ABSTRACT

Precision agriculture may be able to increase food production to help feed an expanding world whilst simultaneously trying to reduce input costs and provide environmental sustainability.

This paper reviews the European research into varying the inputs into cereal production. Direct injection crop sprayers are gaining in popularity and allows the operator to select specific pesticides for specific areas. Grain yield monitoring on combine harvesters along with soil fertility maps allows the farmers to implement site specific crop management.

Precision agriculture is technically feasible as well as being economically and environmentally justifiable.

The authors are:

ANDREW J LANDERS, Head, Dept of Agricultural Engineering, Harper Adams University College, Newport, Shropshire, formerly Senior lecturer at the Royal Agricultural College, Cirencester, Glos. UK and

DAVID STEEL, Land Agent, Smiths Gore, The Kings Lodging, Minster Precincts, Peterborough, former student at the Royal Agricultural College, Cirencester, Glos, UK.

INTRODUCTION

Increased efficiencies in agricultural productivity, via plant breeding, mechanization and management techniques, have all been important factors in maintaining the balance between food supply and world demand. However, essential nutrient and pest control will continue to be the critical link between production of food to meet world needs and long term agricultural sustainability. Coupled with this need to increase output, farmers are coming under increasing pressure to produce food resources in an environmental friendly manner, by the reduction of pesticides and nutrition inputs. Farmers are therefore faced with the problem of trying to increase production to meet world population rises and combat falling world prices whilst, at the same time, trying to reduce inputs to save costs and provide environmental sustainability.

One solution to the problem outlined above is to manage fields as variable units tailoring input levels to as small an area as possible, and no longer treating fields as homogenous production areas. A considerable amount of research work has been completed over recent years on systems to precisely target inputs such as fertilizers and agro-chemicals, according to the localised requirement within the field. These systems, variously described as spatially variable field operations, site specific agriculture, precision agriculture and custom-prescribed operations take into account variations in soil quality, nutrient levels and pests that occur on the majority of arable fields. The whole concept is based around the production of field maps for arable crops and the technology is already well established to produce these. The process starts with the monitoring of yield, which usually takes place in the clean grain elevator of the combine harvester, where the crop is carried into the grain tank. This feature has been available on Massey Ferguson combines since the mid 1980’s although grain weighing has been available as an option for other combine users since the early 1980’s.

The next stage is to produce the map, firstly however, the exact location of the combine has to be recorded. This can be done by various methods, but the method favoured at the moment is via the global precision system (GPS) satellite programme initiated in the 1970’s by the USA Department of Defense. Using the 24 satellites orbiting the earth, the combines location can be pinpointed to within 5 metres accuracy. Yield and location data are stored on a "smart card" which is then downloaded on to the farm office computer to produce yield maps. Spatial variations in crop yields can then be analysed and the reasons for these changes deduced. The ultimate goal of precision agriculture is then to produce "application maps" and with GPS receivers fitted to grain drills, sprayers and fertilizer spreaders inputs can be varied on the move, thus reducing the need to manage land on the "whole field" basis. These developments are only just starting to become reality, and a considerable amount of research work will be needed before they are commercially available.

SPATIAL VARIATION IN CEREAL CROPS

The Background

European Community policy makers are becoming increasingly keen that environmental considerations should play a role in agricultural policy and it is widely accepted that the future shape of the farming industry will depend substantially on the policy decisions taken over the next few years. Open-ended support incentives have encouraged intensive agricultural production and intensive land use. The major cause of the expansion and intensification of agriculture is artificially high and guaranteed output prices. Farmers in the future will have to operate under management systems which will maximise output and reduce target inputs, whilst causing little harm to the surrounding environment. The increasing awareness of environmental pollution, along with worries about pesticide residues on food has resulted in increasing legislation concerning pesticide use. The Governments of Sweden and Norway aim to reduce pesticides by 50 percent states Nordby (1989); the Swedish Government levies a tax on each pesticide treatment. In Denmark the Government aimed at a reduction of 25 percent of active ingredients in pesticides by 1990 and further 25 percent cut before 1997. (Thonke, 1988) In the Netherlands there is a similar move to reduce pesticide use by 50 percent. Conventional approaches to field operations treat a field as uniform with no alternations to drilling rates or fertiliser or agricultural chemical applications. In fact the vast majority of fields are spatially variable in parameters such as soil type, slope, aspect, moisture distribution and fertility. Thus it is thought that field operations should be targeted according to the locally determined requirement. For such precision operations, involving automatic sensing and control systems, we have coined the term spatially variable field operations.

The total arable area in England and Wales is about 4 million hectares and each year some 1.5 million tonnes of nitrogenous fertiliser and 21.6 thousand tonnes of agro-chemicals are applied. The potential savings by applying inputs more precisely and more efficiently are therefore very significant.

The system starts with the production of yield maps which show the high and low yielding areas of the fields. Explanations of these trends are then determined by a field-scale analysis. The yield map information and data from other sources are processed in the farm computer to produce an application map for drilling, fertilizing and spraying.

The causes of spatial variability — soil

Yield differences from year to year are mostly due to climatic changes while the yield variability between farms is attributed mainly to management practices and that between fields is explained by differences in soil conditions (Tits et al., 1989). If soil type is largely responsible for spatial variation, then reference to accurate soil maps will be an essential part of the management process. In extensive field trials, Evans and Catt (1987) worked with winter wheat where the effects of year, location, variety, nitrogen fertilisation and fungicide were compared, and 50 percent of the variance in grain yield was associated with the site, therefore suggesting that soil conditions account for a large part of the total variation.

Variation in soil temperature between years and sites may be a significant cause of yield variation (Gales 1982). During the Spring, direct solar radiation warms the soil but the rise in temperature depends upon the reflection co-efficient, soil water content and the aspect of the slope. In the same year, soils at different sites may be at different temperatures. Some variation in cereal yields could result from variation in the nitrogen nutrition of the crop. Low yields could be caused by deficiency of nitrogen in the soil, or conditions which prevent nitrogen from being absorbed. Current recommendations on the amount of fertilizer to be applied are based on soil type, winter rainfall and previous cropping. These recommendations are believed to be accurate on average, but often they vary for individual fields. The errors may be because factors other than nitrogen limit the yield or because nitrogen reserves of the soils have been estimated inaccurately. It is common practice to sample fields in the UK to determine the mean nutrient reserves available, and to adopt fertilizer practice in accordance with these reserves (MAFF 1988).

The causes of spatial variability — pest and diseases

For the successful production of arable crops, weed and pest control is essential and is generally carried out by the spraying of liquid pesticides. In the majority of arable fields weed patches may account for only a small percentage of the total area, 30 percent for example, that means 70 percent of the field is being sprayed unnecessarily and with herbicides representing on average 40 percent of crop variable costs, there is clearly considerable scope for cutting agro-chemical bills. While there is very little data available on the spatial distribution of weeds, ADAS trials on black grass at Blockworth Experimental Husbandry Farm in 1985 showed that 70 percent of the field area did not justify spraying but 2 percent of the area justified an extra spray. The spatial variability of pests and disease within a field could vary either by type or density. Wilson and Scott (1982) observed over a 6 year period that black grass and wild oats were found in less than 40 percent of the total field area. A conservative estimate of the savings via selective pesticide application might be 35 percent. Not only are weeds non-uniformly distributed across agricultural fields but they also tend to occur in patches which will remain relatively stable in size and location from year to year. For example, Wilson and Brain (1990) reported that over a 10 year period, blackgrass grew in well defined patches on a large commercially operated farm . The patches were found to be stable and there was little evidence to suggest that new patches formed under conventional herbicide treatments. It is clear therefore that the very presence of weeds and pests in a crop cause yield differences. In order that a weed control strategy can be implemented it is necessary to accumulate historical data on weed location, density and species and develop field weed maps, which if stored in digital form on a computer, can be used to selectively apply pesticides to different parts of a field. There are a number of ways in which spatial data on pests can be collected; field walking and satellite images are perhaps the most common.

Other causes of yield variation

The weather in any season, particularly in pre-harvest and drilling stages will significantly alter soil structure and composition, thus affecting yield. The topography of fields also has an effect upon yield, angle, aspect and altitude all have significant influences. The nature and condition of field drainage is also of importance when deciding upon explanations for yield patterns.

Yield measuring and mapping

The primary stage on the "precision farming" concept is an accurate measurement of grain yield within the field. In the UK the Massey Ferguson combine uses a mass flow yield meter. This system measures mass flow in tonnes per hour and per hectare. A small radioactive source is positioned on the elevator leading to the grain tank. Opposite the source is a sensor to record the level of emissions; with no grain flowing the sensor records maximum emissions and when grain flows through the radiation window, low levels are recorded. The difference in radiation count is then translated into grain yield. Yield mapping is the natural progression from accurate yield monitoring. By correlating yield data with information about the combine’s position, a highly accurate yield contour map can be drawn showing the variations on crop performance between the fields or within individual fields. Yield maps should not be used alone to form a sound basis on which to make firm decisions on the variable application of inputs. It forms the start of the decision-making process. Having built up yield data over three or four years and established exactly where the high and low yielding areas of the field are, the reasons for the patterns can be established. The Massey Ferguson combine records the following data every 2.8 seconds whilst combining is in process: longitude, latitude, yield, time, day, east west ground speed, north south ground speed, navigation mode and GPS mode.

Vehicle positioning

The system that appears to be best suited to agricultural requirements in terms of meeting the specification at acceptable cost levels is Global Position System (GPS). The resolution of the satellite signals is down-graded by the US Department of Defense by restricting information published on the transmitted code sequence, this procedure known as selective availability (SA) produces a resolution of about 100m. These errors can largely be neutralized by using two GPS receivers in "differential" mode. One receiver, the base, is placed in a known position whilst the second receiver, termed the mobile, is placed at the unknown position. The static receiver computes its position including errors and then calculates a correction signal based on the known location. The correction signal is transmitted to the mobile by short wave radio signals such that the computer position of the mobile can be corrected to give a more precise position (within 5m). A growing number of farmers in UK are using the Massey Ferguson combine with yield recording and GPS systems. A survey by Steel (1994) revealed some interesting opinions by the farmers using such systems. The greatest future potential use was thought to be variable application of herbicides based on information supplied by yield maps. Other causes of yield variation included soil compaction from field operations, this will probably be one of the first problems to be tackled by sub-soiling. Another farmer acknowledged that a number of years worth of data will be required before fixed patterns within fields will emerge. The two initial variables that were evident from the first yield maps produced were compaction and the shadowing effect of trees of the edges of the fields. It is planned as a result of this to take advantage of the current set-aside option of taking field boundaries out of production. These unsown areas in the future will be used for turning at the end of each pass. Another farmer stressed the importance of keeping the whole concept simple to operate and interpret. The price of equipment must also be significantly reduced if uptake is to be increased. It is thought that the productive capacity of the farm rather than its size will be key factor in determining whether particular farmers will spatially vary application of inputs. Another farmer interviewed stated that his overriding aim in using the GPS yield mapping systems is to ultimately make better and more effective use of the inputs used on the farm, if the potential to increase yield by using more inputs is evident, then this will be undertaken.

Spatially variable field operations — spraying

Thompson et al., (1990) suggests that the amount of herbicides applied could be reduced by targeting weed infested areas or applying a low dose rate to the whole field and normal dose rate to weed patches. It will be necessary to assess the agronomy of patch spraying of weeds over a number of crop cycles and seasons. Gaultney et al., (1989) and Shonk et al., (1991) developed a real-time soil organic matter sensor which used a narrow bank light source and measured reflectance from the soil with a photodiode. Sudduth et al., (1990) developed a prototype sensor using near infra-red reflectance techniques to determine the organic matter content of the soil surface. The application rates of soil acting herbicides must often be increased on high organic matter soils because the cation exchange capacity of the organic matter increases the absorption of a herbicide and decreases its effectiveness. A microprocessor can be developed to interface with the controller of an injection crop sprayer.

Petry and Kuhnbauch (1988) developed a technique for the discrimination and quantitative registration of a weed population. Different spectral information of the plant and soil was used. A video camera was used to deliver separate image signals for the red, green and blue parts of the spectrum which became digitised in a video card.

The advent of direct injection sprayers and computer based information systems will allow farmers to spot treat patches of weed or disease. A direct injection sprayer is essentially a conventional crop sprayer fitted with an injection system which dispenses pesticides at a given rate into the sprayer water pipe line. Landers (1988 & 1992) states the major advantages of injection sprayers are:

1. the reduction in environmental pollution due to the elimination of tank and pipeline washing;
2. a reduction in the operator contamination which occurs with conventional sprayers. Ideally the pesticide would arrive on the farm in large 35 litre containers and be connected directly to the pesticide injection pump, resulting in a closed system;
3. an electronic controller which allows each pesticide pump to deliver each product. The injection pumps can be switched on/off as when required to spray a patch of weeds;
4. the amount of pesticide applied (dose rate) can be adjusted on the move, allowing a higher dose rate to be applied to a high infestation of weeds or disease and vice-versa.

Spatially variable field operations — fertilising

The spatially variable application of fertilizer again requires the development of treatment maps based on expert knowledge historical field information and soil data. Yield maps will also play an invaluable role as a firm base of information from which the specific application maps can be accurately produced. Throughout Europe there are several methods of distribution of liquid or dry fertilisers including broadcasting placement, surface distribution and drilling; spinning disc fertilisers spreaders are perhaps widely used form of distributor.

Spatially variable field operations — drilling

Today more and more main line drill manufacturers are offering electronic monitoring control systems to improve the efficiency of their machines. The two most common new specifications are an infinitely variable seed rate adjustment and on the move seed rate changes. These facilities aim for maximum yield by being able to tune the seed rate to different soil types and conditions within the same field. As with spraying and fertilizers, variable drilling within field will be dependent upon accurate application maps built up from yield data and other sources and reliable location systems.

Goering (1992) states that as technology advances it is likely that fields will be treated on a grid-cell basis with cells maybe as small as 5m square, and applications will be adjusted to each cell. However, it may be that at first farmers will concentrate on large cell sizes to tackle the most noticeable crop yield differences. This will be particularly appropriate in the immediate future as the fertilizer spreaders and large sprayers currently available will not be effective based on small cells such as 5m square.

CONCLUSIONS

Precision agriculture will result in more appropriate use of pesticides and rates being used, with an overall reduction in application rates, thereby satisfying environmentalists, legislators and farmers. Computer aided pesticide application is part of an integrated farming system, besides its use for crop spraying it provides the user with information such as the location of grain drills, fertiliser spreaders and combine harvesters, enabling the cost of production to be reduced and crop yields to be measured.

Clearly the larger farms will have the field sizes and the machinery to derive the most benefits from precision agriculture, therefore uptake levels will be highest amongst this category. Even the small farmers could indirectly use precision agriculture by a co-operative system. In this case a local bureau of information could be stored on a computer from which participating farmers could derive their relevant application maps.

Precision agriculture may also be just as relevant for potatoes, sugar beet and harvesting forage crops. If farmers with a mixed cropping pattern could see the potential benefit, targeting inputs to specific areas, then the market place for precision agriculture could successfully be widened. The question of when precision agriculture will become common place is much harder to try to estimate, it basically hinges on how fast necessary research work is completed when new technology reaches the market place.

The potential benefits of spatial yield data to the rural land manager will also be numerous. Rent negotiations between landlords and tenants could now be backed up with clear evidence of the production potential of the land. With yield mapping now in its third successive year, those farmers who originally took up the system are building up the essential data bank of yield information, and will soon be able to see the occurring patterns of yields. Precision agriculture will allow for the targeting of inputs to specific areas of fields, enabling farmers to remain successful in an increasing but competitive industry.

REFERENCES

Evans, R., and J.A. Catt. 1987. Causes of crop patterns in Eastern England. Journal of Soil Science, 1987, 38, pp 309-324.

Gales, K., 1982. Yield variation of wheat and barley in Britain in relation to crop growth and soil conditions — a review. Journal of The Science of Food and Agriculture. 1983, 34, pp 1085-1104

Gaultney, L.D., J.L. Shonk, D.G. Schulze, G.E. Van Scoyoc. 1989. A soil organic matter sensor for prescription chemical application. In: Land and Water Use: proc of 11^th CIGR conf. Dublin, Sept 1989 (Dodd, V.A., Grace, P.M., eds), pp 3065-3072. Rotterdam: Balkema.

Goering, C. 1992. How much and where. Agricultural engineering, July 1992, pp 13-15

Landers, A.J. 1988. Closed system spraying — the Dose 2000. In; Aspects of Applied Biology, 18, Weed control in cereals and the impact of legislation in pesticide application, pp 361-369

Landers, A.J. 1992. Computers Aided Pesticide Application (CAPA) — a spraying system for the future? Agricultural Engineer, Autumn 1992. pp 68-70

MAFF Ref book 209, 1988. Fertiliser recommendations. London: HMSO.

Nordby, A, 1989. Application and control of the distribution of plant nutrients and pesticides. In: proc 11^th Internat Congress on Agric Eng, CIGR, Dublin, Sept 1989 (Dodd, V.A., Grace, P.M., eds), pp 2073-2080. Rotterdam: Balkema.

Petry, W., W. Kuhbauch. 1988. Messung des Unkrautdfeckungsgrades durch echtfarbentuchtige quantitative Bildanalyses auf Personalcomputern. Kali-Briefe (Buntehof), 19 (4), 311-323.

Shonk, J., L.D. Gaultney, D.G. Schulze, G.E. Van Scoyoc. 1991. Spectroscopic sensing of soil organic matter content. Trans ASAE, 34 (5): 1978-1984.

Steel, D. 1994. Precision agriculture, a viable option for future arable farming? Unpubl. thesis, Royal Agricultural College, Circencester Glos.

Sudduth, K.A., J.W. Hummel, R.C. Funk. 1990. Soil organic matter sensing for precision herbicide application. In: Pesticide formulations and application systems: 10^th vol. ASTNSTP 1078 (Bode, L.E., Hazen, J.L., Chasin, D.G. eds), pp 111-125. Philadelphia: Amer Soc for Testing and Materials.

Tits, M., H. Delcourt, F. Vervaeke, R. Vansichen, J. De-Baerdemaeker 1989. Grain yield maps and related field characteristics. In: proc 11^th Internat Congress on Agric Eng, CIGR, Dublin, Sept 1989 (Dodd, V.A., Grace, P.M., eds), pp 2791-2795. Rotterdam: Balkema.

Thompson, J.F., J.V. Stafford, P.C.H. Miller. 1990. Selective application of herbicides to UK cereal crops. For: Winter meeting 1990, of the American Society of Agricultural Engineers, paper number 901629, St. Joseph, MI: ASAE.

Thonke, K.E., 1988. Research on pesticide use in Denmark to meet political needs. In: aspects of Applied Biology 18, Weed control in cereals and the impact of legislation on pesticide application, pp 327-329. Warwick: Assoc. of Applied Biologists.

Wilson, B.J. and P. Brain. 1990. Weed monitoring on a whole farm — Patchiness and the stability of distribution of Alopecurus Myosuroides over a 10 year period. Proceedings of the European Weed Research Society Symposium, Helsinki. pp 42-45

Wilson, B.J. and J.L. Scott. 1982. Population trends of Avena Fatua and Alopecurus Myosuroides on a commercial arable and dairy farm. Proc. 1982 British Crop Protection Conference — Weeds, Brighton. pp 619-628. Farnham: British Crop Protection Council.