 Satellite imagery and data processing

1.2.1 Satellite imagery

The search for worldwide satellite data was conducted and found only two kinds of moderate-high resolution satellites that acquired the images of the Qianjin farm from June to July 2013. One was the TM image of Landsat-8, July 6, 2013; the other was a SPOT-5 image of July 16, 2013. Table 1 and Fig.2 show the basic information about remote sensing satellites.

Table 1 Basic information of remote sensing satellites

The essential processes of satellite images, radiation calibration, atmospheric correction and geometry correction were conducted in order to effectively extract spectral information. The obtained spectral information was then utilized to calculate vegetation indices (VI) in light of bands combination. Since the Landsat-8 TM image contained the panchromatic and multi-spectral bands, and the panchromatic band had a relatively high spatial resolution (15 m), data fusion was done in order to improve the spatial resolution of the study site (Fig.3) by Gramm-Schmidt Spectral Sharpening[23].

Fig.2 Study site in the SPOT-5 (a) and Landsat-8 TM (b) images

Fig.3 Original and fused study site images of Landsat-8 TM

1.2.2 Vegetation indices

Remote sensing technology has provided an effective means for continuous surface observation. With this technology, the spectral characteristics can provide subtle information on crop physiological status at a point of time[24]; and spectral features of the temporal change can reflect the changes in crop physiology within a certain period of time[3]. Considering the visible light, green (G), red (R), and near infrared (Nir) are the most important bands with satellite Landsat-8 and SPOT-5. In this study, vegetation indices, NDVI, SAVI, TVI, MSAVI, and MTVI were calculated based on the bands combination, and then these remote sensing characteristics were used to evaluate crop vegetation growth status[25-26]. The calculation forms of the VIs and the basis of selected bands are shown in Table 2. In addition, a change in the information of spectral features at two points of time was also calculated using normalized formula (T2-T1)/ (T2+T1).

Table 2 The basis of selected bands and VIs

1.3 Instruments and field experiments

1.3.1 Aircraft and sensors

(1) The aircraft M-18B (Fig.4) was used to spray a mix of fungicide and plant growth regulator in the field. Its performance parameters are listed in Table 3.

Fig.4 The M-18B aircraft

Table 3 Performance parameters of the M-18B aircraft

AirplanetypeSprayswath/mFlightheight/mFlightspeed/(m·s-1)NozzleLoadcapacity/TM-18B453～7m50Atomizer2.5～3

(2) The placed platform with water sensitivity paper (WSP) and, droplets deposition analyzer (DDA) were used to measure the actual ground spray effect in the field. As seen in the Fig.5(a) and (b).

Fig.5 The placed platform with WSP (a) and DDA (b)

(3) A GPS sensor was used to record the sampling positions in the field.

1.3.2 Layout of ground sampling points

The layout of ground sampling points was shown in Fig.6. Ten sampling points with 5 m intervals, 0, -5, -10, -15, -20, -25, -30, -35, -40, and -45 m, were arranged along with the spray swath to measure droplets deposition. Five measurement points were positioned at 5, 10, 20, 50, and 100 m in the direction of drift, which was used to measure to reflect droplets drift characteristics. The 0 m position represented the beginning point of drift direction, the -45 m position as the border of upwind direction, and the 100-m position as the farthest drift point. In addition, three sampling lines with 35.5 or 38.7 m intervals were also arranged in the experiment to evaluate the spraying effects using moderate-resolution satellite imagery.

Fig.6 Layout of ground sampling points

2 Results and analysis

Due to the lack of significant development of leaf blast on rice at the time of aerial spraying, a plant growth regulator was added into the fungicide and applied to the tested field to promote the growth of rice plants and enhance host resistance against the disease. At 10 days after aerial spraying, changes in the biophysical features of rice plants with time were extracted from satellite imagery, and then combined with droplets deposition data at different ground sampling points. These data were used to analyze the effects of aerial spraying.

2.1 Spray effect analysis with ground droplets deposition

Tables 4 and 5 listed the relationships of spectral features of single phase and temporal change of vegetation growth after spraying to droplets deposition data, respectively. Ground droplets deposition consisted of droplets deposition points density (DDPD, points·cm-2) and droplets deposition volume density (DDVD, μL·cm-2).

Table 4 The relationship between spectral features

Note: *This reflects a significant correlation between variables (p-value<0.05)

Table 5 The relationship between spectral features of

DDPD: droplets deposition points density; and DDVD (droplets deposition volume density); *This reflects a significant correlation between variables (p-value<0.05)

As shown in Tables 4 and 5, spectral characteristics of single phase (July 16), NDVI, SAVI, TVI, MSAVI, and MTVI had a significant correlation with droplets deposition data. The correlation coefficients of NDVI were 0.315 (p-value=0.035) and 0.312 (p-value=0.038), respectively, which were obviously greater than others. However, there were only lower correlation between single bands (G, R, Nir) and droplets deposition data. For spectral features of temporal change, there were also greater correlation coefficientsRof 0.309 (p-value=0.041) and 0.312 (p-value=0.038) between NDVI and DDPD, MSAVI and DDPD, respectively. The correlations were not consistent in VIs for single phase and temporal change.

2.2 Trend analysis with flight path and droplets drift direction

Fig.7(a) and (b) are spatial interpolation maps of droplets deposition points density (points·cm-2) and droplet deposition volume density (μL·cm-2), respectively. Fig.7(c) and (d) are spatial interpolation maps of single phase spectral feature (NDVI) and temporal phase change feature (MSAVI).

As shown in Fig.7(a) and (b), the greatest biological activity of rice plants appeared within the spray swath (-45～0 m). Their biological activity declined gradually with the direction of droplets drift. The normal growth of plants appeared in the distance of 50 to 100 m where there was a little product-spraying effect. There was a similar trend in the changes in the field for DDPD and DDVD. Moreover, the spatial interpolation map of NDVIs [Fig.7(c)] clearly showed more vigorous growth of rice plants located in the north area than in the southern area, which was similar to ground measurements of droplets deposition. This appeared to be the results of the influence of aerial spraying. The spatial interpolation map of MSAVIs [Fig.7(d)] also showed a similar trend, but the spatial change did not present a strong consistency with NDVIs. Through the data analysis of single phase feature [Fig.7(c)] and the phase change feature [Fig.7(d)], VIs extracted from satellite imagery effectively captured the changes in the growth of rice crop after the aerial spraying, thus VIs can be used to evaluate the changes in plant growth.

Fig.7 The spatial interpolation maps of droplets deposition data and spectral features

(a): Spatial interpolation map of DDPD; (b): Spatial interpolation map of DDVD; (c): Spatial interpolation map of NDVIs; (d): Spatial interpolation map of MSAVIs

3 Discussion

Food security and safety have always been a top priority in China for the past 50 years. It is a tremendous challenge for China to ensure adequate and safe food supply for its almost 1.4 billion people. The development of precision pesticide aerial application technology can be optimal means toward this goal. With the introduction of aviation aircrafts and large-scale application of aerial spraying, we have been provided more opportunities to promote the development and utilization of precision agricultural production technologies, including precision pesticide aerial application technology, to produce adequate and safe food. Owing to the restrictions of aviation flight policy and less emphasis on the research of aerial spraying technologies, fewer scholars and researchers in China have investigated the use of an airplane or satellite platform to detect disease and insect pests and evaluate the effects of aerial pesticide spraying on control of pests and reduced use of pesticides. The results of the current study found that change patterns detected in the performance of rice crop were consistently similar in the spatial interpolation maps of NDVIs and MSAVIs and the maps of ground droplets deposition data. It was clearly shown that the growth of rice plants located in the north area was better than that of rice plants in the southern area after 10 days of aerial spraying. Although the correlation coefficients were the lower (only 0.315 or less) between NDVI or MSAVI and DDPD, the satellite imagery still provided helpful information after the pesticide spraying, which can be utilized to assess the effects of aerial spraying in the large-scale field. Moreover, the results of our study also indicate that the vegetation indices, NDVI and MSAVI can serve a useful tool to effectively evaluate the efficacy of pesticides. These results are in agreement with the report of Huang et al.[18]with the use of the VIs to effectively assess phytotoxicity severity in the area sprayed aerially with glyphosate.

However, when compared the spatial distribution of single phase feature [Fig.7(c)] with the phase change feature [Fig.7(d)], the spatial trend does not present a strong consistency. This may be caused by the following reasons: (1) the randomness of the spraying effect on the rice crop and the growth of the crop; and (2) the spatial resolution of the 15-m, single pixels contained a variety of features that could cause differences between the results of the evaluation index. Moreover, the spatial distribution of MSAVIs was easily influenced by imagery spatial resolution; thus, further research needs to be conducted to explain the results of MSAVI. Secondly, although it was feasible to evaluate the aerial spraying effects using satellite images, there were no obvious disease characteristics to be detected due to the lack of the disease development when the measurement was made. The aerial spraying also caused no phytotoxicity damage to rice plants. Therefore, these factors could lead to the lower correlation coefficients between VIs and droplets deposition data. Thirdly, the satellite images of 15-m spatial resolution can reflect the aerial spraying effect in the large-scale field, however, the assessment accuracy needs to be improved by using high-resolution satellite image (1～3 m). With the emergence of multi-source satellite data in China, including high resolution and a short revisit period, it is very useful to analyze the characteristics of vegetation stress on a large scale, especially for the assessment of aerial spraying in the field. This will tremendously be helpful to the development and improvement of China’s agricultural aviation applications.

4 Conclusions

In this study, an airplane M-18B at the 4-m flight height was used to spray a mix of agrichemicals that can control rice leaf blast and promote the vigor of plants in the field. Remote characteristics were extracted through calculating the VIs, and then combined ground measurement data of agrichemical concentration, to explain the reliability of evaluating the aerial spraying effect using satellite remote sensing imagery. The results of this study showed that the correlation coefficients were 0.315 and 0.312 between NDVI and DDPD, MSAVI and DDPD, respectively. The spatial interpolation maps of NDVIs and MSAVIs had a good consistency with those maps of ground droplets deposition data. It clearly shows that rice plants in the spray swath had the greatest growth vigor. Therefore, it is feasible to evaluate the aerial spraying effects using satellite images.

[1] Acharya K, Dutta A K, Pradhan P. Australian Journal of Crop Science, 2011, 5： 1064.

[2] Mirik M, Michels G J, Jr Kassymzhanova-Mirik S, et al. Computers and Electronics in Agriculture, 2007, 57(2): 123.

[3] Rakhesh Devadas. University of New England, Armidale, New South Wales, Australia, 2009.

[4] Zhao Jinling, Huang Linsheng, Huang Wenjiang, et al. Europe Journal of Plant Pathology, 2014, 139: 407.

[5] Lan Yubin, Thomson Steven J, Huang Yanbo, et al. Computers and Electronics in Agriculture, 2010, 74: 34.

[6] Lan Yubin, Huang Yanbo, Martin Daniel E, et al. Applied Engineering in Agriculture, 2009, 25(4): 607.

[7] Zhou Zhiyan, Zang Ying, Luo Xiwen, et al. Transactions of the Chinese Society of Agricultural Engineering, 2013, 29(24): 1.

[8] Xue Xinyu, Lan Yubin. Transactions of the Chinese Society for Agricultural Machinery, 2013，44(5): 194.

[9] Zhang Dongyan, Lan Yubin, Chen Liping, et al. Transactions of the Chinese Society of Agricultural Machinery, 2014, 45(10): 53.

[10] Mirik M, Michels G J, Jr Kassymzhanova-Mirik S, et al. Computers and Electronics in Agriculture, 2006, 5： 86.

[11] Huang Wenjiang, Lamb David W, Niu Zheng, et al. Precision Agriculture, 2007, 8: 187.

[12] Yang Chwen-ming. Precision Agriculture, 2009, 11(1): 61.

[13] Larsolle A, Muhammed H H. Precision Agriculture, 2007, 8: 37.

[14] Huang Linsheng, Zhao Jinling, Zhang Dongyan, et al. International Journal of Agriculture and Biology, 2012, 14: 697.

[15] Zhang Minghua, Qin Zhihao, Liu Xue. International Journal of Applied Earth Observation and Geoinformation, 2003, 4(4): 295.

[16] Qin Zhihao, Zhang Minghua. International Journal of Applied Earth Observation and Geoinformation, 2005, 7: 115.

[17] Huang Y, Thomson S J, Ortiz B V, et al. Biosystems Engineering, 2010, 107(3): 212.

[18] Ortiz B V, Thomson S J, Huang Y, et al. Computers and Electronics in Agriculture, 2011, 77(2): 204.

[19] Huang Y, Thomson S J, Hoffman W C, et al. International Journal of Agricultural and Biological Engineering, 2013, 6(3): 1.

[20] Zhang H, Lan Y, Lacey R, et al. International Journal of Agricultural and Biological Engineering, 2009, 2: 1.

[21] Zhang Jingcheng, Huang Wenjiang, Li Jiangyuan, et al. Precision Agriculture, 2010，12(5): 716.

[22] Yuan Lin, Zhang Jingcheng, Shi Yeyin, et al. Remote Sensing, 2014, 6： 3611.

[23] http://www.cresda.com/.

[24] Zhang Jingcheng, Pu Ruiliang, Yuan Lin, et al. PLoS ONE, 2014, 9(4): e93107.

[25] Yuan Lin, Zhang Jingcheng, Zhao Jinling, et al. Optik, 2013, 124: 4734.

[26] Tyler J Nigon, David J Mulla, Carl J Rosen, et al. Computers and Electronics in Agriculture, 2015, 112: 36.

[27] Maurer. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XL-1/W1, ISPRS Hannover Workshop, 21- 24 May 2013, Hannover, Germany.

S127

A

基于中分辨衛(wèi)星影像的農(nóng)用航空噴藥效果評估

張東彥1, 2, 3，蘭玉彬4, 5，王 秀1, 3，周新根5，陳立平1, 3*，李 斌1, 3，馬 偉1, 3

1. 北京農(nóng)業(yè)信息技術(shù)研究中心，北京 100097 2. 安徽省農(nóng)業(yè)生態(tài)大數(shù)據(jù)工程實(shí)驗(yàn)室，安徽大學(xué)，安徽 合肥 230601 3. 農(nóng)業(yè)部農(nóng)業(yè)信息技術(shù)重點(diǎn)實(shí)驗(yàn)室，北京 100097 4. 華南農(nóng)業(yè)大學(xué)工學(xué)院，廣東 廣州 510642 5. Texas A&M AgriLife Research and Extension Center, Beaumont, TX 77712, USA

遙感技術(shù)能被用于大尺度作物化學(xué)噴藥效果檢測，這為精準(zhǔn)農(nóng)業(yè)航空施藥發(fā)展提供了重要的技術(shù)支撐。利用M-18B農(nóng)用飛機(jī)在4米的飛行高度噴施化學(xué)農(nóng)藥混合劑(殺菌劑和植物生長調(diào)節(jié)劑)，去控制水稻爆發(fā)性疾病--葉片紋枯病和促進(jìn)水稻植株的生長。施藥一周后，噴藥區(qū)的衛(wèi)星影像被獲取并計(jì)算植被指數(shù)，同時采集了地面化學(xué)農(nóng)藥的藥液沉積量。分析了藥液霧滴沉積量和植被指數(shù)的關(guān)系，結(jié)果顯示，單相光譜特征(NDVI)和液滴沉積點(diǎn)密度(DDPD點(diǎn)·cm-2) 的相關(guān)系數(shù)是0.315，p-value為0.035； 時間變化特征 (MSAVI)和液滴沉積體積密度(DDVD μL·cm-2)之間的相關(guān)系數(shù)是0.312，p-value為0.038。另外，水稻生長活力最旺盛的范圍都出現(xiàn)在噴灑區(qū)域內(nèi)，植株活力隨著藥液漂移距離的增加逐步減少。同時，相同的變化趨勢也出現(xiàn)在霧滴沉積量與光譜特征的空間變化插值圖中。由此得知，從衛(wèi)星圖像中計(jì)算的植被指數(shù)NDVI和MSAVI，可以用來評估大尺度農(nóng)田的農(nóng)用航空藥液噴灑效果。

衛(wèi)星影像； 植被指數(shù)； 航空噴藥； 霧滴沉積； 漂移

2015-06-20，

2015-10-25)

Foundation item： National Key Technology Research and Development Program of the Ministry of Science and Technology of China (2014BAD06B01-17), National Natural Science Foundation of China (41301471), and International Postdoctoral Exchange Fellowship Program (20130043)

10.3964/j.issn.1000-0593(2016)06-1971-07

Received： 2015-06-20; accepted： 2015-10-25

Biography： ZHANG Dong-yan, (1982—), Associate Professor in Beijing Research Center of Information Technology in Agriculture e-mail: zhangdy@nercita.org.cn *Corresponding author e-mail: chenlp@nercita.org.cn