Assessment of Three Commercial Over-the-Row Sprayer Technologies in Eastern Washington Vineyards

Washington winegrape growers are rapidly adopting vineyard management technologies such as mechanical pruners and harvesters, but they have been slower to adopt new chemical application technologies. This study generated technical information about commercial over-the-row sprayers and their deposition and drift to allow growers to select and optimize sprayers for different vineyard systems and winegrape canopies. Three commercial sprayer technologies (multi-fan heads, pneumatic, and electrostatic) were evaluated for canopy deposition and drift in the 2016 and 2017 production seasons. Data were collected in Vitis vinifera Chardonnay and Riesling vineyards at two application timings, early season and midseason, to determine sprayer deposition patterns in opposed and unopposed applications and in-field aerial and ground drifts. All sprayer technologies showed consistent in-canopy deposition and drift patterns at both application timings. Regardless of sprayer technology, the most deposition was in the upper canopy rather than in the fruiting zone. Similarly, the most aerial and ground drift occurred in the row closest to the sprayed row, indicating that drift is relatively low with all three evaluated sprayer technologies.

Thus, it is important to conduct regional studies to evaluate sprayer performance based on local 97 canopy architecture and environmental differences. This study assessed three commercially 98 available over-the-row, air-assisted sprayers to gather data that can be used for sprayer 99 optimization in wine grape canopies.

101
General experimental design. 102 For all sprayers (Figure 1), canopy deposition, aerial drift, and ground drift were evaluated at two 103 timings, early season (BBCH 55 or 65), and mid-season (BBCH 75, 77, or 79;Lorenz et al. 1994). 104 Specific canopy and drift assessment methods are described below in "Spray deposition and drift 105 collection and processing." Sprayer evaluation comparisons are made only within a single sprayer 106 type. Thus, no comparisons are made among sprayer types.   In 2016, due to rapid early season growth at the time of application, six nozzles were used for all 148 applications dates at a spray volume rate of 702 L/ha. In 2017, the early season applications 149 occurred when canopy sizes were smaller, therefore only four nozzles arranged in a square pattern 150 were used at a spray volume rate of 468 L/ha. In the late season applications, six nozzles were used 151 with a spray volume rate of 702 L/ha ( Table 2). The sprayer calibration and optimization were 152 done with direct onsite assistance of manufacturer representatives, and done so according to their 153 regional specifications. Fan heads were set facing forward, towards the front of the sprayer, at an 154 approximately 45 o angle into the canopy, not facing directly towards the opposing fan head (Fig. 155 1A). Air velocity was recorded approximately 0.5 m from the nozzle to represent the distance from 156 the sprayer to the canopy at the canopy closest to the sprayer on the opposed row (24 km/hr) and The second sprayer was a Gregoire Speedflow Progress, distributed by Blueline Manufacturing  (Table 2). This pneumatic sprayer was 171 equipped with a Gregoire air-shear, DynaDiff diffuser nozzle. The sprayer calibration and 172 optimization were done with direct onsite assistance of manufacturer representatives, and done so 173 according to their regional specifications. The fabric tubes were set facing backwards, away from 174 the front of the sprayer, at an approximately 20 to 30 o angle into the canopy, not facing directly 175 towards the opposing fabric tubes and nozzles (Fig. 1B). Air velocity was recorded approximately 176 0.5 m from the nozzle to represent the distance from the sprayer to the closest canopy r on the 177 opposed row (26 km/hr) and the unopposed row (23 km/hr). Though individual sprayers and rate controllers differed between years changed, application rate, engine speed, and operating pressure 179 were all maintained at a consistent rate.  vinifera 'White Riesling' on a 1.5 m by 2.4 m, vine by row spacing. The canopy was trained to a 186 strict VSP, and the vineyard's irrigation and pest management programs followed grower standard 187 practices for the site. Vine-row-volume measurements were also taken at site 2, as previously 188 described.

190
Site 2 sprayer. The On Target sprayer, manufactured by On Target Spray Systems Inc. (20-32 191 horsepower requirement; Mt. Angel OR, USA), was a two row, over-the-row electrostatic sprayer 192 (Fig. 1C). This sprayer was operated by a Kubota ® M8540 tractor (86 horsepower;Gainesville,193 GA, USA). The sprayer was equipped with eight PVC tubes, each with five electrostatic pneumatic 194 nozzles. All nozzles were used the entire season for application at a rate of 198.3 L/ha (Table 2).

195
Prior to field tests, the nozzles were checked with a voltmeter (ASIMT33D-CA, AstroAI,Brea,196 CA, USA) to ensure 1000 V of electricity was present at each nozzle as is required for electrostatic 197 charge of the droplets. This sprayer did not utilize a rate controller, and was calibrated and 198 optimized with direct onsite assistance of manufacturer representatives, and done so according to 199 their regional specifications PVC tubes, and thus nozzles, were oriented at 90° to the vine row, 200 directly opposite the opposing tube and nozzles (Fig. 1C). Air velocity was not recorded for this 201 sprayer.

203
Travel speed calculations. For all sprayers, tractor speed was calculated before each spray trial 204 application. The tractor was operated at the same settings during a tracer application with the power 205 take-off (PTO) engaged and at the rotations per minute (rpm) of the intended application with a 206 full sprayer tank. The specified gear and tractor rpm for each application was recorded (Table 2) 207 and adjusted to account for changes throughout the growing season, including spray volume 208 delivered and canopy size. The tractor speed was timed over a set-length course (91.44 m) and 209 calculated to kilometer per hour (km/hr).

211
Weather and canopy measurements. 212 Environmental parameters including wind speed (km/hr), wind direction (°), relative humidity 213 (RH, %), and air temperature ( o C) were collected continuously during spray applications. During  (Tables 3, 4, and 5) data at 0.02 Hz. The weather station was mounted 218 approximately 3 m above ground level. On days that the all-in-one ATMOS 41 weather station did 219 not record weather data, the corresponding wind direction data was pulled from Washington State 220 University's AgWeatherNet (weather.wsu.edu) weather station network since the handheld 221 anemometer does not collect that type of data. At site 1 the AgWeatherNet "Paterson West" station  water at a concentration of 500 mg/L was applied during field trials. Tank samples were collected 236 pre-and post-application and used to determine tracer concentration, as well as to normalize data 237 across spray dates. Canopy deposition, aerial drift, and ground drift was determined by collecting 238 spray deposition on 5 × 5 cm plastic cards (card placement described below). This method of using 239 plastic cards has been shown to be appropriate in low to moderate volume spray applications, and   Poles were constructed from schedule 40 2-cm PVC pipe and metal alligator clips (5.1 cm by 1.1 246 cm) ( Fig. 2A and B) to attach the afore mentioned plastic collection cards. At site 1, 15 canopy poles were placed in two rows (opposed spray row and unopposed spray row) to accommodate the 248 three-row sprayers (Fig. 2E). Site 2 had 15 canopy poles in a single row since it was a two-row     Powdery mildew disease ratings. 287 Cluster disease ratings of grapevine powdery mildew, Erysiphe necator, were conducted for each 288 sprayer in both years. Disease ratings were visually estimated as percent surface area infected.

289
Ratings were collected in-field on 30 clusters per treatment replicate typically collected between 290 (25 Aug and 1 Sept each year; typically immediately pre-harvest). The level of disease present at all sites and both years were considered acceptable by the associated commercial entity. In 2016 292 at Site 1, clusters in the Quantum Mist trial had an average incidence and severity of powdery 293 mildew of 1% and 0.3%, respectively. In 2016 at Site 1, clusters sprayed by the Gregoire had an 294 average incidence and severity of powdery mildew of 3% and 0.3%, respectively. In 2017 at Site 295 1, clusters in the Quantum Mist trial had an average incidence and severity of powdery mildew of 296 80% and 16%, respectively. In 2017 at Site 1, clusters sprayed by the Gregoire had an average 297 incidence and severity of powdery mildew of 62% and 15%, respectively. In 2017 at Site 2, clusters 298 sprayed by the On Target had an average incidence and severity of powdery mildew of 70% and 299 10%, respectively.   The interaction between height above the canopy and distance from the sprayer was not significant 327 (p = 0.97 in 2016 and p = 0.06 in 2017). Height above the canopy (0.3 to 0.9 m) did not influence 328 aerial drift either (p = 0.81 in 2016 and p = 0.08 in 2017), so aerial drift data were analyzed as 329 distance (rows) from sprayer (Table 6). In 2017, ground drift data were different between the two 330 quadrants of the vineyard (p = 0.04) and were analyzed separately. The amount of ground drift did 331 not statistically differ by distance from sprayer in 2016 or the south quadrant in 2017 (Table 6), 332 but it was significant in the north quadrant in 2017. While not always statistically significant, most 333 ground drift was collected in the row closest to the sprayed row, and the least in the third row from 334 the sprayed row.

Mid-season.
In both years of the study, significantly more deposition was collected in the upper 336 canopy zones of opposed and unopposed row applications (Table 7). Total collected deposition in 337 the opposed row (42.4 ng/cm 2 ) was less than the unopposed row (81.7 ng/cm 2 ) in 2016 (p < 338 0.0001), but deposition was not different between rows in 2017 (p = 0.45; opposed 55.6 ng/cm 2 ; 339 unopposed 52.9 ng/cm 2 ). In 2017, vineyard quadrant data was pooled within vineyard collection 340 area as no significant difference was seen in canopy deposition between quadrants (opposed row, 341 p = 0.71; unopposed row, p = 0.40) or ground drift between quadrants (p = 0.56).

343
The interaction between height above canopy and distance from the sprayer were not different for  (Table 7) and in both years the row closest to the sprayed 352 row had more drift than the rows further away.  (Table 8). Total collected canopy deposition in the opposed row (86.8 ng/cm 2 ) was more than the unopposed row (64.2 ng/cm 2 ) in 2016 (p < 0.0001). Total 358 canopy spray deposition was not different between rows in 2017 (p = 0.67; opposed 50.5 ng/cm 2 ; 359 unopposed 52.5 ng/cm 2 ). In 2017, vineyard quadrant data were pooled within vineyard collection 360 area as no significant difference was seen in opposed row canopy deposition (p = 0.15), aerial drift 361 (p = 0.08), or ground drift (p = 0.06). In 2017, unopposed row data were significantly different 362 between the two quadrants (p = 0.002), thus were not pooled. In 2017, opposed row and north 363 quadrant unopposed row had more canopy spray deposition in the fruiting zone than the upper 364 canopy zones (Table 8) (Table 8). Ground drift 370 between vineyard quadrants was not significantly different in 2017 (p = 0.08), and therefore 371 pooled. Ground drift in downwind rows was significant in 2016, but not in 2017 (Table 8).

373
Mid-season. In 2016, except for DOY 210, both the opposed and unopposed rows had more spray 374 deposition in the upper canopy than in the fruit zones (Table 9). Total collected canopy deposition 375 on DOY 174 and 210 in the opposed row (72.5 ng/cm 2 and 54.8 ng/cm 2 , respectively) was more 376 than the unopposed row (60.6 ng/cm 2 and 45.8 ng/cm 2 , respectively) in 2016 (p = 0.03 and p = 377 0.04, respectively). Total collected deposition in the opposed row was significantly less than the 378 unopposed row in 2017 (p < 0.0001; opposed 47.9 ng/cm 2 ; unopposed 58.9 ng/cm 2 ). In 2017, vineyard quadrant data for opposed row applications were pooled as no significant differences 380 were seen between quadrants in canopy deposition (p = 0.10), aerial drift (p = 0.17), or ground 381 drift (p = 0.35). In 2017, unopposed row application was different between quadrants (p = 0.02) 382 but followed the increased deposition pattern in the upper canopy than fruiting zone pattern.  (Table 9). In 2017, the data on aerial drift from the two quadrants 390 were pooled as they were not different from each other (p = 0.17). Distance away from the sprayer 391 also influenced ground drift (Table 9), and more drift was collected in the row closet to the sprayer.  (Table 10).

403
Aerial and ground drift were not significantly different between the two rows (p = 0.64 and p = 404 0.68, respectively) and were pooled for analysis. There was no interaction between height above 405 the canopy and distance from the sprayer (p = 0.86), and height above the canopy alone did not 406 significantly influence aerial drift (p = 0.84). Distance from the sprayer did influence aerial drift, 407 with a gradient of more drift collected in the rows closest to the sprayer (Table 10). Distance from 408 the sprayed row influenced ground drift (Table 10) with more drift in the closest row.  (Table 11).

418
Aerial and ground drift were not significantly different between the two rows (p = 0.21 and p = 419 0.71, respectively) and were pooled for analysis. There was no interaction between height above 420 the canopy and distance from the sprayer (p = 0.56), and height above the canopy alone did not 421 significantly influence aerial drift (p = 0.44). Aerial drift was highest in the row closest to the sprayer (Table 11). Distance from the sprayer influenced ground drift (Table 11)    performance as it relates to deposition and drift. In our experiments, there was variable canopy 462 deposition between the opposed and unopposed rows, but those differences were not always 463 consistent (that is, we did not always see consistently higher deposition in opposed relative to 464 unopposed rows). While there have been concerns about spray deposition with multi-row sprayers 465 (Franson 2010), our results indicate that concerns about lack of deposition in unopposed rows is not warranted. In fact, in several cases we found equal to more deposition in the unopposed rows, 467 and this could be explained by the time between passes, which can allow droplets from the first 468 spray pass to dry before the second pass which can lead to increased deposition (Deveau 2016).

469
However, it should be noted that the application evaluations were conducted in generally optimal 470 spray conditions (moderate wind speeds) and that spray deposition in unopposed rows could be 471 altered if sprays were made under higher wind speed conditions that were stronger than the air 472 generated by the sprayers.  spray patterns. Each year, the sprayer was set so that the spray arms and nozzles were targeted to 497 account for canopy size, as described in materials and methods. However, in each year, a different 498 machine was used, and each machine and rate controller combination is unique. Rate controllers 499 were calibrated within their limitations, but travel speeds had to differ each season to achieve 500 similar application rates. The difference in tractor speed (Table 1)  Being able to modify the spray boom so that the 3-nozzle arrangement to target the small canopy 504 along the cordon is an advantage of this machine. During early season when the canopy was 505 smaller, the collection cards in the upper canopy were actually above the existing canopy that was 506 still close to the cordon. The early season sprayer set up adapted the typical 5-nozzle arrangement 507 of the machine to only use three nozzles that were directed towards the cordon and lower canopy,  Aerial drift was low in all years and application timings relative to the spray deposited to the 520 canopy, with almost no difference as distance increased from the sprayer. This was not the case 521 for ground drift, where there was typically more drift collected in the row closest to the sprayer.

522
However, there was only one instance where distance from the sprayer did not influence ground 523 drift (Table 8), but relative to other observation dates for ground drift, overall drift at this time 524 point was very low relative to canopy deposition across all collection distances (<1 ng / cm 2 ).   (75) a Three-row sprayers consist of two rows with opposing spray applications, and two half rows with unopposed spray. Two-row sprayers consist of only opposed spray rows. See Fig. 2E and F. All sprayers were over-the-row. b Growth stages based on the extended BBCH scale (Lorenz et al., 1994). BBCH 55 and 65 were defined as early season application timing; BBCH 75, 77, and 79 are defined as mid-season application timing.        11.3 a 0.3 ab 0.3 b 0.03 a In 2016, spray data was collected on DOY 2016-1 is DOY 174 and 2016-2 is DOY 210. In 2017, collection occurred on DOY 206 and 227; but there was no difference between the 2017 collection dates, so data is pooled between collection dates. 17.2 a 3.8 b 0.6 b 0.013 a In 2017, spray data was collected on DOY 157. b Ground data was pooled by row because year and DOY were not significant.