Parameters optimization of crop protection UAS based on the first industry standard of China

The effective swath width (ESW) and the droplet penetration rate (DPR) directly affect the spraying quality, the spraying efficiency and the control effect of pests and diseases during the crop protection unmanned aircraft system (CPUAS) application. However, the ESW and DPR are not constant with the changes of the flight speed (FS) and the flight height (FH). In order to investigate the ESW and DPR of the CPUAS P20, four levels of FS (3 m/s, 4 m/s, 5 m/s and 6 m/s) and three levels of FH (1.5 m, 2.0 m and 2.5 m) experiments were carried out according to the first industry standard of China for the CPUAS in the wheat field. The results demonstrated that the ESWs were negatively correlated with the FS and the FH. Most of the ESWs were over 2 m in the 12 treatments, in which the maximum one was 3.25 m (3 m/s, 1.5 m). The DPRs were negatively correlated with the FH under the same FS, the average value of the DPRs was 48.37%, in which the maximum one was 78.34% (4 m/s, 1.5 m) and the minimum one was 25.5% (6.0 m/s, 2.5 m). The statistical analyses showed that the FS had significant impacts on the ESWs (0.01<p-value<0.05) while there were no significant differences among different FH treatments (p-value >0.05). The impacts of both FS and FH on the DPRs were extremely significant (p-value<0.01), and the interactive impacts were significant (0.01<p-value<0.05). Therefore, it is concluded that reducing the FS could increase the ESWs, and reducing the FH could increase the DRPs at the same FS. In conclusion, the maximum spraying efficiency of P20 was 4.342 hm/h with 6 m/s FS and 1.5 m FH in case of satisfying the requirement of DPRs. This study provided scientific references for guiding the CPUAS spraying.


Introduction 
The CPUAS has the advantages of superior mobility, wide adaptability and high efficiency without the restrictions of the crop types or growth periods, especially suits for the paddy fields and the mountainous areas [1][2][3] . Benefit from the development of the modern electronic communication, the control technologies [4][5][6] and the national policy [7] , the CPUAS has developed rapidly in China in recent years [7,8] , not only the technical level but also the application area are already the first around the world [9,10] .
The ESW and DPR are two important indicators in the CPUAS application, directly affect the spraying quality, the spraying efficiency and the control effect. In the past few years, some studies have been undertaken to investigate and improve the CPUAS application. Huang et al. [11] used a simulation-based approach to study the aerially applied crop protection drift, and set a near-optimal offset of the flight trajectory to reduce the drift and increase the near deposition. Qiu et al. [12] arranged a test by two factors three levels to find out the factors and degree of influence affecting the CPUAS spraying deposition. The results showed that the FH, FS and the interaction between the two factors all affected the deposition and uniformity. Xue et al. [13] carried out trails to measure the aerial spray deposition and drift in the paddy field, the results showed with the assistant of the downwash of the rotors, the under layer deposition could take up of 92.8% of the upper one and 90% drift droplets were located within a range of 8 m. Al-Heidary et al. [14] discussed the droplet and spraying to investigate the influence on field sprayer drift, including droplet size, droplet velocity, droplet evaporation, droplet diameter distribution and spraying height, spraying top angle, which provide a reference to the aerial spraying. Zhang et al. [15] proposed a new approach based on computational fluid dynamics (CFD) to investigate the aerial spraying drift when the FS was 3 m/s, FHs were 5 m, 6 m and 7 m. The study proved that the approach was feasible and suggested that buffer zone should be reserved considering the downwind drift. Hou et al. [16] designed the parameter controller of the CPUAS for improving the droplet density on citrus trees, and the experimental results showed that the FS presented the most significant effect, the established model predicted an optimal spraying height of 1.27 m and maximum droplet density of 35.39 droplets/cm 2 . Wang et al. [17] used three measurement methods(sampling frame, Petri dishes, rotary impactors) to test the sediment and aerial spraying drift under different FH, FS and wind speed, the analysis results could provide data support for the CPUAS application under different conditions.
These studies indicated that appropriate operation parameters could improve the pesticide droplet deposition and the penetration, thereby achieved a fine control effect of pests and diseases. However, the ESW and the DPR vary with the FH and FS, while in actual the application of the CPUAS the ESW usually was determined by the manufacturer's suggestions or the experiences of the operators [18] , ignoring the influences of parameters. Therefore, studying the effects of different heights and speeds on the ESWs and DPRs can guide the CPUAS to spray pesticides better, reducing the pesticide usage and improving the pesticide utilization efficiency.
Due to the lack of standards special for the CPUAS, some results and conclusions were concluded and tested according to some relevant standards as ASAE S3451.3, MH/T1002 and MH/T1040 [18][19][20][21] , which had a certain reference significance for the CPUAS application.
The first agricultural industry standard Technical Specification of Quality Evaluation for Crop Protection UAS was promulgated on June 1st, 2018 in China [22] , which is the most authoritative worldwide until now. In this article, four levels of FS and three levels of FH experiments were carried out to test the ESWs and DPRs of the CPUAS P20 (Guangzhou XAG Co., Ltd, China) in the wheat field. The aims were to understand the spraying quality under different parameter conditions and determine the optimal parameter combinations for the CPUAS application, providing a decision basis for controlling pests and diseases in different parts and growth periods of the crops.

Test site and the environment
The experiments were carried out in the wheat field of Sihong agricultural demonstration base (33.3636°N, 118.2599°E) in Jiangsu Province, China, on April 16, 2019. The wheat variety is Qianmai 33, and was sowed in the field (60 m×120 m) with seed rate 225 kg/hm 2 . The wheat was on the growth period of heading stage with an average 65 cm height. The wind speed ranged among 0.5 to 1.5 m/s, the average temperature was about 14°C and the average relative humidity was 65%, which met the test condition requirements of the CPUAS swath width.

Experimental equipment and materials
The main parameters of P20 are shown in Table 1. The portable weather meter Kestrel 4500 (the Nielsen-Kellerman company, US.) was used to record the wind speed, the temperature and the relative humidity. The Rhodamine B solution with 0.05% mass fraction was sprayed in the experiments. Droplets were collected by the water sensitive paper (WSP). The DepositScan (DS) software [19,23] was used for analyzing droplet deposition density and coverage.

Sampling arrangements
The experiments were designed according to the standard Technical Specification of Quality Evaluation for Crop Protection UAS (NY/T3213-2018). The whole experimental area was divided into flight acceleration area, sampling area and stop spraying area. The flight acceleration area and the stop spraying area were both 50 m long in order to ensure the P20 could accelerate to a predetermined speed and stop timely. Three repetitions in the sample area with a 10 m interval were arranged along the vertical direction of the flight route. A total of 15 sampling points were arranged on each line. The sampling points labeled S1 to S15 were symmetrically distributed from left to right on both sides of the flight route. To the experiences, the ESWs of P20 would be not less than 1.5 m, so the interval distances among S1 to S6 were set as 0.20 m, S6 to S7 were set as 0.25 m, and S7 to S8 were set as 0.50 m for improving the experiment efficiency (the right side sampling points arranged same as the left side ones). The sampling layout was set as Figure 1 showed.

Figure 1 Layout of droplet sampling cards (top view)
The WSPs were fixed horizontally on the upper and lower layers at each sampling point without overlapping as Figure 2 showed, 15 cm vertical distance both to the top canopy of the wheat and the ground. The swath widths were measured by the collecting droplets from the upper layer WSPs and the penetration rates were calculated by the collecting droplets from both the upper and layer WSPs. When each spraying test was finished, the WSPs were collected and put into the self-sealing bags and brought back to the laboratory for analysis. Figure 2 Sketches of WSPs fixed for collecting droplets

Experimental treatments
Considering the actual applications, the FS was set four levels of 3 m/s, 4 m/s, 5 m/s and 6 m/s, the FH was set three levels of 1.5 m, 2.0 m and 2.5 m. The CPUAS P20 flew from the acceleration area to the stop spraying area perpendicularly along the center line of the sampling area with autonomous mode [4] . Twelve treatments (A1 to A3, B1 to B3, C1 to C3, D1 to D3) with different parameter combinations were carried out. All the treatment parameters are showed in Table 2. The flow volume per unit is constant recommended in Table 1 as the spraying rate matches with the FS.

Statistical analysis
The WSPs were scanned to JPG images as shown in Figure 3. The droplet coverage density on each WSP was calculated by the DS, the deposition uniformity and penetration rate were analyzed further.
a. Upper WSPs b. Lower WSPs Figure 3 WSPs collecting the droplets The first sampling point of droplet quantity not less than 15 droplets per square centimeters (cm 2 ) was judged as the boundary of the ESW each line. If the quantities of the two adjacent sampling points jump changed, such as the current point's droplet quantity was far more than 15 droplets/cm 2 , while the next one was less than 15 droplets/cm 2 , interpolation processing was carried out to calculate the effective spray amplitude boundary.
The deposition uniformity was evaluated with the coefficient variation (CV) of coverage rates [24] on the WSPs calculated from the DS within the ESW. The CV calculation equation is as follows.
where, S is the standard deviation of the droplet coverage rates on the WSPs each repetition; X i is the coverage rate of each WSP in the repetition, and X is the average of X i .
The droplet penetrability into the canopies was expressed by the DPR calculated by the following equation.
where, x i , X i are the coverage of the upper layer WSP and the lower WSP each sampling point within the ESW range;  is the penetration rate, respectively.

Test result data
The maximum and average swath widths were calculated which were the maximum and average ones of the three repetitions in each treatment, respectively. The CV of coverage rates and the penetration rate were calculated according to Equations (1)-(3). In this article, the average swath width was used as the ESW in order to ensure accuracy. The test result data were shown in Table 3. From the Table 3, it could be seen that the ESWs are among 1.79 m (D3) to 2.96 m (A1) which did not reach to 3 m, and the maximum swath width of each treatment was more than 2 m, of which the maximum one was 3.25 m (A1). The CVs were all exceeding 50% of which the minimum one was 52.91% (B2) and the maximum one was 96.77% (C2), which meant the deposition uniformity fluctuated greatly within the ESWs. The DPRs of the twelve treatments had no obvious correlation with the changes of the FSs, of which the maximum value was 78.34% (B1) and the minimum one was 25.50% (D3). Comparing Figure 4 with Figure 5, it could be seen that the ESWs with the same height varied much more than the ones at the same speed.

Effects of FS and FH on ESWs
By the two-way analysis of variance (ANOVA) results, the ESWs were significantly affected by the FS, while the FH effect and the interactive effect between FS and FH were non-significant (   (4), the F value was 69.869, the significance F value < 0.01 in the regression analysis indicating that the regression function (4) was extremely significant (**).

PDR analyses 3.3.1 The DPR changes
The average DPR was 48.37%, which indicated that the coverage of the lower layer was about to half of the upper one, and the maximum one was 78.34% (B1). The DPRs decreased with the FH increased under the same FS showed in Figure 6, while change trends were not consistent with the FS under the same FH in Figure 7. It could be concluded out the negative correlations between DPR and FH at the same FS. From this perspective, the average DPR was 60.99% at 1.5 m FH, 46.04% at 2.0 m, 38.12% at 2.5 m, indicating that FH was the main factor affecting penetration.

The Effects of FS and FH on DPRs
The two-way ANOVA results (Table 6) showed that both FS and FH significantly affected the DPRs (p-value < 0.01) extremely, and the influence of FH on DPR was greater than that of FS. The interactive effect between FS and FH was significant (0.01<p-value <0.05).

Droplet deposition uniformity
The droplet deposition uniformity was as another indicator to evaluate the spraying quality. In this article, the CV was used to investigate the droplet deposition uniformity by Equation However, the p-values of ANOVA analysis were larger than 0.05 taking the FS and FH as the independent variables, the CV as the dependent variables, which indicated that the CV variations were not affected by the changes of FSs or FHs. Figure 8 showed the droplet deposition uniformity by CVs. Figure 8 Deposition uniformity of each treatment by CVs

Spraying efficiency
The CPUAS spraying efficiency is accessed by the effectively sprayed area per hour calculating based on the ESW and the FS as Equation (6).
where, S is the effectively sprayed area per hour, hm 2 /h; v is the FS, m/s. In Table 7 all the spraying efficiencies are calculated. The maximum efficiency is 4.342 hm 2 /h appears in S 6 . When the penetrability (no larger than 53.64%) meets the requirement, the parameter combination of D1 (6 m/s, 1.5 m) should be preferred to ensure the maximum spraying efficiency.

Discussion
Different requirements are required of the deposition and penetration for different growth periods and different diseases in chemical crop protection applications because of the crop density and the disease occurrence location. Take the wheat as an example, stripe rust, powdery mildew and scab (or head blight) are the major diseases that are harmful to the wheat [25][26][27] . For the occurring time, the stripe rust would occur from the wheat tillering stage to the filling stage if it occurs seriously [31][32][33][34] , the powdery mildew mainly occurs between the heading stage and the milky stage of wheat [32][33][34] , and the scab mainly occurs from the heading stage to the filling stage [35] . For the disease occurrence locations, the stripe rust occurs in the middle and lower parts of the wheat, the powdery mildew occurs and develops from the bottom to up layer, and the scab is concentrated on the upper layer (spikelets) of wheat. Therefore, the selection of suitable operating parameters can better ensure the control effect. When controlling the scab, priority is given to increase the ESW, and the FS is appropriately reduced. When controlling the powdery mildew, the penetration of droplets is considered as an effective consideration, the FH should be reduced possibly. When controlling the stripe rust, the ESW and the penetration should be both considered for the parameter optimization combined with the growth period of wheat.

Conclusions
The ESW and DPR are two most concerned indicators to evaluate the CPUAS spraying, which represent the droplet deposition on the top canopy and inside the canopy. Achieving the accurate ESW and DPR could provide scientific references for guiding the CPUAS spraying, consequently increase the CPUAS spraying quality and reduce the pesticide waste. The FS and FH are two most important parameters in the CPUAS practical application by manual control. The parameter combination experiments suggested that the ESW and DPR of the CPUAS could be optimized by changing the FS and FH. For the CPUAS P20, the maximum ESW was 2.96 m (3 m/s, 1.5 m), the maximum DPR was 78.34% (4 m/s, 1.5 m). To achieve the maximum spraying efficiency, FS 6 m/s and FH 1.5 m could be used. In this study, the relationships during the FH, FS, ESW, DPR and the droplet deposition uniformity were studied.
(1) The ESW is negatively correlated with the FS and FH as a whole. The ESW variations at the same FHs with different FSs are greater than those of at the FSs with different FHs. In this study, the maximum ESW variation amplitude is 0.95 m (1.5 m FH) while the maximum one is only 0.26 m (3 m/s FS). The FS has significant effect on the ESW. The reason may be that when the FS is slow, the droplet initial horizontal velocity is slow, and the droplet falling is more susceptible to the rotor wind field, thus can effectively deposit on the crop canopy. However, when the FS is fast, the droplet with fast initial horizontal velocity may escape from the range of the wind field and drift away. The FH has no significant effect on the ESW. The ESW is larger when the FH is lower at the same FS. The reason may be that the droplet transport distance from release to deposition is shorter for better deposition with less evaporation. Therefore, it can be concluded that appropriately reducing the FS could effectively increase the ESW.
(2) The DPR is negatively correlated with the FH at the same FS. The DPR change trend is not monotonous at the same FH. Both FH and FS have significant effects on the DPR, and the effect of FH is greater. The reason may be that the wind field disturbs the crop making it easier for the droplet to deposit on the lower part of the exposed crop. Another reason may be that when the FH is lower the whole transport distance is shorter, the DPR is increased with the assistant of downwash of the wind field. Therefore, it can be concluded that reducing the FH could improve the droplet penetration.
(3) In this study the average CVs of droplet deposition in the 12 treatments is 66.65%, and the variation trend of the variation coefficient was not regular. Also, the ANOVA results showed that the uniformity of deposition was not affected by the FS or FH. The reason may be that the spraying uniformity was affected not only by FH and FS, but also by environmental wind speed, spraying volume and droplet particle size, etc, which needed further studies.
(4) The ESW is one of the most important indicators in this study, and it was tested and determined according the first industry standard of China special for the CPUAS. Some other ESW evaluation methodologies had been adopted previously, such as "minimum acceptable deposition coefficient variation values determinate method", "50% effective application rate determinate method", "identifying the largest range of coverage rates greater than the average coverage rate method ", and the results were different by different methods. In addition, the experimental results and conclusions could be as a reference basis, the influence of weather factors on the results should also be considered in actual applications.