Corn is the most cultivated and consumed cereal in the world, with a production of more than 1,000 million tons and a harvested area of almost 200 million hectares in 2016 (García-Lara & Serna-Saldivar, 2019). The planting operation is one of the most important tasks undertaken by corn producers. It should result in a plant stand with the desired density that emerges quickly and uniformly. Plant spacing uniformity and emergence rate are the characteristics most commonly used by producers to evaluate planter performance (Staggenborg et al., 2004). Other planting factors influencing corn stand establishment include seed spacing, uniform seed depth, seed quality, planter speed, insects, diseases, desired seeding rates and optimum soil environment for rapid germination and uniform emergence, including soil water and temperature (Lauer & Rankin, 2004); variable-rate seeding, multi-hybrid planting and machinery should also provide high performance under different field and operating conditions (Virk et al., 2019).
Precision Agriculture (PA) only takes into account in-field variability, while Smart Farming (SF) goes further by basing management tasks not only on location but also on data, enhanced by context and situation awareness, triggered by real-time events (Wolfert et al., 2014). Fountas et al. (2006) mentioned that farmers need to think systematically about their information needs and costs, alternative sources, and the value of information by identifying what information needs to be collected before making decisions. In this sense, Saiz-Rubio & Rovira-Más (2020) mentioned a common way to manage field data displayed on maps and culminated with a practical solution using Geographic Information System (GIS) software.
Regarding the indicators for the planting operation, the measurements to control and evaluate in the precision planting are the uniform in-row seed spacing and seed depth. However, many times this control is not performed, or is only partially performed, due to the time required and the tedious work of the measurements. It may also be susceptible to human errors. Concerning seed mapping, it has not been studied and discussed in the research works. This map constitutes a strategic data source to have information on the quantitative and/or qualitative aspects of the planting operation.
This review can contribute to a conceptual synthesis of the methodologies for evaluation and measurement to determine the performance of the precision planter. In this sense, technology can provide tools for monitoring and reporting planting data, making corrections in real time if required, being able to use this information in the context of PA or SF to improve decision-making with the GIS-type software. The main aims of this work were to review: 1) methodologies to evaluate precision planting; 2) technologies for in-row seed spacing measurement system for the precision planter; and 3) the systems that allow to obtain corn ‘seed mapping’ in the planting operation for PA or SF.
Planter performance factors include variability around the target drop points (drop error), seed failure, multiple seeds falling at the same time, seed bouncing and rolling in the furrow, and in-row seed movements when covered with soil (Panning et al., 2000). Singh et al. (2005) mentioned that the planter performance depends on the seed spacing uniformity in the furrows, which is difficult to measure in the field due to soil cover after planting.
These measured seed spacing data and descriptive statistics can be used for planter performance evaluation. These statistical indicators are arithmetic mean, standard deviation (SD), and coefficient of variation (CV). Regarding these indicators, Kachman & Smith (1995) concluded that the use of mean value and SD is not an appropriate summarized method to evaluate precision planting performance.
For corn crop planting, Lázaro et al. (2005) concluded that SD levels below 5 cm did not produce grain yield loss and that the relationship between SD plant spacing and grain yield loss is not linear. On the other hand, a study by Nielsen (2001) evaluated the effect of plant spacing variability, SD treatment from 5.1 to 30.5 cm, on grain crop yield. The results of that field research indicated that about 62 kg/ha are lost for each centimeter increase in the SD of the plant-to-plant spacing. Regarding operating variables in the field, Staggenborg et al. (2004) indicated that corn yield decreased at a rate of 93.7 kg/ha per km/h as the planter speed increased from approximately 7.2 to over 11.3 km/h.
Another methodology for evaluating seed spacing uniformity is the Standard ISO 7256/1 (1984). It is based on the theoretical seed spacing (xref). This value assumes that there are no missing, multiple or variant seeds, and its value is a planter manufacturer's specification. With this value, the intervals for determining the multiple index (D), the missing index (M) and the quality of the feed index (A) are defined. For the calculation of these indicators it is necessary, from each measurement of the seed spacing, to determine to which interval it corresponds. The sum of the three indexes is equal to 100%. The precision (P) indicator of the Standard differs from the usual CV in that it uses the xref as the denominator instead of the sample mean. Another difference is that in the numerator the SD corresponds to the uses of the seed spacing data included in the interval corresponding to the A index. The standard of making at least 250 seed spacing measurements to evaluate planter performance is mentioned.
Another methodology based on xref is called coefficient of precision (CP3) for planter performance. Panning et al. (2000) mentioned that the interval for CP3 is determined only with the seed spacing value within, xref ± 1.5. The CP3 value (in %) can be calculated as the ratio between the number of measurements within the interval and the total number of measurements of seed, or plant, spacing.
A variant of the calculation of the indicators used in the ISO Standard methodology is the one mentioned by Ding et al. (2016), who used the method based on the GBT/6973-2005 Standard (China) but with the ‘standard time’ of the metering seeds instead of xref. The standard time is calculated by an equation using the variables of the seed meter rotation speed and the seed plate holes number of the seed meter.
Some authors have mentioned quantitative and qualitative indicators for the evaluation of precision planting. Gil & Carnasa (1996) considered that any seed placed at a distance of ± 20% of xref will produce a viable plant with a non-negative effect on yield. On the other hand, Weirich Neto et al. (2015) concluded that it is convenient to adopt corn planting as acceptable when the A-index value is above 90%. Kachman & Smith (1995) mentioned that a practical upper limit to the P-indicator value is 29%, which would indicate that the entire seed sp acing is evenly distributed within the target range. However, Yazgi & Degirmencioglu (2014) mentioned qualitative performance criteria based on quantitative indicators to evaluate precision planters. By using D, M and A indexes (in %), it is possible to classify planting as Very Good, Good, Moderate and Insufficient.
Karimi et al. (2015) mentioned that the sticky belt test stand is a method commonly used by researchers as a reference technique to test seed spacing for each planter configuration. The planter unit is placed on the moving belt covered with an adhesive material, usually grease, so that impacted seeds remain in the grease. The belt is then stopped and seed spacing measurements are recorded manually (Alchanatis et al., 2002). Panning et al. (2000) reported that seeds can bounce or roll and that the grease on the belt can minimize this. The authors concluded that laboratory test methods could be useful in the planter with poor seed metering uniformity. The results of laboratory and field tests could be used to improve planter performance.
Different technologies have been developed for in-row seed spacing measurement systems. According to the available technologies, a new conceptual proposal could be mentioned to classify seed spacing measuring systems for the precision planters. The proposal is based on two methods, indirect and direct measurement (
| Measurement strategy | Sensor characteristics | Seed/s | References | |
|---|---|---|---|---|
| Indirect method | Optical-type seed sensors | Infrared (IR) optical-type | pelleted sugar beet |
Kocher et al., 1998 |
| Digital fiber | corn chickpea rape |
Barut & Yiğit, 2008 Hajahmed et al., 2011 Ding et al., 2016 |
||
| Laser | corn | Meng et al., 2016 | ||
| Non-optical-type seed sensors | Line scan camera | cotton | Alchanatis et al., 2002 | |
| Machine vision (image processing) | corn | Li & Lin, 2006 | ||
| High-speed camera system | wheat, soybeans | Karayel et al., 2006 | ||
| Capacitive | corn | Zhou et al., 2012 | ||
| Hall-effect | - | Goldman et al., 2013 | ||
| Piezoelectric | corn | Huang et al., 2013 | ||
| Radio wave-type (microwave) | - | Sauder & Plantamura, 2014 | ||
| Acoustic | corn, pelleted tomato, wheat | Karimi et al., 2015 | ||
| Inductive proximity | - | Tolón Becerra et al., 2016 | ||
| Direct method | Open furrow | Seed-firming wheel (rolling on seed) | part of a tuber, corn, etc. | McCloskey, 2018 |
| High-speed camera (image) | corn | Badua et al., 2019 | ||
| Closed furrow | X-Ray technique | barley, rape, soybeans | Campbell & Baker, 1989 | |
| Ground penetring radar | corn | Mapoka et al., 2018 | ||
The indirect methods consist of measuring seed spacing before the seed reaches the soil, i.e., the furrow. The direct methods consist of measuring seed spacing with the seed located in the furrow. The measurements strategies for the indirect methods could be classified into optical and non-optical seed sensors. While that, the direct methods measurement strategies involve the open or closed furrow.
As for non-optical type seed sensor technology, Karimi et al. (2015) mentioned that acoustic signals are generated by seed impact and evaluate seed spacing. Mapoka et al. (2019) reported that a typical acoustic sensor is composed of a transmitter and a receiver, which contain a piezoelectric material responsible for generating and receiving sound waves at frequencies below 20 kHz, infrasound, and above 20 kHz, ultrasound. Badua et al. (2019) mentioned the technique of overlapping images using common points to create a single image with a field of view to measure seed spacing. Mapoka et al. (2019) related that ground penetrating radar (GPR) is a high-frequency electromagnetic signal and consists of a transmitter and receiver antennas. The reflected signals per seed are received by the GPR receiver as a function of time.
Optical-type seed sensor technology is based on the use of the visible or non-visible electromagnetic spectrum to detect the passage of seeds. Seed monitoring systems with infrared (IR) optical-type sensor are the dominant commercial technology in the electronic equipment for planter monitors. This system allows the planter operator to monitor and diagnose problems during precision planting. Shearer & Pitla (2014) mentioned that there are two technologies most commonly used in seed monitoring systems. One is IR optical-type sensors that detect seed shape and another is microwave seed sensors that detect the seed mass using high-frequency radio waves.
There are three alternative seed delivery concepts for the singular seed meter on the precision planter. As shown in
For optical-type seed sensor, Kumar & Raheman (2018) indicated that IR technology is better due to higher accuracy, small size, lower power consumption, low cost, and simple control of input/output signals. Sauder & Plattner (2006) mentioned that, as for the photoelectric seed sensor, when a seed passes between the light emitting source and the light receiver the light beam is interrupted and detected. The planter monitor processes signals to determine the seeds number or seeding rates, as well as monitors the time between seeds (pulses) to determine seed spacing (Sauder & Plantamura, 2014). Körösi et al. (2019) mentioned that optical-type seed sensors can be classified into two groups according to the control scheme of the sensor light sources. The first group includes seed sensors in which the light sources operate continuously and with a constant light intensity over time. The second group includes seed sensors in which the intensity of the light sources is controlled by periodic signals (
The methodology used to estimate in-row seed spacing with strategic indirect measurement is calculated using the following equation:
where ss: in-row seed spacing, ∆t: the time interval of consecutive seeds measured at the seed sensor, ps: planter travel speed, and e: error. The measuring system consists of a sensor for each seed tube to calculate the seed time interval (∆t) and a sensor to measure the ps. These signals are processed in the planter monitor together with the GPS signal to geo-referenced data. To calculate the error (e) of seed spacing is the actual measured value minus the calculated or estimated value. Goldman et al. (2013) mentioned that the processor, planter monitor, could use a setting based on time parameters to improve the accuracy of seed spacing measurement in the furrow. As for the adjustment parameter, Klenin et al. (1986) mentioned that the time required for the seed to reach the furrow is seed residence time in the seed tube multiplied by an empirical factor (1.05 to 1.15).
Therefore, an alternative to the adjustment parameter is the time (tc) to reach the seed in the furrow given by the free-fall equation. Result of the equation of uniformly accelerated motion, with approx. vo = 0,
where tc: seed falling time between the bottom of the seed delivery device and the furrow (s), H: distance in vertical direction of seed falling (m), and g: gravity acceleration (9.81 m/s2).
For the case of the planter with the seed delivery device, the seed sensor is located at the top (
where: L, distance from the seed sensor to the bottom of the seed delivery device; ds, displacement speed of the belt or brush on the seed delivery device. The term (L/ds) is the travel time for the seed to reach the bottom of the delivery device. The ds (m/s) can be calculated with the equation
where: n, pulley rotational speed of the seed delivery device (rpm); Dp, diameter of the driving pulley (m).
Some limitations have been mentioned in the use of the technology with IR optical-type seed sensors. Kocher et al. (1998) mentioned that seeds with an effective diameter of less than about 3 mm have not consistently blocked enough light beams to trigger the phototransistors reliably. While St Jack et al. (2013) reported that if two seeds fall next to each other, the IR light beam from the sensor would only be cut once and multiple seeds would be incorrectly recorded as a single seed. Liebich et al. (2017) developed a seed sensor in which light is emitted from the diodes transversely to a receiving axis of the sensor receiving unit, bundled through the perforated screens. The light guided into the straight triangular prism-shaped reflector element is deflected by total reflection to form a light band of parallel light beams inside the seed tube with flat intensity. On the contrary, Wilhelmi et al. (2014) developed a seed sensor provided with at least one emitter and several light receivers. The seed size and shape can be inferred from the number of trajectories blocked by the receivers in the array that are cut in each direction and the time it takes for the seed to pass through the seed detection plane in the seed tube.
Steffen et al. (1999) evaluated in the laboratory the influence of two corn seed coatings on the accuracy of a commercial planter monitor, and the authors concluded that one of them undercounted seeds by almost 4.6%. Tevs et al. (2018) mentioned that, during field operation, a layer of dust can accumulate on the sensor optical windows, causing the light from the emitters to refract in unpredictable directions, so that seed counts are doubled or not detected. On this, Kjartanson (2014) indicated that during the planting operation dust is stirred up and this can generate a seed-like signal resulting in a false count. If a puff of dust with 95% transmissivity travels through the IR optical-type seed sensor with a passage time similar to that of a seed the result will be a signal virtually indistinguishable from that produced by a falling seed. To the above, Kostić et al. (2018) mentioned that the sensitivity of the IR optical-type seed sensor should be properly adjusted to avoid erroneous seed reading induced by external light sources, i.e. sunlight or ambient light. On the above problem, a solution by Körösi et al. (2019) proposed an IR optical-type seed sensor whose sensitivity is automatically regulated (
In the case of the seed meter device with seed falls by gravity, Sauder & Plattner (2006) mentioned that factors like sensor holes may act as catch points when the seed drop in the seed tube; manufacturing tolerances of the seed tube or sensor elements and improper installation can affect in-row seed spacing. Deere (2007) used a translucent seed tube that allows the seed sensor to be mounted completely outside of the seed tube, and eliminates the need for seed sensor holes.
Considering the current trend of increasing travel planter speed, in which the rotational speed of the seed meter is increased, it is necessary to consider planting performance. Moody et al. (2003) used two seed sensors in the seed tube, one at the top and another in the bottom, to evaluate seed spacing. These authors concluded that the largest component of seed spacing variability was introduced over the top seed sensor, possibly due to variations in seed exit speed of the meter. In this sense, Virk et al. (2019) mentioned that the results of field data analysis suggested that seed meter performance starts to degrade at higher meter speeds (>35 rpm) indicated by higher CV values (>30%) for seed spacing. On the other hand, Kostić et al. (2018) determined in the laboratory that seed meter rotation speed had the most significant effect on seed spacing deviation, followed by corn seed varieties.
The term microwave is defined as an electromagnetic radiation between radio waves and infrared waves in the electromagnetic spectrum having frequencies between 300 MHz and 300 GHz and wavelengths between 1 m and 1 mm (The American Heritage Dictionary, 2020). About this technology, Shearer & Pitla (2014) mentioned that given the dusty environment of the planting operation, the radio wave-type sensors perform better as they are not prone to dust and can accurately differentiate between single and double seeds. Itagi et al. (2018) reported that sensors with this technology were able to count with accuracy higher than 99% and with high seeding rates over 100 seeds/s.
Tables 2 and 3 show the state of the art in measurement technologies for the indirect method. These could be classified as optical and non-optical seed sensor technologies. Also, the methodologies for evaluating the corn seed planting in the laboratory (Lab) and field tests are mentioned. These results refer only to corn seeds since other seeds were used in some of these investigations. Three groups can be identified to increase the system performance order: (i) quantitative measurement technologies where only seed counts can be determined; (ii) electronic measurement systems (MES) to evaluate the planting performance through the D, M and A indexes, as indicated in the ISO 7256/1 and GBT/6973-2005 standards; these MES are used with different sensors, capacitive, piezoelectric or IR optical-type; and (iii) electronic seed spacing measurement system (SSMES) to measure in-row seed spacing; these use different technologies evaluated in laboratory conditions through a belt test stand with or without adhesive material on the surface to retain the seeds.