An Overview of Precision Agriculture in the United States of America

Precizní zemědělství ve Spojených státech amerických

Green Cary J., Stewart Mike W., Ping Jianli

Abstrakt

Přehled současných aktivit v oblasti precizního zemědělství ve Spojených státech amerických je shrnut v tomto příspěvku. Precizní zemědělství je hodnoceno na základě aktuálních literárních poznatků předních odborníků. Je diskutován jeho význam při hodnocení půdní úrodnosti, výskytu plevelů, integrované ochrany rostlin, rostlinné produkce a použití závlah. Vložené citace základní literatury umožní získat další podrobnější informace v této oblasti. Význam precizního hospodaření je uveden na příkladu aplikace P a K hnojiv. V závěru práce jsou shrnuty současné výzkumné aktivity autorů v této oblasti.

This paper is intended to provide an overview of current activities in the area of precision agriculture in United States of America. However, due to the requisite brevity of this paper, many details are necessarily omitted. The reader is therefore encouraged to obtain further information from detailed reviews such as those by Pierce and Nowak (1999) and Robert et al. (1996).

Precision agriculture is defined as "The application of technologies and principles to manage spatial and temporal variability associated with all aspects of agricultural production for the purpose of improving crop performance and environmental quality” (Pierce and Nowak, 1999). Following a thorough review of the precision agriculture literature, these authors state that no precision agricultural systems currently exist, rather the current effort represents the application of separate components. Following is a brief overview of precision agriculture activities in the USA.

Current Precision Management Activities

It is generally observed that soil properties vary spatially within agricultural fields. Variable rate application of soil amendments offers a potential opportunity to manage this variability. However, to be feasible, variable application of soil amendments must increase yield to a sufficient degree to pay for increased sampling and application costs. Soil chemical parameters such as phosphorus (P), potassium (K), and soil pH often exhibit lower temporal than spatial variability; this suggests that these soil properties should be more amenable to variable rate application than nitrogen (N), which has higher temporal variability (Pierce and Nowak, 1999). A recent summary article on precision agriculture (Pierce and Nowak, 1999) suggests that due to the relatively high amount of spatial variability and relatively low amount of temporal variability, variable rate lime application may be an appropriate application of precision agricultural technologies, if it can be done cost-effectively.

Precision management of P and K generally involves sampling soils in a systematic way to assess in-field variability and then using conventional fertilizer recommendation philosophies to determine application rates (Hergert et al., 1997). The suitability of conventional recommendation philosophies is still in question (Hergert et al., 1997). Pierce and Nowak (1999) state that, "There appear to be no standards regarding the underlying agronomic principles that should be guiding the development and application of precision agriculture.”

Regarding precision management of N fertilizers, temporal variability is an important consideration. Pan et al. (1997) reported that the temporal variability of N can be larger than the spatial variability. Although precision N management may be more difficult due to this temporal variability, its feasibility is enhanced due to the potential environmental issues associated with usage of N fertilizers. After reviewing the current literature, Pierce and Nowak (1999) report that current precision N management research can be broken down into three approaches. The first approach involves determination of available N levels in grid samples and then interpretation based on traditional recommendation philosophies. The second approach involves determination of application rate based on observed N responses on replicated strips of varying N fertilizer rates. The third approach involves determination of application rate based on observed light reflectance or chlorophyll content.

With regard to precision weed management, it is often observed that weeds are localized within fields. Factors affecting this localization include topography, environment, weed biology, and management practices (Johnson et al., 1995). Johnson et al. (1997) summarize three precision management strategies currently in use. The first approach involves mapping areas of weed infestations and then using GIS to develop the appropriate application protocol. The second approach involves real-time detection of weeds. This detection is facilitated either by optical sensors or image analysis. The third approach involves determining herbicide application rate based on soil chemical and physical properties.

With regard to precision integrated pest management, it is commonly observed that insect populations are both spatially and temporally variable. Fleischer et al. (1997) state that the variability is due to an interaction between four factors: population dynamics, population genetics, biotic environment, and abiotic environment. An approach to supersede the sampling problem is summarized by these authors: sampling on as small a grid pattern as feasible, analyzing data for spatial dependence and looking for consistent patterns. If this process yields unacceptable data, the sampling pattern be modified and the process repeated. They maintain that appropriate sampling strategies can be developed and that, while the costs may be high, appropriate cost-benefit analyses have yet to show whether or not the sampling costs will be prohibited.

Approaches to precision crop management include planting different cultivars within a given field and varying seeding rate within a given field. Bullock et al. (1998) stated that profitability due to precision crop management requires knowledge of the changing relationships between yield and plant density at various locations within a field. These authors suggest that this is quite difficult in practice. Variable rate seeding has been attempted for corn based on topsoil depth (Barnhisel et al., 1996) and for wheat based on landscape position (Fiez and Miller, 1995). After reviewing the current literature, Pierce and Nowak (1999) concluded that while there is potential for successful precision crop management, the early field studies provide no tangible support.

Irrigation systems for precision water management are being developed. Application maps are being developed based on soil and climate data, plant growth sampling and modeling, and irrigation scheduling models (Evans et al., 1996). According to Evans et al. (1996) the delivery technology exists, but the impediment to the successful application of precision irrigation is the inability to successfully interpret data and develop appropriate recommendations. These authors conclude that because water distribution is influenced by many factors (e.g., wind) and that water and N inputs are relatively cheap, "It is probably not economically feasible to use site-specific management only for water and/or N”.

A Case Study

In eastern North Carolina, Heiniger (1998) utilized a large field to compare uniform and variable rate P and K fertility management. Intensive soil sampling on 30 m x 243 m grids was done to determine nutrient levels, pH, and other soil properties. The field was then divided into 16 subunits each with matching soil conditions. Eight of the subunits were randomly selected to receive site-specific applications of P and K, and eight were selected to receive uniform applications.

Where variable rate applications were made the total P applied increased and K decreased compared to uniform applications. The author's explanation for the difference in overall fertilizer needs was based on the distribution of P and K levels. Whenever the average soil test level is greater than the median or mid-point, there will be an increase in the amount of fertilizer applied using site-specific management. This is because the number of samples that test low are overshadowed by a few high to very high samples, thus underestimating the need for nutrients in uniform field management. Conversely, when most of the samples test high, but with a few very low testing samples, the average soil test is low. In this case, site-specific nutrient management will result in less fertilizer used. An economic analysis of the two management schemes was also performed. While variable rate application of P and K was more costly than uniform application, it also produced about 481 kg ha-1 more corn. With variable rate P and K application net profit was increased by $24.37 ha-1 over uniform application.

Factors Affecting Yield of Irrigated Cotton in West Texas

Research has been conducted on two irrigated cotton fields near Lubbock, Texas during 1998 and 1999. Field 1 contained three soil types: Acuff loam (Fine-loamy, mixed, thermic Aridic Paleustolls), Amarillo fine sandy loam (Fine-loamy, mixed, thermic Aridic Paleustalfs) and Olton clay loam (Fine, mixed, thermic Aridic Paleustolls). Both fields had center pivot LEPA systems.

Both fields are under traditional management with uniform applications of nutrients, pesticides, and irrigation water. A grid system of 1.02 hectares (101 by 101 meters) was the basic unit for soil sampling. Sampling positions were georeferenced by means of DGPS.

Soil samples were collected in early summer by compositing three samples within the center of each grid at depths of 0 - 15, 15 - 30, 30 - 61 cm. Soil properties were determined by using standard soil testing methods. Yield data were collected by hand on a 4 m2 area from 2 rows of cotton near the soil sampling points for each grid. Other than sample collection, all activities were conducted by the producers according to their normal management practices.

In both fields and both years, soil nitrate was highly variable (CVs of 0.44 and 0.90 in Field 1, 0.34 and 0.32 in Field 2). Concentrations of Zn, P, Mn, Ca, and Mg were less variable (Ping and Green, 1999; Ping and Green, 2000).

Cotton yield averaged 956 (C.V. = 0.19) and 893 (C.V. = 0.18) kg/ha in Field 1 in 1998 and 1999, and averaged 1085 (C.V. = 0.20) and 1117 (C.V. = 0.15) kg/ha in Field 2 in 1998 and 1999. Growing season rainfall varied greatly between the two years and influenced spatial yield patterns. Furthermore, differences in elevation within fields influenced water distribution during the wetter growing season in 1999. The soil variables related to lint yield were Ca content, Ca saturation, K saturation, depth to HCl reaction layer, depth to caliche layer (petrocalcic horizon), and soil texture.

To further understand the relationships between cotton yield and soil parameters, principal component analysis (PCA) was used. The PCA result for Field 1 in 1998 showed that the first principle component accounted for 51% of the variance among 11 variables which were correlated to lint yield. Principle component analysis indicated that soil texture, Ca content, soil pH, and P content are affecting the lint yield more than other factors. The higher lint yields tended to be associated with soils having higher sand content, higher potassium saturation, greater depth to caliche layer and HCl reaction layer. The presence of caliche can limit rooting depth, thereby potentially reducing yield potential. Factors such as higher levels of Ca and soil pH, and lower levels of potassium are consistent with the presence of caliche. Therefore, their influence on yield may be due to their association with the caliche.

Summary

Precision agriculture research is active in the USA in an effort to increase the production efficiency of agriculture and improve environmental stewardship. Research is addressing issues such as soil fertility, weed science, integrated pest management, crop, and irrigation management. Problems to be addressed include temporal variability, development of appropriate sampling protocols, appropriate recommendation capability, and demonstration of the benefit.