Simplified calculation of a theoretical spray coverage
Before interpreting spray coverages obtained under experimental spray trials, it was considered useful to provide a highly simplified estimate of the theoretical spray coverage (elimination of all physical and environmental effects). Consider a spray application of 100 L/ha of a pesticide formulation, which is equal to 10 mL/m2. If the average droplet diameter is 100 μm, then 10 mL is approximately 19 million droplets, and the theoretical spray coverage would be 0.6 m2/m2 or 60 %. This calculation is based on the assumption that the total surface area is completely flat and horizontal (the total surface area of a three-dimensional crop canopy is ignored), that the droplets are uniformly distributed, and that the entire volume of pesticide formulation is deposited on target surfaces (no drift and/or evaporation). In Western Australia, commercial pesticide spray applications in field crops are often conducted with 70–80 L/ha water carrier rates and combinations of nozzle orifice size and tank pressure, which delivers droplets with an average diameter of 200–250 μm; this means that the theoretical (highest possible if no drift and evaporation occur) spray coverage is 20 % or lower. These simple calculations appear to justify some general concern about the performance of pesticide spray applications when the potential spray coverage is that low. Furthermore, it highlights an aspect of pest management practices, in which more quantitative decision support tools, assessment methods, and clearer guidelines are needed.
Calculation of spray coverage
In this study, quantification of spray coverage was based on water-sensitive spray cards (5.1 cm × 7.6 cm, Syngenta, Wilmington, DE, USA), which are coated with bromoethyl blue and turn from yellow to blue/purple depending on dosage of water (Hoffman and Hewitt 2005; Nansen et al. 2011). Each spray card was labeled and stored in dry dark conditions prior to analysis in the laboratory. Spray coverage was a percentage measurement of blue/purple on individually digitized yellow spray cards. Analysis involved digitization using a color scanner at a spatial resolution of 5000 pixels/cm2, and image analysis using Image J 1.45 s (http://imagej.nih.gov) to determine the percentage of blue pixels (water droplets) on each spray card. Using threshold settings in ImageJ, the color image was converted into an 8-bit black/white image, and the following YUV color space settings were used as thresholds: Y = 172, U = 255, and V = 255. This method of quantifying spray coverage of pesticides has been widely used (Hill and Inaba 1989; Degre et al. 2001; Cunha et al. 2012, 2013; Hoffman and Hewitt 2005; Garcia et al. 2004; Martini et al. 2012; Nansen et al. 2011; Sánchez-Hermosilla and Medina 2004), and Hill and Inaba (1989) also found a strong and positive correlation between spray coverage and dosage of active ingredient on sprayed surfaces.
Experimental spray trials
Experimental spray applications were conducted on 11 separate days between June 2012 and June 2014 at three locations in Western Australia (Albany, −35.026920, 117.883801; Shenton Park, −31.950397, 115.798016; and Mingenew −29.198569, 115.438181). On each day of spray applications, we collected data with all 12 combinations of nozzle type and orifice size, so that we have the same number of observations and replications from all 11 days of spray applications. For each experimental spray application, two water-sensitive spray cards, about 1.2 m apart, were secured in a horizontal position 15 cm above a bare soil surface immediately before each spray application. In this study, the complexity of agricultural fields and its impact on pesticide movement was ignored, as we quantified spray coverages on horizontally placed spray cards above bare ground. This enabled direct comparison of effects of spray settings and weather conditions, but it also means that predicted spray coverages should be considered “relative” to an experimental standard.
Only water or water with the labeled rate (0.1 % by volume) of a non-ionic surfactant (SP700 surfactant, Genfarm, Landmark Operations, New South Wales, Australia) was sprayed. This adjuvant was chosen because we had anecdotal evidence of its widespread use in Australian agriculture. We tested effects of the following spray settings: (1) sprayer speed (15–35 km/h), (2) flat fan nozzle type (AIXR110, TP110, XR110, and TT110), (3) nozzle orifice size (02, 03, and 04), and (4) water carrier rate (35–140 L/ha). Regarding nozzle orifice size. The values for nozzle orifices (÷10, e.g. 0.3) indicate the liquid flow rate of the nozzle in US gallons per minute at 2.76 bar pressure (1 US gallon = 3.785 L).
Four commonly used nozzles in Australia are the Spraying Systems® Teejet® TT, TP, XR, and AIXR. TP is a tapered flat fan nozzle, XR is an extended range flat fan nozzle, and TT is a Turbo Teejet nozzle with an anvil shape which reduces liquid velocity thereby increasing average droplet diameter. The TP and XR nozzles have a similar droplet size spectrum at the same pressure and flow rate, but the spray velocities and coefficient of variation patterns may have differed. The XR denotes “extended range” and can be used at a greater range of pressures than the TP series. The TT (Turbo Teejet) nozzles produce larger droplets than the TP and XR series under similar conditions. The AIXR is an extended range flat-fan nozzle with air inclusion, creating larger droplets than the TT, TP, and XR series for the same flow rate by introducing air bubbles into individual droplets. If these air bubbles remain in the droplet to the point of deposition on a leaf, they may facilitate its spread to cover a larger area prior to drying.
The total data set consisted of spray coverage from 1796 water-sensitive spray cards and accompanying weather variables and spray settings. However, for 308 of the water-sensitive spray cards, the nozzle flow rate was unknown, so these data were not included in analyses involving this explanatory variable.
In all spray trials, a weather station (Kestrel 4500 Pocket Weather Tracker) was used in an open area of the field being sprayed to collect on-site weather data at 50 cm above ground level in 1-min intervals. The measurements included ambient temperature and dew point (°C), relative humidity ( %), wind speed (km/h), and barometric pressure (mmHg). These variables were chosen because they are readily available or could be easily measured with a standard weather station.
Statistical analysis of spray coverage data
All analyses of spray coverage data were conducted using PC-SAS 9.1 (SAS Institute, Cary, NC), and the objective was to quantify the relative influences of spray settings and weather conditions on both spray coverage and spray deposition efficiency (spray coverage/spray carrier rate). Spray coverages were arcsine transformed prior to statistical analyses. The water carrier rates ranged from 35 to 140 L/ha (131 different rates), and for average comparisons, the applications were divided into 10-L water carrier classes. All water carrier rates from 100 to 140 L/ha were grouped into a single water carrier class due to the low number of observations (N = 63).
It is important to highlight that spray coverage may not be linearly correlated with water carrier rate, as an increase in water carrier rate increases the chance of a droplet landing on a part of the water-sensitive card that already has a droplet (Hewitt and Valcore 2002). Thus, an initial regression analysis (PROC REG) was used to examine the direct effect of water carrier rate on spray coverage. PROC ANOVA was used to examine the effects of nozzle type, and use of a particular adjuvant was examined for each of the eight water carrier classes. PROC REG with linear, quadratic, and cubic responses was used to examine effects of sprayer speed on spray coverage. The same multi-regression approach was also used to examine effects of water carrier rate and nozzle flow rate on the obtained spray coverage. Finally, PROC REG was used to examine effects of all spray settings and weather variables on obtained spray coverages. In this latter analysis, forward stepwise selection was used to only select explanatory variables that contributed significantly to the regression fit.
The SnapCard app can be downloaded from Apple iTunes and Android Google Play stores by searching for snapcard. A manual for download, installation, and use of SnapCard is available at http://snapcard.agric.wa.gov.au, and a brief description is provided below. There are two complementary and inter-connected components of SnapCard: (1) a website and (2) a smartphone app. The interconnection of the two components is based on a user-defined login. The main functionalities of the website are (1) to introduce different combinations of spray settings and environmental conditions (Fig. 1c) to predict spray coverage with four different nozzle types, and (2) record-keeping of results from previous spray applications. In spray coverage predictions, it is only possible to generate predictions within the minimum and maximum ranges of the data used to generate the predictive models: nozzle orifice size (2, 3 or 4), sprayer speed (15–35 km/h), water carrier rate (50–90 L/ha), adjuvant (no = 0, yes = 1), barometric pressure (985–1025 mm Hg), relative humidity (15–85 %), ambient temperature (10–37 °C), and wind speed at ground level (0–30 km/h).
The main functionalities of the smartphone app are to allow field prediction of spray coverage and to conduct quality control of completed spray applications by photographing water-sensitive spray cards in the field using the smartphone camera (Fig. 1d). SnapCard records the GPS coordinates of the spray card and digitizes and quantifies droplet coverage. The user takes a photo of each spray card and crops the photo to process only the area of interest, i.e. the spray card. After quantification of spray coverages obtained from actual spray applications, these estimates can be compared directly with model predictions (based on spray settings and environmental conditions) as a form of quality control. Treatment information, personnel, spray settings, environmental data, predicted coverage, measured coverage, spray card photo, and GPS coordinates can all be archived on the secure SnapCard website.