(PDF) Spatiotemporal normalized ratio methodology to evaluate the impact of field-scale variable rate application

Precision Agriculture https://doi.org/10.1007/s11119-022-09877-4 Spatiotemporal normalized ratio methodology to evaluate the impact of field‑scale variable rate application L. Katz1,2,3,4   · A. Ben‑Gal4 · M. I. Litaor3,5 · A. Naor3 · M. Peres6 · I. Bahat1,7 · Y. Netzer8,9 · A. Peeters10 · V. Alchanatis1 · Y. Cohen1 Accepted: 14 January 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022 Abstract Wide assimilation of precision agriculture among farmers is currently dependent on the ability to demonstrate its efficiency at the field-scale. Yet, most experiments that compare variable-rate vs uniform application (VRA and UA) are performed in strips, concentrated in a small portion of the field with limited extrapolation to the field scale. A spatiotem- poral normalized ratio (STNR) methodology is proposed to evaluate the impact of VRA compared with UA for on-farm trials at the field scale. It incorporates a base year in which the whole plot is managed with UA and consecutive years in which half of the plot is man- aged with UA and the other half is managed with VRA. Additionally, a novel normalized relative comparison index (NRCI) is presented where the ratios of VRA/UA sub-plots are compared between a base year and a consecutive year, for any measured parameter. The NRCI determines the impact of VRA on variability using statistical measures of dispersion (variability measures) and on performance with statistical measures of central tendency (performance measures). Variability measures with NRCI values lower or higher than 1 indicate VRA management decreased or increased variability. Performance measures with NRCI lower or higher than 1 indicate subplot impairment or improvement, respectively due to VRA management. The methodology was demonstrated on a commercial drip irrigated peach orchard and a wine grape vineyard. NRCI results showed that VRA drip irrigation reduced water status in-field variability but did not necessarily increase yield. The benefits and limitations of the proposed design are discussed. Keywords  Variable rate application · Precision irrigation management · Normalized relative comparison index · Stem water potential · Variability · Performance measures Abbreviations CWSI Crop water stress index MC Management cell NRCI Normalized relative comparison index PA Precision agriculture PAS Precision agriculture system * L. Katz

Extended author information available on the last page of the article 13 Vol.:(0123456789) Precision Agriculture PM Performance measure STNR Spatiotemporal normalized ratio SWP Stem water potential TSS Total soluble solids UA Uniform application VM Variability measure VRA Variable rate application VRDI Variable rate drip irrigation WP Water productivity Introduction Precision application of agricultural inputs, such as water, fertilizers, pesticides, and her- bicides, strives to optimize crop profitability and sustainability by increasing yield and improving input efficiency while concurrently decreasing spatial and temporal variability and diminishing adverse environmental effects. Wide assimilation of precision agricul- ture among farmers is currently dependent on the ability to demonstrate its efficiency at the field-scale. Quantitative tools are required to evaluate data from field trials to estab- lish whether precision strategies successfully fulfill their target optimization goals, e.g. if they have succeeded in increasing profitability, reducing spatial variability, and decreasing adverse environmental effects. In 2020, the International Society of Precision Agriculture (ISPA) supported a study carried out by Prof. Francesco Marinello from the University of Padova to define a “precision roadmap”. The most important objective suggested by the 50 researchers worldwide engaged in the study was “design of on-farm variable rate applica- tion (VRA) trials that can be implemented by farmers as ’business as usual’ together with protocols, methods, and tools to analyze them for economic and environmental benefits compared to uniform application. This will include research aimed at optimizing the uti- lization of data and understanding the learning needs and economical improvements that would encourage the precision agriculture (PA) approach adoption”. This vision clearly states the necessity for a method that can simply evaluate the added-value of VRA over UA for applied research and practical implementation at the field scale. There is presently no putative established methodology to evaluate the impact of VRA compared with UA man- agement on commercial-scale on-farm trials. Currently, three approaches exist for the comparison of VRA to UA (Table 1): 1. Strip approach: Application performed in strips, mostly in a small portion of the field. 2. Whole plot approach: Application performed for an entire plot. 3. Subplot approach: Division of a field into two or more subplots with application per- formed in each subplot. The strip approach is the most common technique for comparison, stemming directly from randomized complete block experimental design where the input is applied in strips, usually on a relatively homogenous part of the field (Colaço & Molin, 2017; Cordero et al., 2019; Dammer et al., 2009; Esau et al., 2014; Liakos et al., 2020; Ma et al., 2014; O’Shaughnessy et  al., 2015; Stamatiadis et  al., 2018; Vellidis et  al., 2016; Yang, et  al., 2001; Zaman et al., 2011). 13 Table 1  Comparison of uniform application (UA) to variable rate application (VRA) in literature Approach Area of PA Research Authors Crop Compared Parameters Method of Compari- Length of Base year son (UA vs. VRA) research (years) Strip Precision irrigation O’Shaughnessy et al., Cotton (Gossypium Yield, irrigation amt, Mean comparison 2 No Precision Agriculture (2015) hirsutum) WUE between manual and automatic irrigation scheduling methods Vellidis et al., (2016) Peanuts (Arachis Yield, irrigation amt Mean comparison 1 No hypogaea) between UI and VRI areas Precision fertilization Yang, et al., (2001) Grain Sorghum (Sor- Yield, economic ANOVA 2 No ghum bicolor) return Ma et al., (2014) Maize (Zea mays) Yield, NUE, soil Mean comparison 2 No mineral N Colaço & Molin, Citrus trees Input consumption, Comparison of total 6 Yes; not incorporated (2017) yield, soil fertility inputs between in comparison treatments Stamatiadis et al., Durum wheat (Triti- Yield, elements in Mean comparison 1 No (2018) cum durum) leaves and grain, between strip treat- NUE, net return ments during 2016; comparison between treatments in dif- ferent landscape locations Cordero et al., (2019) Maize (Zea mays) Yield, NUE, net return Comparison per site 3 Base year data not used and year in this study Liakos et al., (2020) Apple (Malus domes- Yield, soil and fruit Mean changes 3 Base year used to tica) quality properties between UA and determine the amount VRA per year (2011 of fertilizer for sub- and 2012) sequent year and for soil analysis, not for comparison 13 Table 1  (continued) Approach Area of PA Research Authors Crop Compared Parameters Method of Compari- Length of Base year son (UA vs. VRA) research 13 (years) Precision applica- Dammer et al., (2009) Winter wheat Yield, application Mean comparison 1 No tion of rate, savings fungicide and her- Zaman et al., (2011) Wild blueberry (Vac- Percent area coverage Mean comparison 1 No bicide cinium angustifo- (PAC), weed height with t-test lium) Esau et al., (2014) Wild blueberry (Vac- Yield, plant growth ANOVA 1 No cinium angustifo- parameters lium) Subplot Precision irrigation Sanchez et al., (2017) Grape (Vitis vinifera) Yield, WUE Gain of VRI vs UI 4 Yes; not incorporated per year in comparison Ortuani et al. 2019 Grape (Vitis vinifera) Yield, fruit quality ANOVA 1 No and maturation parameters, WUE Whole plot Precision fertilization Yost et al., (2017) Corn (Zea mays), Yield Percent difference 20 Yes; 5 year average soybean (Glycine between pre- PA max), wheat (Triticum and post-PA systems aestivum) Precision Agriculture Precision Agriculture Using the strip approach, a large number of treatments and repetitions can be success- fully incorporated into the experimental design, and thus conventional statistical analysis is possible. Methods of comparison between application techniques include a simple means comparison, t-test, and ANOVA, depending on the number of treatments. Due to the small size of the experimental strips relative to the entire plot, the full scope of spatial informa- tion of a field remains unknown. Additionally, strip experiments usually span from one to three years, limiting the evaluation of temporal variability between treatments. The whole plot approach is applied for long-term studies where fields are converted from conventional to precision management. In the research of Yost et  al.,  (2017), the whole plot approach was applied to a corn-soybean rotation plot for 20 years. The entire plot was farmed for ten years using conventional agriculture (pre-precision agriculture sys- tem – PAS, UA method) followed by ten more years using the PAS system (VRA of fertili- zation). Due to rotation, each crop was grown for approximately five years pre-PAS and an additional five years with PAS. UA and VRA methods were compared by calculating the percent difference of yield and yield CV between averaged pre-PAS and post-PAS for each specific plot. This type of experiment is spatially representative of the entire plot. As such, it is currently perhaps the optimum approach for determining the full potential for precision management of a specific field, realizing the potential of both its spatial and temporal vari- ability. However, experiments with such a degree of spatial and, especially, temporal data are rare due to logistical limitations (Yost et al., 2017). Trade-offs clearly exist between the size of the experimental plots, their spatial repre- sentation of the variability within the field, and the experiment’s duration. In strip experi- ments, small plots are used with a large number of repetitions relative to plot size and the experiments are commonly short-term. Increasing plot size improves the spatial represen- tation. However, in experiments with such large plots, the number of repetitions per treat- ment is usually small, and longer experiment length is necessary to compensate for the relatively small number of samples. Due to temporal variability, a reference is needed by which to calibrate the meteorological changes that occur in the field from year to year and affect the yield and spatial variability patterns of various parameters. The subplot approach attempts to calibrate temporal changes by incorporating a base year. A base year is defined as a growing season when UA is performed on the entire experimental area, preceding the use of VRA in part of the entire field. This approach was presented once by Sanchez et  al., 2017. Subplot experiments are both spatially and temporally more representative than the strip approach because the plots can be signifi- cantly larger than individual strips and the experiment duration is typically at least slightly longer than strip experiments. Subplots only represent a portion of a field (different from the whole plot approach), and therefore the comparison between UA and VRA subplots remains important. The precision irrigation research of Sanchez et al. (2017) used the sub- plot approach and that lasted four years. Following one base year, a vineyard was divided into two subplots, with UA and VRA management plots for an additional three years. Every year, changes in yield and water use efficiency were compared between the UA and VRA plots. The spatial variability of the vineyard was not identical between subplots dur- ing the base year and this difference was not taken into consideration in the comparison between the two subplots. This practice limits the use of base-year data to relatively similar subplots and prevents the use of such an approach in more heterogeneous fields. Limitations of the existing approaches are summarized as follows: 13 Precision Agriculture 1. The strip approach is small scale, implemented in restricted strips in the field, with results difficult to be extrapolated to the field scale. 2. The whole-plot approach is by definition conducted at the field scale yet meaningful results from its application may take a long time (10 years in the single trial found in the literature). 3. The subplot approach is also by definition field scale but requires plots that can be divided into two similar subplots. The primary objective of this paper was to propose a design for on-farm trials incorpo- rated into a spatiotemporal normalized ratio (STNR) methodology to evaluate the impact of VRA on in-field variability and on performance measures compared to UA at the field- scale. The methodology proposed belongs to the subplot approach and is composed of (1) a base year, in which the whole plot is managed with UA; (2) consecutive years in which half the plot is managed with UA and the other half is managed with VRA; and (3) a new normalized relative comparison index (NRCI). It is hypothesized that this design would enable spatiotemporal normalization of the VRA impact compared to UA and would ena- ble evaluation of the potential of precision management at the field-scale. Secondary objec- tives were to demonstrate the proposed methodology on a commercial drip irrigated peach orchard and a wine grape vineyard, present results and discuss the benefits and limitations of the proposed design. Materials and methods Spatiotemporal normalized ratio methodology to compare VRA and UA in field‑scale trials The on-farm trial STNR methodology is composed of the following steps (Fig. 1): Experimental design A commercial plot is divided into two equal subplots. In the first season, denoted base – year (t0), both are managed with UA. During succeeding years (ti), one subplot is man- aged with VRA and the other by UA. Sampling Target parameters (p) such as yield, plant/soil status, or resource-based parameters includ- ing water, fertilizer, or pesticides are collected in the base-year and in the succeeding years by grid-based sampling. This allows for adequate representation of both the average and the variance in the entire plot and each subplot (Kerry et  al., 2010). Alternatively, algo- rithms for optimal sampling locations may be used to significantly reduce the number of sampling points (Minasny & McBratney, 2006; Ohana-Levi et al., 2021). Type of measure Statistical measures of the collected parameters are calculated. These types of measures (m) are divided into two groups: variability measures (VM) and performance measures 13 Precision Agriculture Base year t0 t1 tn Season Experimental design (ti, i = 1… n) Plot is halved Management VRA* VRA VRA VR type A UA UA UA VRA UA *VRA = UA at Base year VRA Sampling UA Measure Type of measure Variability Performance (m) e.g. yield, resource-use- Parameter e.g. plant/soil status, yield efficiency (dispersion stascs: stdv, var, cv) (central tendency stascs: (p) mean, median and mode) Normalized relave NRCI comparison index (NRCI) NRCI result Interpretaon NRCI < 1 NRCI = 1 NRCI > 1 m VRA management VRA management Variability decreased variability increased variability (++) VRA has no (--) added value VRA management VRA management (-) increased ag. Performance decreased ag. performance performance (--) (++) Fig. 1  A scheme of the developed Spatiotemporal Normalized Ratio methodology for the evaluation of var- iable rate application (VRA) compared with uniform application (UA) management where VRA/UA plot ratios are compared at base-year ­(t0) and additional year ­(ti) for any target parameter (p) and type of meas- ure (m). VRA* refers to the VRA plot where UA was applied during the base year (PM). The VM are measures of field variability and are calculated using dispersion statisti- cal measures, such as standard deviation, variance, or coefficient of variation (CV). The PMs are measures of field performance and are calculated using statistical quantifiers, such as mean, median, or mode. VM and PM can be calculated to any target parameter (p). 13 Precision Agriculture Normalized relative comparison index A normalized relative comparison index (NRCI) is proposed to spatiotemporally and quantitatively evaluate the impact of VRA on a field (Eq. 1). Variably applied inputs may include water, fertilizer, pesticides, herbicides, or others. The index compares the ratio of VRA/UA of a measure (m) and any target parameter (p) at ti (numerator) with the same ratio at t0 (denominator), where t0 and ti represent seasons before and during VRA imple- mentation, respectively. ⎛ VRAti ⎞ ⎜ UAt ⎟ Normalized relative comparison index (NRCI) = ⎜ VRA∗i ⎟ (1) ⎜ t0 ⎟ ⎝ UAt0 ⎠mp VRA* indicates the specific VRA subplot within the field where UA was implemented dur- ing the base year (t0). This ratio allows for the normalization of the inherent differences between the two subplots of the field as well as any different environmental conditions between years, thus enabling the calculation of the relative contribution of VRA. Interpretation An NRCI value of 1 indicates no change, i.e. VRA did not affect the type of measure of a specific parameter mp thus VRA had no added value. NRCI values lower and higher than 1 are interpreted based on the specific research targets. Variability measures with NRCI values lower or higher than 1 indicate that VRA management decreased or increased vari- ability, respectively. In contrast, performance measures NRCI values lower or higher than 1 indicate the VRA management impaired or improved the field, respectively. Initial demonstration of the spatiotemporal normalized ratio methodology in a peach orchard and vineyard The developed on-farm trial design STNR methodology was evaluated for two experimen- tal data sets, from VRDI trials in a peach orchard and a wine grape vineyard (Table  2). These were chosen because, while both crops are profitable, they have different objective functions and irrigation strategies. The objective in the peach orchard was primarily to ben- efit profitability by increasing water productivity (WP) while maintaining yield and sec- ondarily to decrease the variability of both SWP, yield and WP. The objective in the vine- yard was primarily to promote fruit quality for wine production by reducing the variability of SWP, and secondarily to increase yield and WP in cases where yields were lower than target yield. The variability and performance measures were calculated using the CV and the mean statistic, respectively for all parameters at each site (Table 2). The CV statistic was specifically incorporated to null orders of magnitude between years (e.g. extremely low orchard yield in 2018). 13 Precision Agriculture Table 2  Summary of STNR methodology components for the peach orchard and vineyard on-farm trials: experimental design, sampling and type of measure Method Components Peach Orchard Vineyard Experimental Design Base year ­(t0) 2017 (UA in both parts of field) Additional years ­(ti) 2018, 2019 (UA in south and VRA in north) 2018 (UA in east and VRA in west) VRA type Drip irrigation Sampling Grid-based with management cells Type of Measure Variability Measures (VM) (statistic: CV) Yield (weight), tree water status (SWP in stage III) Yield (weight), vine water status (SWP in stages I and III), Brix, cluster weight Performance Measures (PM) (statistic: mean) Yield (weight), WP Yield (weight), WP, Brix, cluster weight 13 Precision Agriculture Fig. 2  a Mishmar Hayarden peach orchard: 22 management cells (MC) (orange delineated cells) * 6 meas- urement trees (red points) = 132 total measurement trees. b Mevo Beitar vineyard: 20 MCs (purple deline- ated cells) * 6 measurement vines (purple points) = 120 total measurement vines (Color figure online) Case study 1: peach orchard Research area A field experiment was conducted between 2017 and 2019 in a 4  ha commercial late- harvest peach (Prunus persica cv. 1881) orchard. The orchard is located next to Mishmar Hayarden village (33.01°N; 35.60°E) in the Upper Galilee region in Israel (Fig. 2a). Eleva- tion ranges from 171 – 188 m ASL, the general slope is 5% to the northwest and within the orchard the slope ranges from 0 – 11.3%. Precipitation is limited to winter months, typically from October until April. The total precipitation for the rainy seasons prior to experimental growing seasons in 2016–2017, 2017–2018 and 2018–2019 was 228, 377 13 Precision Agriculture 2017 2018-2019 2017 VRDI UI UI UI e a b 2018 -0.60 -0.80 d -1.00 VRDI SWP (MPa) -1.20 -1.40 -1.60 PH -1.80 -2.00 I II III UI -2.20 15-Oct 25-Oct 15-Sep 25-Sep 18-May 28-May 17-Jun 27-Jun 28-Apr 8-May 17-Jul 27-Jul 16-Aug 26-Aug 6-Aug 5-Oct 7-Jun 7-Jul 5-Sep c 2019 Target Upper limit - turn on irrigaon Lower limit - turn off irrigaon f Fig. 3  a Uniform irrigation (UI) was applied to the entire Mishmar Hayarden peach orchard in 2017 and the southern subplot in 2018 and 2019. b Variable rate drip irrigation (VRDI) was implemented in the north plot in 2018 and 2019 to control 11 management cells (MC) (orange). c VRDI was applied according to a target value and lower (blue) and upper (red) limits that differed between growth stages I-III and post- harvest (PH). d The system included a manifold with 11 electronic valves to control the north subplot. e UI was applied to the entire Mevo Beitar vineyard in 2017 and the east subplot in 2018. f VRDI was imple- mented in the west subplot in 2018 allowing control over MC A-J (purple) (Color figure online) and 603 mm, respectively (Gadot meteorological station, 33.03° N; 35.62° E). The orchard was planted in 2007 with 2.6 m and 5 m between trees and rows, respectively. The rows run from northwest to southeast. The orchard was divided into two subplots: north and south and each subplot was further divided into 11 management cells (MC) with dimensions of 35 m X 35 m (approximately 78 trees each) (Fig. 2a). Six trees in each MC (total of 132 trees, 66 trees in the north and south plots) were defined as measurement trees. Irrigation system and decision‑making Uniform irrigation (UI)  The entire orchard (season 2017) and the southern plot (seasons 2018 and 2019) were irrigated by the grower according to common local commercial prac- tice (Fig. 3a). The general irrigation strategy for peach was one of excess irrigation to ensure that the trees were not under stress. Stage I is the initial stage of fruit growth when cell- division occurs, followed by stage II, the pit-hardening lag-stage. Stage III is the primary fruit growth-stage when cell expansion takes place (Naor, 2006). Following the harvest, during the post-harvest (PH) stage, the tree accumulates carbohydrates, which affects flower bud development for the following year (Naor et al., 2006). Most of the annual irrigation is applied during growth stage III, the primary fruit growth stage, when approximately 100% and sometimes more of the reference evapotranspiration ­(ET0) is returned to the orchard. Irrigation amounts (crop evapotranspiration, ­ETc) were based on the multiplication of Agri- cultural Extension Service of Israel (Shaham) crop coefficients (­ kc) and calculated E­ T0, cal- culated according to the Penman–Monteith equation ­(ETc = ­kc * ­ET0) (FAO56). Irrigation 13 Precision Agriculture decision-making occurred weekly. Irrigation was applied daily using Uniram (Netafim™, Tel Aviv, Israel) integral pressure compensating drippers (2.3 L ­h−1) every 0.5 m with one dripline per row. Water meters for the entire plot (2017) and the southern plot (2018 and 2019) were manually read weekly during the season. Actual dates for phenological stages for the years 2017 and 2019 are shown in Online Resources Figure S1. For season 2018, the stage dates were: 6 May 2018 – 21 May 2018 (I), 22 May 2018 – 27 June 2018 (II), 28 June 2018 – 19 Aug 2018 (III) and 20 Aug 2018 – 21 Oct 2018 (PH). Variable rate drip irrigation  In March 2018, a differential irrigation system was installed that included 11 electronic valves and water meters (Fig. 3d). A computer control system (Dream2, Talgil Ltd., Kiriyat Motzkin, Israel) was used to give direct control over MCs 1–11 and each valve controlled a single MC. The term variable rate drip irrigation is used to define the system and refers to the fact that the rate of water application can be spatially variable between different sites in a specific field and temporally variable between dates of application. However, the flow rate of the applied water is stable and constant. The north plot of the orchard was differentially irrigated during seasons 2018 and 2019 (Fig.  3b). During 2018 and 2019, irrigation decisions were made twice per week according to the calculated ­ETc for the entire orchard and then by correcting for each MC according to its average SWP. The SWP values were compared to pre-defined target values for each growth stage with upper and lower limits defined as 10% of the target value (Fig. 3c). During most of growth stage III of season 2019, when the water requirement was at its peak, calculated SWP values, based on thermal imaging, were used for the correction (Online Resources Fig- ure S2). Irrigation was applied using Uniram (Netafim™, Tel Aviv, Israel) integral pressure compensating dripper (1.6 L h­ −1) every 0.5 m with two drip lines per row. The computer control system automatically recorded daily amounts of water applied to each plot. Addi- tionally, the water meters were manually read approximately once every week during the season to calculate the accumulated irrigation of each MC. Data collection and analysis The data collected for the peach orchard is presented in Table 3. Stem water potential Stem water potential measurements were performed during 2017 – 2019 (Table 3). Three to six trees were sampled per MC for both the north and south plots. During 2018, three trees were sampled in five MCs. Plant water status was evaluated by directly measur- ing SWP using a Scholander-type pressure chamber (Arimad, MRC Ltd., Holon, Israel). One to two shaded leaves were covered with an aluminum foil zip-lock bag 1.5 h before the measurement (Naor et  al., 2006). Measurements were performed between the hours of 12:30–15:15. A stage III comparison was made between the first available measure- ment day (denoted beginning) and the last available measurement day (denoted end). SWP results were averaged per tree (if more than one leaf was measured) and then per MC for each measurement day (Online Resources Figure S1). 13 Precision Agriculture Table 3  Summary of Mishmar Hayarden peach orchard data collection: parameters, years, measurement dates and number of samples Parameter Year Stage III Frequency/ Measurement Dates Number of sampled trees per north (N) and south (S) plots SWP 2017 09 July, 20 July, 06 Aug, 07 Aug, 29 Aug 66 (N), 66 (S) 2018 2 day/week 33 (N), 15(S – 11 July, 18 (28 June – 19 Aug) July) 2019 2 day/week  ~ 44 (N), ~ 33 (S) (27 June – 29 Aug) N only – 03 July, 10 July, 17 July, 29 Aug Thermal imaging 2017 20 July, 29 Aug Entire orchard 2018 05 Aug Entire orchard 2019 21 July, 25 July, 29 July, 01 Aug, 04 Aug, Entire orchard 08 Aug, 12 Aug, 15 Aug Yield 2017 22–25 Aug, 27–29 Aug 66 (N), 66 (S) 2018 12–13 Aug, 19 Aug 66 (N), 66 (S) 2019 14–15 Aug, 23 Aug, 25–26 Aug, 28 Aug 66 (N), 66 (S) Yield The total yield and number of fruit of all 132 measurement trees were determined during the commercial harvest. Peaches were selectively harvested according to ripeness. There- fore, the harvest spanned several nonconsecutive days (Table 3). On each harvest day, all the ripe fruit (50% reddish fruit coloring – the guidelines were provided by the grower) was collected from each tree. Fruit weight and number of fruit per tree were recorded. Yield and total number of fruit were summed for each tree and averaged for every MC. The irri- gation water productivity is the efficiency by which the peach orchard or vineyard produces a crop as a function of input quantity, not including soil water reserves (herein water pro- ductivity, WP) calculated using Eq. 2: [ ][ ] WP = [Yield]∕ Irrigation ton m−3 water (2) where Yield units are ton h­ a−1 and Irrigation is the actual amount of water given to a spe- cific field in ­m3 ­ha−1. Descriptive statistics (mean, standard deviation, and CV) were calcu- lated per MC for the north and south plots (Online Resource Table S1). Case study 2: wine grape vineyard Research area A field experiment was conducted over two years (2017, 2018) in a highly variable 2.4 ha Vitis vinifera cv. ‘Cabernet Sauvignon’ vineyard, established in Mevo Beitar (31.43° N; 35.06° E) in the Judaean hills region in Israel (Fig. 2b). Elevation ranges from 676–700 m ASL, slope ranges from 1–22%, and soil depth ranges from 0.35–1.60  m. Vines were planted in 2011 in a northwest-to-southeast direction with vine and row spacing of 1.5 m and 3.0  m, respectively. The vineyard was divided into east and west subplots and then each subplot was further divided into 10 MCs with dimensions 30 m X 30 m. Six vines 13 Precision Agriculture per MC were defined as measurement vines. Precipitation is limited to winter months, typically from October until April. Average yearly precipitation from 2013–2020 at the site was 537 mm. Annual precipitation for rainy seasons 2016–2017 and 2017–2018 was 369 mm and 452 mm, respectively (Tzuba meteorological station, 31.78° N; 35.12° E). Irrigation system and decision‑making Uniform irrigation  Drip irrigation was uniform for all 20 MCs in the vineyard during 2017 and for the east subplot during 2018 (Fig. 3e). The general management strategy was regu- lated deficit irrigation to ensure that the vines were under sufficient stress to produce high- quality grapes for wine. During stage I, the berries undergo cell division, and the maximal size potential of each berry is determined. The importance of maintaining minimal drought stress in this stage was emphasized in Cabernet sauvignon and Merlot vineyard in semi-arid conditions (Munitz et al., 2020; Netzer et al., 2019). During stage II, there is an initial accu- mulation of sugar and acid in the berries. Cell expansion takes place to a small degree dur- ing stage II although it is generally considered a lag stage (Kennedy, 2002). Stage III is the main growth stage of the berries and critical in terms of anthocyanin and polyphenols pro- duction and production of additional quality components. Irrigation during stage III is con- sidered necessary to a degree but must also be sufficiently regulated to ensure grape quality (Castellarin et al., 2007). Post-harvest irrigation is purely technical. Irrigation amounts were calculated using a model based on lysimeter-derived E ­ Tc and additional weekly field LAI measurements (Munitz et al., 2019; Netzer et al., 2009). ­ET0 was calculated according to the Penman–Monteith equation and the irrigation factor was corrected by average plot SWP. Irrigation was applied using Uniram (Netafim™, Tel Aviv, Israel) integral pressure compen- sating dripper (1.6 L h­ −1) every 0.75 m. The dates for phenological stages for the years 2017 and 2018 are shown in Online Resources Figure S4. Variable rate drip irrigation  The VRDI system was installed in spring 2018 and included 11 electronic valves enabling the independent irrigation of each of the 10 MCs in the west sub- plot and one for the entire east UI subplot (Fig. 3f). All of the valves were controlled by an irrigation controller (Dream 2, Talgil, Kiriyat Motzkin, Israel). Irrigation management was according to an optimal SWP curve for all phenological stages determined suitable for the cultivar, region, and winery requirements (Munitz et al., 2017). MCs in the western block (A-J) were differentially irrigated according to their individual weekly SWP. Data collection and analysis The vineyard data is concentrated in Table 4. Stem water potential Stem water potential measurements were performed during 2017—2018 on six vines per MC every one-two weeks (Table  4). SWP was measured before irrigation close to solar noon (from 12:00 to 14:30), using a portable pressure chamber (Arimad 3000, MRC, Holon Israel) according to the procedures described by Boyer (1995). Evaluation of years 2017 and 2018 for stages I and III was made by comparing the first available measure- ment dates at the beginning and end of these stages. Measured stem water potential results were averaged per MC for each measurement day (Online Resources Figure S4). A map 13 Precision Agriculture Table 4  Summary of Mevo Beitar vineyard data collection: parameters, years, measurement dates and number of samples Parameter Year Stage I Frequency/ Measure- Stage III Frequency/Measurement Dates Number of sampled vines ment Dates per west (W) and east (E) plots SWP 2017 1 day / 2 weeks 1 day / 2 weeks 16 Aug – 05 Sep 60 (W), 60 (E) 07 June – 07 July 2018 1 day / week (W) 1 day / week 08 Aug – 29 Aug 60 (W), 60 (E) 1 day / 2 weeks (E) 30 May – 27 June Yield (Mechanical harvest) 2017 09 Sep Entire vineyard 2018 05 Sep Entire vineyard TSS 2017 06 Sep 60 (W), 60 (E) 2018 04 Sep 60 (W), 60 (E) Cluster weight 2017 06 Sep 60 (W), 60 (E) 2018 04 Sep 60 (W), 60 (E) 13 Precision Agriculture visualization of the averaged MC values at the end date of stages I and III each year is shown in Fig. 8. Yield and water productivity Grapes were harvested mechanically each season (Table  4). Fruit was measured with a continuous weighing system to provide a yield map (ton h­ a−1) at a spatial resolution of 5 m. The average yield per MC was calculated from the yield map. The WP was calculated using Eq. 2. Fruit quality TSS: One day before the mechanical harvest, 36 randomly chosen bunches per MC, were hand crushed for determination of TSS using a digital refractometer (Palette, Atago Co. Ltd., Japan) (Table 4). Degrees Brix is a measurement of sugar content and the winery tar- get range is 23.5–25° Brix for this specific plot. Cluster weight: Prior to the general harvest, 12 measurement vines in each MC were harvested separately when the fruit TSS reached a minimum of 22.5° Brix (Table 4). The number of bunches per vine and total yield were recorded using a HS-15 K hanging scale (Universal Weight Enterprise Co., Ltd., Taiwan). Descriptive statistics (mean, standard deviation, and CV) were calculated for the above parameters per MC per measurement day for the west and east plots (Online Resources Table S2). Lastly, the % difference of each measured parameter between the UA and the VRA plots for a specific year was calculated using Eq. 3. Percent difference The percent (%) difference of each measured parameter between the UA and VRA subplots for a specific year was calculated using Eq. 3. This is the conventional method of compari- son between treatments and calculated for the strip and whole plot approaches. (VRA − UA) % Difference = (3) UA The % difference incorporating the base-year normalization NRCI was calculated using Eq. 4: ( ) ( ) VRA VRA∗ UA t − UA t0 ( ) (4) i % Difference in NRCI = = NRCI − 1 VRA∗ UA t0 where t0 and ti represent years before and during VRA implementation, respectively. VRA* indicates the specific VRA plot within the field where UA was implemented during the base year (t0). These percent difference calculation methods are used to compare the pro- posed STNR methodology with the other approaches. 13 Precision Agriculture to (2017) t1 (2018) t1 (2019) Irr. = 1,087 Avg. = 570 Avg. = 734 SE = 18 SE = 22 Irr. = 589 Irr. = 763 Fig. 4  Map visualization of total irrigation (mm) in Mishmar Hayarden peach orchard in years 2017 – 2019: during the base year (2017) uniform irrigation (UI) was implemented in the south subplot and during years 2018 and 2019 variable rate drip irrigation (VRDI) was implemented in the north subplot for irriga- tion of management cells (MC) Results Case study 1: peach orchard Accumulated irrigation The irrigation amount for both the north and south subplots during the base year (2017) was 1,087 mm (Fig. 4). The irrigation amounts for the south UI plot were 589 and 763 mm during 2018 and 2019, respectively, while the average irrigation of the north VRDI subplot was 570 mm in 2018 and 734 mm in 2019. During 2018 the MCs that received more irri- gation were located in the northwest corner and the east area of the north subplot. During 2019, most of the north and east cells were differentially irrigated with larger amounts of water as a result of relative stress in these areas as indicated by the calculated SWP. Normalized relative comparison index NRCI values for all measures and parameters are presented in Table  5. In the peach orchard, VRDI reduced the variability of water status during stage III in both 2018 and 2019 (Fig. 5). The SWP NRCI values were reduced from the beginning to end of stage III (Table  5). This implied that the VRDI generally succeeded in bringing each MC closer to the target SWP and thus reducing the water status variability in the north subplot as compared to the south subplot. Additionally, the NRCI yield variability values were sub- stantially low, with values of 0.41 and 0.23 in 2018 and 2019. The NRCI values of yield and WP variability were also considerably lower than one. The peach orchard NRCI of the performance measures in 2018, however, were less than one, indicating that the reduction in water status variability did not improve yield or WP during the 2018 season. The NRCI results for these parameters during the 2019 season were greater than one and thus suggest that VRA management seemingly improved both the yield and WP. Comparison of percent difference results Substantial differences were recorded for all orchard variability measures between the per- cent difference using the NRCI (Eq. 5) and using the conventional method for comparison 13 Precision Agriculture Table 5  Normalized relative comparison index (NRCI) values of variability measures (VM) and perfor- mance measures (PM) of the peach orchard and the vineyard Type of measure and parameter Peach orchard Vineyard t0 = 2017 t0 = 2017 NRCI ­(ti = 2018) NRCI ­(ti = 2019) NRCI ­(ti = 2018) Variability Parameters (VP) (using CV) Yield 0.41 0.23 1.07 Water Productivity (WP) 0.48 0.40 1.82 SWPmeasured  Beginning of stage I 1.78  End of stage I 0.56  Beginning of stage III 0.94 1.83 2.23  End of stage III 0.30 0.25 0.91 Fruit quality  Brix 1.56  Cluster weight 1.05 Performance Parameters (PP) (using mean) Yield 0.90 1.29 0.94 Water Productivity (WP) 0.91 1.47 1.04 Fruit quality  Brix 1.00  Cluster weight 1.00 to (2017) t1 (2018) t1 (2019) Avg. = -1.23 Avg. = -1.15 Avg. = -1.41 North SD = 0.12 SD = 0.05 SD = 0.08 CV = 0.096 CV = 0.040 CV = 0.056 South Avg. = -1.22 Avg. = -1.13 Avg. = -1.56 SD = 0.06 SD = 0.07 SD = 0.17 CV = 0.049 CV = 0.062 CV = 0.111 a b c Fig. 5  Measured stem water potential (SWP) (MPa) per management cell (MC) of Mishmar Hayarden peach orchard at end dates of stage III during years 2017 (a), 2018 (b) and 2019 (c) between VRA and UA subplots (Eq. 4) (Fig. 6a). However, the trend between the two cal- culation methods was similar for both 2018 and 2019 seasons for yield, WP and SWP (end of stage III) parameters. For example, the conventional percent difference of yield vari- ability during season 2019 − 57%. However, following plot normalization using base year data a 77% decrease was revealed. The comparison of the 2018 SWP stage III beginning and end measurements showed a different trend. No practical difference was calculated between the beginning and end of stage III (− 34% and − 35%, respectively) using the con- ventional method of comparison. However, a significant difference was calculated between the beginning and end measurements (− 6% and − 70%, respectively) when the base year 13 Precision Agriculture 2018 NRCI - 1 100% 2018 %Diff 83% 2019 NRCI - 1 80% 2019 %Diff 60% 47% 47% 42% 40% 26% 29% 20% % Difference 3% 0% -1% -6% -10% -9% -20% -18% -26% -40% -32% -34% -35% -52% -50% -60% -59% -57% -60% -70% -80% -75% -77% -100% Yield (CV) WP (CV) SWP - Stage III SWP - Stage III End (CV) Yield (avg.) WP (avg.) Beginning (CV) a Variability Measures b Performance Measures Fig. 6  Comparison of percent difference in normalized relative comparison index—% Diff NRCI (NRCI-1) and percent difference—%Diff ((VRA-UA)/UA) for 2018 and 2019 in measures of variability (a) and meas- ures of performance (b) in the Mishmar Hayarden peach orchard (2017) to (2018) t1 Irr. = 122 Avg. = 47.9 Irr. = 47.3 SE = 4.4 Fig. 7  Map visualization of total irrigation (mm) in Mevo Beitar vineyard in years 2017 and 2018: dur- ing the base year (2017) uniform irrigation (UI) was implemented and in 2018 variable rate drip irrigation (VRDI) was implemented in the west subplot for irrigation of management cells (MC) data was incorporated using the NRCI. Minor or no difference between percent difference calculations was evident for orchard performance measures (Fig. 6b). Case study 2: vineyard Accumulated irrigation The irrigation amount for both the west and east subplots during the base year (2017) was 122 mm (Fig. 7). The irrigation amounts for the east UI subplot were 47.3 mm during 2018 while the average irrigation of the west VRDI subplot was 47.9 mm. During 2018, the MC that received more irrigation was located in the north-east corner of the west subplot. Addi- tionally, the cells located in the middle row of the west subplot received less irrigation than the other cells. 13 Precision Agriculture Stage I t1 (2018) to (2017) End End Avg. = -0.95 Avg. = -0.97 Avg. = -1.01 Avg. = -0.75 SD = 0.16 SD = 0.10 SD = 0.07 SD = 0.06 CV = 0.171 CV = 0.104 CV = 0.072 CV = 0.077 a West East b Stage III t1 (2018) to (2017) End End Avg. = -1.15 Avg. = -1.33 Avg. = -1.23 Avg. = -1.43 SD = 0.19 SD = 0.20 SD = 0.22 SD = 0.22 CV = 0.163 CV = 0.153 c CV = 0.179 CV = 0.152 d Fig. 8  Measured stem water potential (SWP) (MPa) per management cell (MC) of Mevo Beitar vineyard at end dates of stages I and III during years 2017 and 2018 Normalized relative comparison index The VRDI management decreased the variability of SWP during both stages I and III in the vineyard (Fig. 8). The NRCI values at the end dates were both less than one indicating that the SWP variability decreased by the end of these stages. However, the NRCI values of the yield, WP, ° Brix, and cluster weight variability indicated that VRDI was not able to reduce the variability of these parameters (Table 5). Furthermore, the vineyard NRCI val- ues were all equal to or slightly below one indicating that VRDI did not improve the yield, WP, ° Brix, or cluster weight (0.94, 1.04, 1.0 and 1.0, respectively). Comparison of percent difference results Differences were noted between percent NRCI difference and traditional percent difference for vineyard variability measures (Fig.  9a). Variability parameters such, as yield, WP, ° Brix, and cluster weight, had lower percent difference values when plot normalization with base year data was incorporated. A nearly negligible difference between beginning and end Stage I SWP measurements was recorded using the traditional percent difference calcula- tion as opposed to the substantial difference between the percent NRCI difference. The SWP calculations of stage III showed a similar trend. Little or no difference was noted between percent NRCI difference and traditional percent difference for vineyard yield, WP, ° Brix and cluster weight (Fig. 9b). 13 Precision Agriculture 200% 2018 NRCI - 1 167% 2018 %Diff 150% 123% 100% 82% 78% 74% 63% % Difference 53% 56% 50% 19% 7% 3% 7% 5% 4% 0% 0% 0% 0% -6% -9% -6%-3% -4% -4% -50% -44% -100% Yield (CV) WP (CV) SWP - SWP - SWP - SWP - Brix (CV) Cluster Yield (avg.) WP (avg.) Brix (avg.) Cluster Stage I Stage I End Stage III Stage III weight (CV) weight Beginning (CV) Beginning End (CV) (avg.) (CV) (CV) a Variability Measures b Performance Measures Fig. 9  Comparison of percent difference in normalized relative comparison index—% Diff NRCI (NRCI-1) and percent difference—%Diff ((VRA-UA)/UA) for 2018 in measures of variability (a) and measures of performance (b) in the Mevo Beitar vineyard Discussion The developed STNR methodology builds upon and improves the current approaches of VRA evaluation in comparison to UA (Table 6). The STNR methodology integrates spatial base year data from an entire field and enables determination of the contribution of VRA relative to UA. Two subplots of a commercial field can be compared even if they are dis- similar in terms of performance and/or variability measures. These subplots, of interme- diate size, are spatially representative of the field and as a result, the temporal span is a minimum of two years. The methodology was specifically demonstrated with VRDI but is universal by nature and can be applied to other precision application systems of fertilizer, pesticides, seeds as well as other irrigation methods. This methodology built upon the research of Sanchez et  al. (2017) who presented a related subplot approach, where a vineyard was divided into two subplots managed alter- natively by UA or VRA. While they also used a base year to estimate the effect of VRA compared to UA, the yield of the two subplots was interestingly identical, therefore, the base-year data was technically not part of the calculation, as the changes in yield due to VRA could be simply determined. Commercial fields, however, are typically heterogene- ous, making such a comparison difficult. Sanchez et al. (2017) used the mean correlation distance (MCD) parameter introduced by Han et al. (1996) to evaluate the spatial structure of yield and NDVI data. The 2012 base year values for the two subplots in the Sanchez et al. (2017) study were indeed different. Calculated NRCI values based on the data pub- lished by Sanchez et al., 2017 suggest that VRA improved the spatial yield and the normal- ized difference vegetation index (NDVI) NRCI values for years 2013 and 2015, especially for the yield of 2013. In the Sanchez et al. (2017) study, VRA caused impairment during 2014 as both NRCI values of 1.16 and 1.11 for yield as well as NDVI and MCD data were greater than one. The temporal span of subplot experiments is usually a number of years, slightly longer than the minimum two years for the STNR methodology. 13 13 Table 6  Comparison of uniform application (UA) to variable rate application (VRA) approaches. The spatiotemporal normalized ratio (STNR) methodology incorporates data from a year with VRA (­ ti) and prior to VRA implementation (­ t0). VRA* indicates the specific VRA subplot within the field where UA was implemented during the base year ­(t0) Approach Spatial coverage Temporal Span Calculation Base-Year Inclusion Can remote sens- References ing be incorpo- rated? VRAt Spatiotemporal Intermediate size and repre- Minimum 2 years i Yes Yes Current article UAt i Normalized VRA∗t 0 sentative area UAt Ratio 0 per year Methodology Subplot Intermediate size and repre- Few years URA−UA Yes; not part of comparison Yes Sanchez et al., 2017 UA per year sentative area Whole plot Large size and representa- Long-term VRA−UA Yes Yes Yost et al., 2017 UA multi-year average tive area Strip Usually small size and non- Few years Comparison of treatment No Usually not fea- Vellidis et al., 2016, representative area means sible Liakos et al., 2020, Essau et al. 2014 Precision Agriculture Precision Agriculture The temporal span of the whole plot approach is long-term and hence attempts to deal with temporal variability. Temporal variability is common (Sadler et  al., 2005; Sanchez et al., 2017; Yost et al., 2017). Differences between years are primarily the result of varying meteorological conditions (Jiang et al., 2020), biological growth cycles with higher yields followed by lower yields (Aggelopoulou et al., 2010), and due to changes in commercial agro-technical practices. A long-term temporal span enables complete coverage of the field and is significantly more spatially representative relative to other approaches. However, the whole plot approach is not practical for most research settings nor easily implemented at the field scale. The opposite is true of the common strip approach, which is short-term and generally spatially non-representative. The complexity of defining a methodology that can adequately evaluate the improve- ment or impairment of VRA on a field is highlighted by the tradeoff between spatial cov- erage and temporal span. The STNR methodology attempts to address both aspects and provides a short-term spatially representative approach. This requires normalization of the field with base year data and assumes that the base year is representative. A season with an acute underlying problem should not be used as a base year. Only one base year was included in this study and between one to two years of demonstration of the methodology. Contemporary agricultural field research rarely extends beyond three years because of high costs and limited budgets. However, to fully understand and adapt VRA management and the STNR methodology presented here, longer-term experimental studies would be benefi- cial in order to draw specific conclusions about the case studies. Despite these limitations, the STNR methodology provides an improvement over current approaches of VRA evalua- tion and should be further tested. To highlight the importance of normalizing the VRA/UA ratio of any year (ti) with the base-year ratio to calculate the relative contribution of VRA, it is important to assess the differences between the percent difference results with and without incorporating the NRCI. In the orchard during 2018, the percent difference of SWP at the beginning and end of stage III without incorporating NRCI was − 34% and − 35%, respectively, suggest- ing that no effective change occurred during this time span. However, during this stage the most significant VRDI was implemented. Conversely, the percent difference results incor- porating the NRCI at the beginning and end of this stage (− 6% and − 70%) indicated that in fact VRDI management was not only implemented during this stage, but that it had a significant impact on the decrease of in-field SWP variability. Similar to the peach orchard, VRDI management was able to decrease the in-field variability of SWP both stages I and III in the vineyard as evident in the comparison of beginning and end percent dif- ference results. The difference in trend between the beginning and end of stage I and to a slightly lesser degree at stage III. The percent difference was very similar between the beginning and end of stage I (3% and − 6%). The percent difference results incorporating NRCI, however, showed a substantial difference between the beginning and end of this stage (78%, − 44%) during which VRI was implemented. This reinforces the importance of incorporating the NRCI. These results can perhaps be explained by the total pre-season precipitation amounts of 369 mm in 2017 and 452 mm in 2018. The precipitation during 2017 was less than the yearly average of 537 mm, and this, coupled with the regulated defi- cit irrigation strategy likely caused higher variability of vineyard water status (SWP). This, in turn, this may have been the cause for the low SWP end NRCI values. The reduction of orchard variability enables streamlining of agro-technical procedures such as fruit and foli- age thinning and improved efficiency of field inputs such as water and nutrients (Colaço & Molin, 2017). In the orchard, VRDI management decreased the yield and WP variability for both 2018 and 2019 seasons as indicated by the NRCI results. Additionally, the percent 13 Precision Agriculture difference results incorporating NRCI were all lower than the traditional percent difference results. In the vineyard, the VRDI management was unable to decrease the variability of the additional variability parameters, yield, WP, ° Brix, and cluster weight. This was evi- dent both from the NRCI values and in the comparison of percent different values. The NRCI results indicated that VRDI was not able to improve performance in the orchard nor in the vineyard. Additionally, the percent difference results (with and with- out the incorporation of NRCI) were very similar in both case studies indicating that the base-year normalization had little impact on these parameters. This was most likely the result of several factors. Most significantly is that the vineyard yield potential is defined during the previous season, in this case during the UA base year (2017). VRDI was there- fore not expected to affect the average yield or cluster weight during 2018. Unlike most of the results from the current study, yield has been demonstrated to be improved as a result of VRA in projects for both perennial crops (Colaço & Molin, 2017; Sanchez et al., 2017) and field crops (Yang et  al., 2001). VRA has also been shown to improve resource-use- efficiency (Colaço & Molin, 2017). The orchard NRCI yield performance in both the 2018 and 2019 seasons highlighted some of the limitations of this research, especially the challenges in isolating variables while conducting research in commercial fields as opposed to designated research farms. An explanation for the unusually low peach yield during 2018 is thought to be due to: (1) incorrect application of dormancy breaking agent following the winter and (2) extensive foliage growth during 2017, which may have curbed the number of buds. The 2019 results showed an improvement in yield and WP in the north subplot with VRA. However, during the 2019 season, the north subplot fruit was hand-thinned approximately three weeks later than the south subplot due to limitations in human resources. Both subplots were hand- thinned during growth stage II and, according to the grower, the thinning instructions were identical. Seemingly as a result, 62% more fruit was harvested in the north than in the south subplot. Therefore, it cannot be concluded that the increase in peach yield and WP was purely the result of the VRA strategy. Precision application of water, nutrients and pesticides is gaining acceptance in both applied research communities and, to some degree, among growers (Lowenberg-Deboer & Erickson, 2019; Robertson et al., 2012) albeit mostly in field crops. Widening the scope of research to explore how VRA affects additional fruit tree and field crops would strengthen the STNR methodology. Particularly, the incorporation of crops with different target objec- tives (yield, quality, resource efficiency) would add value to such research. Furthermore, it would be valuable to ascertain how VRA affects NDVI and CWSI and how the NRCI can evaluate such parameters. Bellvert et al. (2020) indeed used NDVI data in order to compare regional vineyard spatial variability between years with UA and VRA management. Such spatially high-resolution parameters can potentially be significant in determining the rela- tive effect of VRA over both space and time. The relative ease by which remotely sensed parameters produce high temporal resolution data, suggests that NRCI could be calculated throughout a growing season, possibly improving both the variability and performance measures. Additionally, although the proposed approach is based on commercial-scale sub- plots using the subplot approach, the STNR methodology can potentially be integrated into and enhance the currently prevalent strip approach of experiment setups. The incorpora- tion of a designated UA “base year” area adjacent to the experimental strips with similar characteristics would enable NRCI calibration and could prove valuable for future applied research, especially if there are initial differences between the strips. Lastly, the NRCI ena- bles comparison of parameters of different orders of magnitude (e.g., yield, SWP, NDVI, MCD). Therefore, the index can be used to explore the contribution of VRA at a regional 13 Precision Agriculture scale to map where the precision agriculture community has succeeded and concentrate where efforts are needed. Conclusion A new spatiotemporal normalized ratio (STNR) methodology was presented to compare VRA to UA at the field-scale. The methodology includes an experimental design setup, sampling scheme of relevant parameters, a defined normalized relative comparison index (NRCI), and means of interpretation of the results. The NRCI quantitatively calculates the effect of VRA by comparing VRA subplot data with base-year UA data for any relevant parameter. This enables the normalization of meteorological changes that occur from year to year, which affect average yield and WP, as well as variability patterns of parameters of interest. The interpretation of the index results is based on the specific target goals per field. When NRCI equals 1, VRA offers no added value. Variability measures with NRCI values lower or higher than 1 indicate that VRA management was able to decrease or increase variability, respectively. Performance measures with NRCI values lower or higher than 1 indicate VRA management impairment or improvement of the yield, respectively. The methodology can be incorporated into any precision resource application system. In the current study, it was applied to data sets from a peach orchard and a vineyard to quan- tify the contribution of VRDI management. In the peach orchard, the NRCI indicated that VRDI was able to reduce the variability of peach yield, WP, and the SWP of stage III for both 2018 and 2019 seasons, although the mean yield and WP for 2018 were not improved. Similarly, in the vineyard, SWP variability decreased at the end of stages I and III. How- ever, VRDI management did not reduce the variability of other vineyard parameters such as yield, WP, Brix, and cluster weight. The performance measures for the vineyard were either unchanged or negatively affected by VRDI management. Future research directions can include using the methodology to evaluate how VRA affects remotely-sensed NDVI and CWSI parameters at the field scale and additionally, the incorporation of a base-year UA area and NRCI index to common strip experiments. The precision application toolbox could benefit from the STNR methodology and the NRCI index enabling advancement in decision-making capabilities, ultimately optimizing inputs and reducing the environmental footprint of resources. Supplementary Information The online version contains supplementary material available at https://​doi.​ org/​10.​1007/​s11119-​022-​09877-4. Acknowledgements  The authors would like to thank the peach grower, Shlomo Cohen, for collaborating and allowing the research to be conducted in his orchard; Reshef Elmakais, Tomer Hagai, Shai Levi, Suli- man Farhat, Omer Levi, Ishai Gilad, Ohad Masad, and Shlomi Kfir for field measurements and technical support; Datamap company for imagery acquisition and mosaicking. Additionally, the authors would like to thank the team of Carmel Wineries, Avi Yehuda and Dror Dotan for their collaboration and assistance at Mevo Beitar vineyard and particularly thank Ben Hazut, Matan Golomb and Doron Kleimann for assisting in the field measurements. The authors would also like to thank the anonymous reviewers of the manuscript for their constructive comments. Funding  This research is a part of The “Eugene Kendel” Project for Development of Precision Drip Irriga- tion funded via the Ministry of Agriculture and Rural Development in Israel (Grant No. 20–12-0030). The project has also received funding from the European Union’s Horizon 2020 research and innovation pro- gramme under Project SHui, Grant Agreement No. 773903. 13 Precision Agriculture Declarations  Conflict of interest  The authors declare that they have no conflict of interest. Ethical approval This paper is an expansion of the ECPA 2021 conference proceedings full paper enti- tled “Methodology for comparison between uniform and variable rate application in a drip-irrigated peach orchard”. References Aggelopoulou, K. D., Nanos, G. D., & Gemtos, T. A. (2010). Spatial and temporal variability of yield and fruit quality in apples. Acta Horticulturae, 877, 731–738. Bellvert, J., Mata, M., Vallverdú, X., Paris, C., & Marsal, J. (2020). Optimizing precision irrigation of a vineyard to improve water use efficiency and profitability by using a decision-oriented vine water con- sumption model. Precision Agriculture. https://​doi.​org/​10.​1007/​s11119-​020-​09718-2 Boyer, J. S. (1995) Chapter 3: Thermocouple Psychrometry Measuring the Water Status of Plants and Soil. Academic Press Inc., USA. 495: 49–102. Castellarin, S. D., Matthews, M. A., Di Gaspero, G., & Gambetta, G. A. (2007). Water deficits accelerate ripening and induce changes in gene expression regulating flavonoid biosynthesis in grape berries. Planta, 227(1), 101–112. https://​doi.​org/​10.​1007/​s00425-​007-​0598-8 Cohen, Y., Alchanatis, V., Saranga, Y., Rosenberg, O., & Sela, E. (2017). Mapping water status based on aerial thermal imagery: Comparison of methodologies for upscaling from a single leaf to commercial fields. Precision Agriculture, 18, 801–822. https://​doi.​org/​10.​1007/​s11119-​016-​9484-3 Colaço, A. F., & Molin, J. P. (2017). Variable rate fertilization in citrus: A long term study. Precision Agri- culture, 18(2), 169–191. https://​doi.​org/​10.​1007/​s11119-​016-​9454-9 Cordero, E., Longchamps, L., Khosla, R., & Sacco, D. (2019). Spatial management strategies for nitrogen in maize production based on soil and crop data. Science of the Total Environment, 697, 133854. https://​ doi.​org/​10.​1016/j.​scito​tenv.​2019.​133854 Dag, A.; Alchanatis, V.; Zipori, I.; Sprinstin, M.; Cohen, A.; Maravi, T.; et  al. (2015). Automated detec- tion of malfunctions in drip-irrigation systems using thermal remote sensing in vineyards and olive orchards. In Proceedings of the 10th European Conference on Precision Agriculture, 12–23. Dammer, K. H., Thöle, H., Volk, T., & Hau, B. (2009). Variable-rate fungicide spraying in real time by combining a plant cover sensor and a decision support system. Precision Agriculture, 10(5), 431–442. https://​doi.​org/​10.​1007/​s11119-​008-​9088-7 Esau, T. J., Zaman, Q. U., Chang, Y. K., Schumann, A. W., Percival, D. C., & Farooque, A. A. (2014). Spot- application of fungicide for wild blueberry using an automated prototype variable rate sprayer. Preci- sion Agriculture, 15(2), 147–161. https://​doi.​org/​10.​1007/​s11119-​013-​9319-4 Idso, S. B., Jackson, R. D., Pinter, P. J., Reginato, R. J., & Hatfield, J. L. (1981). Normalizing the stress- degree-day parameter for environmental variability. Agricultural Meteorology, 24, 45–55. Jackson, R. D., Idso, S. B., Reginato, R. J., & Pinter, P. J. (1981). Canopy temperature as a crop water stress indicator. In Water Resources Research, 17(4), 1133. Jiang, H., Hu, H., Wang, S., Ying, Y., & Lin, T. (2020). Understanding the impact of sub-seasonal meteoro- logical variability on corn yield in the US Corn Belt. Science of the Total Environment, 724, 138235. https://​doi.​org/​10.​1016/j.​scito​tenv.​2020.​138235 Jones, H., & Sirault, X. (2014). Scaling of thermal images at different spatial resolution: the mixed pixel problem. Agronomy, 4(3), 380–396. https://​doi.​org/​10.​3390/​agron​omy40​30380 Kennedy, B. Y. J. (2002). Understanding grape berry development. Practical Winery and Vineyard, August, 1–5. Kerry, R., Oliver, M. A., & Frogbrook, Z. L. (2010). Geostatistical applications for precision agriculture. Geostatistical Applications for Precision Agriculture. https://​doi.​org/​10.​1007/​978-​90-​481-​9133-8 Liakos, V., Smith, E., Fountas, S., Nanos, G., Kalfountzos, D., & Gemtos, T. (2020). On-Farm evaluation of variable rate fertilizer applications using yield-based mathematical formulae in a greek apple orchard. International Journal of Fruit Science, 20(sup2), 1–18. https://​doi.​org/​10.​1080/​15538​362.​2019.​17021​ 35 Lowenberg-Deboer, J., & Erickson, B. (2019). Setting the record straight on precision agriculture adoption. Agronomy Journal, 111(4), 1552–1569. https://​doi.​org/​10.​2134/​agron​j2018.​12.​0779 13 Precision Agriculture Ma, B. L., Wu, T. Y., & Shang, J. (2014). On-farm comparison of variable rates of nitrogen with uniform application to maize on canopy reflectance, soil nitrate, and grain yield. Journal of Plant Nutrition and Soil Science, 177(2), 216–226. https://​doi.​org/​10.​1002/​jpln.​20120​0338 Meron, M., Tsipris, J., Orlov, V., Alchanatis, V., & Cohen, Y. (2010). Crop water stress mapping for site- specific irrigation by thermal imagery and artificial reference surfaces. Precision Agriculture, 11(2), 148–162. https://​doi.​org/​10.​1007/​s11119-​009-​9153-x Minasny, B., & McBratney, A. B. (2006). A conditioned Latin hypercube method for sampling in the pres- ence of ancillary information. Computers and Geosciences, 32(9), 1378–1388. https://​doi.​org/​10.​ 1016/j.​cageo.​2005.​12.​009 Munitz, S., Netzer, Y., & Schwartz, A. (2017). Sustained and regulated deficit irrigation of field-grown Mer- lot grapevines. Australian Journal of Grape and Wine Research, 23(1), 87–94. https://​doi.​org/​10.​1111/​ ajgw.​12241 Munitz, S., Schwartz, A., & Netzer, Y. (2019). Water consumption, crop coefficient and leaf area relations of a Vitis vinifera cv. “Cabernet Sauvignon” vineyard. Agricultural Water Management, 219, 86–94. https://​doi.​org/​10.​1016/j.​agwat.​2019.​03.​051 Munitz, S., Schwartz, A., & Netzer, Y. (2020). Effect of timing of irrigation initiation on vegetative growth, physiology and yield parameters in Cabernet Sauvignon grapevines. Australian Journal of Grape and Wine Research, 26(3), 220–232. https://​doi.​org/​10.​1111/​ajgw.​12435 Naor, A. (2006). Irrigation scheduling of peach - Deficit irrigation at different phenological stages and water stress assessment. Acta Horticulturae, 713, 339–349. Naor, A., Gal, Y., & Peres, M. (2006). The inherent variability of water stress indicators in apple, nectarine and pear orchards, and the validity of a leaf-selection procedure for water potential measurements. Irri- gation Science, 24, 129–135. https://​doi.​org/​10.​1007/​s00271-​005-​0016-6 Netzer, Y., Yao, C., Shenker, M., Bravdo, B. A., & Schwartz, A. (2009). Water use and the development of seasonal crop coefficients for Superior Seedless grapevines trained to an open-gable trellis system. Irrigation Science, 27(2), 109–120. https://​doi.​org/​10.​1007/​s00271-​008-​0124-1 Netzer, Y., Munitz, S., Shtein, I., & Schwartz, A. (2019). Structural memory in grapevines: Early season water availability affects late season drought stress severity. European Journal of Agronomy, 105, 96–103. https://​doi.​org/​10.​1016/j.​eja.​2019.​02.​008 O’Shaughnessy, S. A., Evett, S. R., & Colaizzi, P. D. (2015). Dynamic prescription maps for site-specific variable rate irrigation of cotton. Agricultural Water Management, 159, 123–138. https://​doi.​org/​10.​ 1016/j.​agwat.​2015.​06.​001 Ohana-Levi, N., Derumigny, A., Peeters, A., Ben-Gal, A., Bahat, I., Katz, L., et  al. (2021). A multifunc- tional matching algorithm for sample design in agricultural plots. Computers and Electronics in Agri- culture, 187, 106262. https://​doi.​org/​10.​1016/j.​compag.​2021.​106262 Robertson, M. J., Llewellyn, R. S., Mandel, R., Lawes, R., Bramley, R. G. V., Swift, L., et al. (2012). Adop- tion of variable rate fertiliser application in the Australian grains industry: Status, issues and prospects. Precision Agriculture, 13(2), 181–199. https://​doi.​org/​10.​1007/​s11119-​011-​9236-3 Rud, R., Cohen, Y., Alchanatis, V., Levi, A., Brikman, R., Shenderey, C., et al. (2014). Crop water stress index derived from multi-year ground and aerial thermal images as an indicator of potato water status. Precision Agriculture, 15(3), 273–289. https://​doi.​org/​10.​1007/​s11119-​014-​9351-z Sadler, E. J., Evans, D. E., Gerwig, B. K., Millen, J. A., Thomas, W., & Fussell, P. (2005). Severity, extent and persistence of spatial yield variation in production fields in the SE US Coastal Plain. Precision Agriculture, 6(4), 379–398. https://​doi.​org/​10.​1007/​s11119-​005-​2416-2 Sanchez, L. A., Sams, B., Alsina, M. M., Hinds, N., Klein, L. J., & Dokoozlian, N. (2017) Improving vine- yard water use efficiency and yield with variable rate irrigation in California. In Proceedings of the 11th European Conference on Precision Agriculture., 8(2): 574–577. https://​doi.​org/​10.​1017/​s2040​ 47001​70007​72 Stamatiadis, S., Schepers, J. S., Evangelou, E., Tsadilas, C., Glampedakis, A., Glampedakis, M., et  al. (2018). Variable-rate nitrogen fertilization of winter wheat under high spatial resolution. Precision Agriculture, 19(3), 570–587. https://​doi.​org/​10.​1007/​s11119-​017-​9540-7 Vellidis, G., Liakos, V., Porter, W., Tucker, M., & Liang, X. (2016). A Dynamic Variable Rate Irrigation Control System. In Proceedings of the 13th International Conference on Precision Agriculture, 13: 1–9. Yang, C., Everitt, J. H., & Bradford, J. M. (2001). Comparison of uniform and variable rate nitrogen and phosphorus fertilizer application for grain sorghum. Transactions of the American Society of Agricul- tural Engineers, 44(2), 201–209. https://​doi.​org/​10.​13031/​2013.​4676 Yost, M. A., Kitchen, N. R., Sudduth, K. A., Sadler, E. J., Drummond, S. T., & Volkmann, M. R. (2017). Long-term impact of a precision agriculture system on grain crop production. Precision Agriculture, 18(5), 823–842. https://​doi.​org/​10.​1007/​s11119-​016-​9490-5 13 Precision Agriculture Zaman, Q. U., Esau, T. J., Schumann, A. W., Percival, D. C., Chang, Y. K., Read, S. M., et al. (2011). Devel- opment of prototype automated variable rate sprayer for real-time spot-application of agrochemicals in wild blueberry fields. Computers and Electronics in Agriculture, 76(2), 175–182. https://​doi.​org/​10.​ 1016/j.​compag.​2011.​01.​014 Publisher’s Note  Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Authors and Affiliations L. Katz1,2,3,4   · A. Ben‑Gal4 · M. I. Litaor3,5 · A. Naor3 · M. Peres6 · I. Bahat1,7 · Y. Netzer8,9 · A. Peeters10 · V. Alchanatis1 · Y. Cohen1 1 Institute of Agricultural Engineering, Agricultural Research Organization, Volcani Center, Rishon‑LeZion 50250, Israel 2 Department of Soil and Water Sciences, The Robert H. Smith Faculty of Agriculture, Food & Environment, The Hebrew University of Jerusalem, P.O.B. 12, Rehovot 7610001, Israel 3 Department of Precision Agriculture, MIGAL Galilee Research Institute, P.O.B. 831, Kiryat Shmona 11016, Israel 4 Environmental Physics and Irrigation, Agricultural Research Organization, Gilat Research Center, M.P. Negev 85280, Israel 5 Department of Environmental Sciences, Tel Hai College, Upper Galilee 1220800, Israel 6 Northern R&D Center, MIGAL Galilee Research Institute, P.O.B. 831, 11016 Kiryat Shmona, Israel 7 The Robert H. Smith Institute of Plant Sciences and Genetics in Agriculture, The Robert H. Smith Faculty of Agriculture, Food & Environment, The Hebrew University of Jerusalem, Rehovot 7610001, Israel 8 Department of Chemical Engineering, Ariel University, Ariel, Israel 9 Department of Agriculture and Oenology, Eastern R&D Center, Ariel 40700, Israel 10 TerraVision Lab, P.O.B. 225, Midreshet Ben‑Gurion 8499000, Israel 13