Visiting the ‘White Elephant’: a Role Model for practical Precision Agriculture

Precision Agriculture (PA) is a concept with a set of dynamically evolving different techniques. PA is not new, but it is only in a few countries that PA is adopted to relevant degrees (AU, USA, CA, EU, BR). In others the adoption of PA is growing. Since 1996, a large mixed farm in Germany has been constantly implementing and developing its’ own solutions in PA across the farm. 

The following case study, written in July 2014 by Armin Werner (Lincoln Agritech, Lincoln/NZ) and Katarina Apostolidi (ZALF, Germany) provides more detail on this very unique example of PA integration and should inspire thinking and discussion. 

Introduction
WIMEX is a German family owned and operated corporation that runs a large farm in the eastern part of Germany.This farm demonstrates one of the most advanced examples in precision agriculture (PA) adoption in Europe, and likely beyond. The operation’s entire information management is software based and all equipment for the arable production is operated as PA-enabled. All tractors, combines, other self propelled equipment and most of the implements ‘understand’ how to read and produce maps - e.g. yield maps and prescription maps for variable rate applications. Own RTK stations allow highly precise positioning and therefore ensuring spatial operation of most machinery.

The family corporation is actually a group of companies. Their main business area is the production of broiler hatching eggs for the European and international markets. The need to control the entire supply chain in order to guarantee the highest quality of the food for the chicken saw WIMEX introduce the arable farming business two decades ago. WIMEX’s own feed mills convert most of the feed raw materials that are produced on their land. The farm is structured as six agricultural production companies (farms), covering around 6,400 ha (arable: 6,000 ha grassland: 400 ha). An additional service company (AGROSAT) they own provides information management, remote sensing and developments of software tools in precision agriculture for the farm as well as for external clients.

Located in the centre of Saxonia-Anhalt, the soil quality of most of the company’s paddocks is rather high. Yet with 470mm annual rainfall in the rain shadows of the Harz Mountains water is a limiting factor most years. Combined with some high variability of soil quality within the fields (mainly texture and potential rooting depth), the yields can vary substantially. Irrigated land covers 25% of the farm land (1,500 ha) with large centre pivots. The number of permanent employees of the company is 31. Grain maize, winter wheat, winter rape and winter barley covers 64% of the cultivated land. In addition, vegetables (700ha), triticale and other crops are also produced. The arable products are channeled as their own produced feed to the chicken farms. Organic carbon and nutrients are returned as manure, but the hatching egg production is a separated operation from the farming business.

General Information Flowon the Farm
The size of the cultivated land and the resulting need for effective management urged WIMEX early on to adopt advanced information technology solutions. This included their use of precision farming techniques for the land cultivation since 1996. In the subsequent years the farm experimented with yield mapping, adopted some GPS soil sampling procedures, as well as remote sensing techniques. In 1999 management zones were specified on most of the paddocks of the farm. In 2005, they developed their own fertilizer models through AGROSAT and these were made commercially available with AgroSense Smart software to manage the zones on PDAs with GPS support (smartphones were not available at that time). The owner ran the company with “management by numbers” and either bought or developed information management systems for the whole farm.

A complex farm management information system (FMIS) supports the farmland operation including smooth information flows within the farms to managers and staff as well as to suppliers and retailers. As such, the FMIS also supports plant protection decisions and records all treatments. Seeding and fertilizing information as well as application maps are fed into the system. The system is mostly developed in-house and allows information sharing among the employees now with smartphones, Pads/tablet-PC or the office PC.

The farm utilizes relevant information and their smooth flow through self-owned weather stations, a network of soil moisture sensors in the irrigated paddocks, ISO-Bus on the majority of the machines, combine harvesters with GPS based yield mapping, geo-referenced satellite images, ECa-mapped soil properties, auto guidance at most larger tractors, telemetric solutions of expensive equipment, management zones of almost all 600+ fields and variable rate application of fertilizers, pesticides and seeds.

Knowledge management and decision making on WIMEX farms
The general strategy of the corporation is to achieve the best possible economic returns from the arable land and maintain the product quality at the same time. This does not mean that the farmer is looking only for the lowest cost. Dominating the production are long term goals to achieve highest profits, high efficiency of resource use in the production and effectiveness of the farm management. In order to achieve these goals, farm staff continuously tap into a large number of own databases and continuously nourish their own knowledge sources.

WIMEX farms participate in a regional farmer’s consultation group that provides support to a number of farms within a specific agricultural area.

WIMEX utilize the concept of management zones for site specific cropping decisions. WIMEX distinguished a level of up to five different management zones in each field. These zones are static, in the sense that have been defined ten years ago and have barely changed ever since. For the definition of management zones, information from geo-referenced satellite images, information from soil analysis, and yield measurements were taken into account. Zone delineation is revisited on a regular base.

A WIMEX farm manager utilizes management zones as well as information from all the knowledge sources mentioned earlier, in order to define a cultivation strategy for each specific zone which includes a strategy for seeding, fertilization, N-fertilization and plant protection. Additional information currently available at WIMEX is soil moisture and weather data. Technologies that are already in use on all fields are high precise guidance and auto-steering, variable pesticide spraying and fertilization with minimum overlapping. Variable rate irrigation of the centre pivots is gradually introduced. 

In WIMEX, the cropping decisions are made prior to each season for each specific field and within-field zone separately. The individual farm manager has access to all historic information of that field and takes into account information concerning soil points[1], soil texture, fertilization history and yield potential for each field (fig. 3). He then estimates the optimal plant density (seed density, row distance). Next, he estimates the required fertilizers for the expected yield of a management zone. Depending on the case, he also decides on applications of growth regulators in small grains. During the year he uses satellite images in order to decide on optimum fungicide use in all types of crop. Finally, he always evaluates his cropping decisions with the data from yield mapping, after the harvesting period.

armin image1

Fig. 1 (left): Image of a field with the 5 different management zones in different colours (source: WIMEX).

In figure 1, the different management zones for a field are indicated by the colours blue, light blue, green, yellow, and red. The lower yielding area is blue, the higher yielding red.

armin image 2
Fig. 2 (left): Yield, Costs and Revenue per soil point group for a winter wheat cultivation, differentiated into soil quality units (BP-classes) (source: WIMEX).

The site specific treatment has brought significant benefits to the WIMEX farm. Some data from winter wheat may serve as an example. Dates are derived from experiments conducted by the farmer, being produced either with site specific treatment (for fertilizers and fungicide) or as a blanket operation. The results are shown in the figure 2 and are differentiated per soil quality unit.

Only for the lowest soil quality (soil rating of 0-30 points, BP), the yield from VRA treatment (29 dt/ha = 2.9 t/ha) was actually lower than the yield without VRA. The costs of non-VRA were also lower per hectare. Thus the revenue without the use of VRA techniques was larger than the one with the use of VRA techniques. Nevertheless, the revenue for soil points 31-40 was almost the same. For the better soil areas of this paddock (soil points more than 41), higher revenues were achieved inthose areas where use of VRA techniques took place.

In figure 3 a screenshot from the farm management information system shows what is provided to support the farm manager to his decision making of N-fertilization. The manager has information on the zone size, the yield per hectare and per zone in absolute and relative values, the N-fertilization in kilos per hectare and the final yield per zone and hectare for the last year. Such data are available in the decision making process for the last 10 years.

The farm manager then has to decide on the N-fertilization for this year and eventually the current situation. For WiMex farm manager, the final decision will be made by taking into account all the above data but, supported by response functions developed on their own data. Final informationwill be collected by inspecting the field itself.

The value of 50 dt / ha (= 5t/ha) of expected yield is based on the farm manager’s own assessment, using the available data. It applies for the certain field, taking into account all historical data (i.e. previous yield, fertilizer base, supply status, water balance). This location is defined by soil type S and a number of fields of 32 and a humus content of 3.3. For all the values ​​for this location results in an expected average yield of 50 dt / ha. Therefore 50 dt / ha for this field is assigned to 100%. The overall impact is thus a likely yield of 5 tons per hectare. In another field, the division can also be quite different. It depends on the composition of the variable soils. These determine the yield level of each zone, and thus the expected return of the field. The percentage classification is independent of the crop type.

armin image 3

Fig. 3 (left): Screenshot from the WIMEX farm management system. For a specific field, per zone, the following information shows a wheat crop (in columns from left to right): a. zone area (ha and % of paddock); b. expected yield distribution (dt/ha = 0.1 t/ha and % of expected average yield); c. N fertilization kg/ha; d. actual yield (dt/ha = 0.1 t/ha)

Future Plans for WIMEX Farms and new Technologies
Since the establishment of the business in the 1990, the WIMEX farm considers managing the information from the land based production along the entire value added chain to be of crucial importance for the future of the farm. WIMEX, through this effort, is trying to minimize costs but, most importantly, to minimize risks concerning the feed quality (contaminants, hygiene) for its hatching eggs production.

WIMEX, a farm with significant experience in different software and systems, has acknowledged that there will always be the dilemma between developing their own software and acquiring software from the market. When WIMEX decides to buy software, this should cover the specific business needs of the farm and has to be able to be properly customized. WIMEX knows that in the farming sector, and for a fast evolving company, the requirements change continuously. WIMEX expects the technology to be user-driven, and not design driven. And expects its’ adoption to be easy and fast.

The choice of in-house development and the identified need to always be platform independent resulted in seeking open source, open software, open data formats and open standards as platforms to develop on. This facilitates migration and interoperability issues and also keeps the company, although highly dependent on technology, independent of a specific software vendor. Specialized employees deal with technology and the FMIS. And, as always in in-house development cases, there is the risk of quick and dirty approaches to solve specific problems. The lack of proper software structure and documentation and the increased complexity are the challenges that WIMEX has to address, concerning in-house development. On the other hand, the changing business and production requirements can only be addressed with the flexibility that in-house development offers.

The information gathered inside this FMIS over the last years is already huge and is expected to grow rapidly. New information will stem from different sources and systems. Therefore, it is of importance for WIMEX to be able to handle such data from different sources. Service Oriented Architecture solutions are expected to be adopted in helping WIMEX to do so. The handling of these data will also require intense computing power. Currently WIMEX is finishing the conversion of their farm information management system into “Cloud computing” techniques. This allows a centralized computing infrastructure. Some of these solutions will be offered for interested clients of AGRO SAT.

The future challenge for WIMEX will be to establish an (almost) closed cycle of nutrients and organic carbon from biomass. Although the farm does not consider oil resources will become scarce within the next few years, phosphorus is expected to do so. Therefore, WIMEX intends to use technology in order to manage, control and optimise matter and nutrient cycles using ICT across business areas. Operating supplies, machinery, sensors and process knowledge will need to be integrated. For this they also operate a large biomass fermenter, converting the chicken manure into energy and high quality (liquid) nutrients. Some of the biogas is used for heat production on the farm and the rest is sold to the close mid-sized town community administration.

In order to acquire process knowledge, WIMEX is already doing In-Farm-Research trying to simulate the site specific soil-crop-climate resources and their interactions. Such knowledge will allow the farm to make the optimal decisions concerning variable rate application and of course have the maximum benefit also in changing climatic conditions.

As a next step, automated exchange of information between the farm databases and the supply chain will be developed by WIMEX. On a trial base some of the retailers have already restricted access to the FMIS. This enables these retailers to check in the database how a batch of vegetables they may buy had been produced (what, how much, when fertilized, pesticides etc).

Summary
As with white elephants, the special way the WIMEX farms use (and develop) PA is extremely rare and hardly found elsewhere. But this is clearly an example that shows that for more than 15 years the stringent development of information based production of a farm is possible and that PA provides the key concepts and technologies for that.

The WIMEX-farmer is always interested in professional discussions, has participated from early on in research projects on his farm and continues to do so. This has helped him to understand, but also ask for new knowledge, new technologies and new production concepts. His constant interest in looking beyond the daily work allows him to identify potential innovations for his farms very early. To get the best out of promising innovations he does not wait until they are established. Instead he intensively tests them, learns and adapts by applying them to his farms. The result is a continuous innovation process on his farms. He does not just follow trends. For the main part, he is ahead of them.

The owner of WIMEX runs the five farms through competent managers. After a decade of setting up the farms on site himself, he now mostly lives and works 500km away from the farm at the corporate headquarters. This is jointly operated by family members. As he has easy access to all data and can extract key information on the actual situation of the farms any time ‘using his fingertips’, he spends most of his time developing new businesses.

Comment
This example of the WIMEX farms also points to a message we try to convey on the adoption of PA on farms in New Zealand (and everywhere else): Precision Agriculture of a farm (or a production system of a farm) will not be ONE technology that was just tried intensively by researchers or another farm. PA is information based production and works best with locally well adapted systems. But uniquely, PA also provides the flexibility necessary for such a move. When pursued in this way, PA is serious business with many benefits for productivity, profitability, management and ecology.

Implementing PA on a farm needs good attention and integrated thinking in designing what kind of approaches and technologies are the best for a very specific farm. Saying this, PA also gives the farmer a genuine opportunity to develop a solution that is unique and most suitable for their specific farming operation. Successful farmers are good at applying such an adaptable farm management strategy to their farming business. And the solutions resulting from tweaking their very own PA will change over time. Sounds all too familiar? Yes, as PA is pure farmingentrepreneurship.

[1] In Germany a rating system had been developed in the early decades of the last century, covering all farm land with a 50m grid, down to 120cm (“Bodenpunkte” (BP) and “Ackerzahl”). This system was started in Germany (and a few other European countries) as a taxation tool but is heavily used now for evaluating arable land. It takes into account soil type, soil condition and distance to a shallow ground water table. The rating can be from 7 to 100. A level of 7 accounts for hardly cultivable land (pure sand), whereas 100 accounts for the most fertile land in Germany (Black Earth), near Magdeburg. (Sources: )