Traceability through all the stakeholders in food production is an issue of increasing importance, being specifically required by the regulations for food safety and quality (EC 178/2002), and for compliance with environmental protection. The agricultural market perceives a need for systems and technologies to automate the currently manual process of producing records of agrochemical inputs loaded into a spraying machine. A novel prototype Automated Agrochemical Traceability System (AACTS) to identify and weigh agrochemicals as they are loaded into crop sprayer has been designed, constructed, fitted to a machine and evaluated with commercial operators. The functional blocks of the system are a 13.56 MHz RFID reader, 1.4 litre self cleaning weighing funnel mounted on a 3 kg load cell, a user interface with a screen and three user command buttons (Yes, No, Back), and a progress bar made of 8 coloured LED’s (green, amber, red). The system is able to trace individual agrochemical containers, associate the product identity with national agrochemical databases, quantify the required amount of product, assist the sprayer operator and control workflow, generate records of sprayer inputs and interoperate with (recommending extensions to) task management standards as set out in ISO 11783-10. The evaluation of the quantity weighing has demonstrated that with such a system, the principal noise component is in the range of 33–83 Hz, induced by the operating tractor engine. A combined 3 Hz low pass digital filter with a second stage rolling mean of 5 values improves performance to allow a practical resolution of 1 gram (engine switched off) to 3.6 grams (sprayer fully operational) with a response appropriate to suit human reaction time. This is a significant improvement over the ±10 grams of the work of Watts (2004). An experiment with 10 sprayer operators has proved that in the majority of cases (92%) an accuracy equal or better than ±5% is achieved regardless of dispensing speed. The dispensed amounts (100.36% of target) and recorded (100.16%) are in accordance with prescribed values (100%; LSD(5%) 2.166%), where amounts dispensed by manual methods (92.61%) differ significantly from prescribed and recorded value (100%). The AACTS delivers a statistically similar work rate (211.8 s/task) as manual method (201.3 s/task; Δt = 10.5 s/task; LSD(5%) 28.2 s/task) in combined loading and recording cycle. Considering only the loading time (181.2 s/task) of manual method, the difference is 30.6 s/task (LSD(5%) 30.1 s/task). In practice this difference is believed to be marginal compared to the time required to load the water, random external events during the spraying session and in time moving, checking and storing paper records. The integrated weighing funnel concept is another significant improvement over previous work. Using this system, the mean duration of measuring per container for all tasks (34.0 s) is approximately half the time (68.5 s) achieved by Watts (2004). The AACTS was rated to be safer than the manual method regarding operator health and safety and risk of spillage. All operators who evaluated the AACTS were interested in purchasing such a system. The work confirmed that an RFID system was an appropriate media for agrochemical identification performing more than 250 product identification operations during operator tests without failure, with a speed of operation <1 s per cycle and reading distance of 100 mm. A specific format for RFID tag data is proposed for adoption, using low cost tags, that combines item level traceability with identification of products independently without access to worldwide databases. The AACTS follows ISO 11783 task management logic where a job is defined in a prepared electronic task file. It is proposed to extend the ISO 11783-10 task file to integrate the records provided by AACTS by handling the tank loads as individual products resulting from loading task and allocating them to spraying tasks. It is recommended to produce a production prototype following the design methodology, analysis techniques and performance drivers presented in this work and develop the features of user interface and records of tank content into software for ISO 11783-10 cabin task controller to deliver business benefits to the farming industry. The results with RFID encourage the adoption of RFID labelling of agrochemical containers. The reader may wish to read this thesis in parallel with Gasparin (2009) who has considered the business and industry adoption aspects of the AACTS.

Historically, Malawi has depended on rain-fed agricultural systems. It is reported that the frequent droughts and unreliable rainfall since early 1990s have caused many small-scale farmers to turn to the wetlands as alternative sites for crop production. There they use low-cost farming methods and various forms of ‘informal’ irrigation. This study, to better understand the water management practices and the socioeconomic characteristics of the wetland farmers, was carried out in the Shire Valley, at the southern tip of Malawi. This covers about 600,000ha and supports around 250,000 farming families. More than half is wetland, characterized by a network of small streams, rivers, and swamps, and a mosaic of many very small farms separated by bush. Phase I mainly documented the agriculture technologies and socioeconomic characteristics of wetland farming and small-scale informal irrigation systems. 200 farmers and other key informants were interviewed. Phase II aimed to define and measure the benefits of the current systems. The major farming systems groups were identified using cluster analysis and focus group discussions were carried out with 7 to 10 members of each. The results were assessed using gross margin analysis. The results show that flood recession agriculture, river diversion and treadle pumps were the commonest water management technologies among the farmers interviewed. Most preferred flood recession and river diversion to treadle pump, citing capital requirements and running costs as major obstacles. However, the government and NGOs were promoting treadle pump technology (mostly) and river diversion, but not recession agriculture. Motorized pumps, introduced under various schemes, were no longer in use due to farmers’ inability to meet fuel costs and repairs. Farmer access to land was largely under the control of individual farmers who pass on ownership to their children under traditional custom. This finding is contrary to the documented land policy which describes chiefs as custodians of the land. Many farmers viewed group farming as a surrender of their land ownership rights. However, team work was seen to be common in river diversion technologies where a committee was usually chosen to manage a main canal traversing several farms. Even under these circumstances, farmers still preferred to manage their plots individually. The economic analysis showed low farmer-benefits, except where flood recession agriculture was used to grow sweet potatoes, although this receives no attention from government or NGOs. Among the problems were the farmers’ inability to afford inputs, promotion of unsuitable technologies, and government controlled market prices. The study found that the increased wetland use was partly a livelihood diversification strategy linked to droughts and the worsening of the economic situations caused by structural adjustments in the early 1990s. This study encourages government or NGOs to promote the technologies that are acceptable to the farmers and seen to benefit them under the local socioeconomic conditions. Locally, these include flood recession agriculture and small river diversions. Reducing production costs and increasing yields through more efficient water use and improved extension services should be encouraged, and subsidizing input costs and freeing market prices would also help.

The use of the trenchless plough drainage implement has increased in the past few years due to its efficiency and cost advantages over other methods. However, the performance of these machines when working in fields with irregular soil conditions is not yet satisfactory. It is important therefore to study the soil parameters and conditions which could affect the implement behaviour under these circumstances. Therefore, a detailed investigation of the soil reaction forces acting upon a scale model of the trenchless plough was conducted under controlled conditions in a soil laboratory. The model was tested first under restricted conditions of movement, in order to observe and determine all the possible soil reaction forces. The tine, due to its geometric characteristics, was classified as a very narrow tine, and an existing model to predict the soil reaction force acting on the front face of these tines was extended to predict the forces on the sides. Since the length of the failure plane ahead of the tine is often required in the investigation of the soil reaction forces, a mathematical solution based on the Coulomb principle of Passive Earth Pressure was presented to estimate the soil failure pattern. There was good agreement between the values of the angle of the shear plane predicted by this method and the experimental data obtained from the glass sided tank tests. Dynamic tests were conducted with the implement assembled with a long floating beam arrangement assisted by a small link (free-link), used between the hitch-point and the pivoted end of the beam. These tests revealed that, when working over irregular soil conditions a better grade control can be obtained if the hitch-point is kept at constant level in reference to a desired line. In the case where field irregularities persist for long (step inputs), corrections in the hitch-point height might be necessary. These tests show that the implement depth changes in different proportion in relation to the hitch-point height. Where no control is imposed on the hitch-point, the path of the implement is attenuated in relation to the hitch-point position, where better results are obtained for high frequency of the hitch-point. A mathematical solution based on these findings and on the dynamic balance of the forces acting on the system was presented. Since it is an interactive method and requires long and repetitive calculations, a computer programme was developed and used to predict the response of the implement under these uneven conditions. Good agreement between data and estimated values suggested that the method is acceptable.

This research aimed at increasing the understanding of the water resources of the Victoria Nile basin and assessing the impact of irrigation and hydropower developments on the Victoria Nile under different rainfall regimes. A Bayesian Network constructed with the HUGIN expert researcher software version 6.9 was used as the decision tool. The Network used a mixture of data, information from previous studies and consultation with experts/stakeholders. A network consisting of 21 nodes was developed and run to determine the impacts of different development scenarios. The Victoria Nile basin in Uganda is the first recipient of the river Nile flow as it leaves Lake Victoria. In this basin, there is potential for 5 large hydroelectric power plants and the basin consists of 70% of the irrigation potential in Uganda and yet it is one of the most lacking in hydrological data in the Nile basin. Further downstream of this basin are two riparian states, Egypt and Sudan which according to the prevailing legislation on the use of the Nile share amongst themselves the entire river flow. The research shows that Irrigation and hydropower developments have modest effects on lake levels and river flows exiting the basin. Rainfall occurrence on the other hand has the largest effect on the lake levels and Victoria Nile river flow exiting the basin. It is shown that in situations of very high water demand, which occurs when annual rainfall is less than 1,200 mm, full irrigation potential is utilized and all 5 hydroelectric power plants are developed, irrigation water need is not more than 7% of the Nile flow from the basin. The effects of hydropower plants are manifested mainly in the socioeconomic impacts in their vicinity, which are found to be large and to increase with the number of plants developed. The current mode of operation of outflows from Lake Victoria which is based on an international agreement between Uganda and Egypt is a satisfactory means of control only during moderate rainfall events and lake levels. However, for extreme conditions of lake levels outside the range of 10.8-11.6 m it is inadequate under increasing demands of hydroelectric power generation.

Donkeys are becoming increasingly important for draught animal power (DAP) in Zimbabwe and the sub-Saharan region mainly because cattle, the traditional DAP source, have suffered high mortalities in recent droughts. However, there is limited information on the extent of use and potential of donkeys for DAP in Zimbabwe. To rectify the deficiency, two surveys and a series of studies were undertaken. In the first survey, a rapid rural appraisal (RRA) was undertaken to assess the status, role and management practices of DAP in smallholder farming areas in the semi-arid regions of Zimbabwe. The second survey assessed the morphological attributes of 335 working donkeys, the relationships between live weight and body measurements and the effect of seasonal fluctuations of herbage availability on live weight and body condition of the donkeys. This was followed by experiments to measure the draught performance at ploughing and to investigate the effects of water and work on dry matter intake (DMI) by donkeys. The results of the RRA showed that large numbers of cattle died during the 1991-92 drought (up to 75 per cent in some areas) and that this had increased the farmers' dependency on donkeys for DAP. The management of donkeys was generally inadequate. For example, there was widespread use of the inappropriate neck yokes on donkeys in some of the areas. The survey on morphological attributes indicated that the typical "Zimbabwean" donkey weighed 142 kg, had a heart girth of 115 cm, a withers height of 105 cm and was normally light grey in colour. Male and female donkeys were similar in size. The single best predictor of live weight was heart girth (r<SUP>2</SUP> = 0.864) followed by umbilical girth (r<SUP>2</SUP> = 0.753). Donkeys lost weight and body condition during the late dry season. When used for ploughing in the wet season, donkey and cattle teams of similar total team weights exerted a similar draught force (863 N (±49.6) and 912 N (±98.2), P>0.05), generated similar power output (743 W (±68.8) and 938 W (± 140), P>0.05), worked at similar speeds (862 m/s (±69.5) and 1012 m/s (±64.3), P>0.05) and had similar effective field capacities (14.9 hours/ha (± 1.30) and 13.6 hours/ha (± 1.46), P>0.05), respectively.

Tractors incorporate draught, controls for two reasons, to limit variations in implement depth and to limit fluctuations in the load, i.e. the draught force, imposed on the tractor by the implement. Efficient operation of this control is important because many farmers demand a consistent tillage depth and also, if the load is controlled, the tractor can be operated at its maximum output without excessive wheel slip or engine stall. The work for this thesis involved a theoretical investigation of draught control response and its effect on overall tractor performance. A computer model was devised to simulate draught control and its effect on dynamic performance of the tractor-implement combination. Input data for the simulation was provided by laboratory measurements. Two series of field work were carried out to measure response of a linkage force sensing experimental control to sinusoidal and random draught variations and the response of a control which sensed driveline torque. Comparisons between measured and predicted results, confirmed the validity of the theoretical analysis. Suggestions for improving control performance, particularly stability, are made and the likely improvement predicted using the computer simulation. Other possible methods of implement control or improvements to draught controls are discussed in relation to the likely requirements of future tractors.

Rio Grande do Sul is the southern state of Brazil. The Southern part of the Rio Grande do Sul state, an area of approximately the size of England, is part of the Campos sub-region of the Rio de La Plata temperate sub-humid grasslands ecosystem. Beef cattle and the rice crop are the main economic activities in this region. The main goal of this thesis was to simulate the dynamic nature of the farm with the partnership between finishing beef cattle and the rice crop that can be carried out in Rio Grande do Sul, Brazil. To achieve this goal crop, livestock and economic models were developed and integrated to simulate farm conditions in the South of the Rio Grande do Sul state. Models as tools for decision support need to be dynamic in concept as the real farm environment. The information base is classified as "natural" and "simulated". The "natural" results from past experience. The "simulated" is based on quantitative formal scientific information. This work presents a framework that deals with adaptive behaviour as a response as natural and simulated information. Decisions about animals, pasture, soil, land use and economics are incorporated. These decisions impact on the biological and economic models and generate scenarios resulting from these decisions. As the farmer's decisions are sequential and dynamic, the model simulates the bio-socio-economic environment in which farm decisions occur at established time steps. The Farm Integrated Decision Model (FIDM) supplies the farmer with information about economic and biological aspects of the farm and asks the farmer about each decision. Therefore, when farmer takes the decision the "natural" state is a pre-condition of the simulation. Case study simulations were made, and the results of the proposed methodology are presented to demonstrate the potential use of this approach, generating different scenarios for the farmer. Dialogue between extension worker and farmer permits the interactive evaluation of existing technologies. Flexibility in model construction for incorporation of new technologies as and when information comes available.

The viability of the <i>campesino </i>maize-cattle production system of Central Mexico is under stress by the North American Free Trade Agreement policies. To survive <i>Campesino </i>farmers are developing alternative production systems and more efficient uses of their land. The objective of this work was "to develop a "Decision-Support System" (DSS) in order to support <i>Campesino</i> farmers in this process. Two biological models, one socio-economic model and a survey database on the target farming system form the DSS. The CERES-Maize model simulated the yield response of three local land-races of maize to different management systems. The second biological model, a dynamic 'hybrid model', which predicts potential intake, digestion and animal performance of individual dairy cows, was used to simulate alternative feeding systems. A multi-period mathematical programming model integrated the outputs of the previous models with the survey database. This model was used to find the optimal combination of resources and technologies that maximised farmers' income. This model consists of 15,884 structural columns and 602 rows. The DSS was successfully in reproducing the functioning of the main components of the farming system. More importantly it simulated the complex interactions observed between the farmers and their crops and cattle, including traditional maize management practices. The model simulated the resulting effects of these practices on the feeding systems for cattle and on the household's labour demand. The DSS selected on a monthly basis the forage type, concentrate type and supplementation level fed to all classes of cattle on the farm. Moreover, it was able to incorporate the seasonal effects on forage quality and availability.

Although these is considerable uncertainty in the literature about the effects of greenhouse gases on the climate there is little doubt that the composition of the atmosphere has changed in recent times. As a result of human activities the concentration of CO<SUB>2</SUB> is likely to double from pre-industrial levels by the middle of next century and as a consequence, global temperatures are likely to rise. The potential importance of a change in the climate for the environment and agriculture and the complexity of the issues that are involved means that it is important to develop analytical tools to study this problem. The principal aim of the study is to evaluate the possible effects of a change in climate on the pattern, structure and viability of agriculture in Scotland. To address this objective it was necessary to evaluate the effects of climate change at a range of different levels of resolution. A linear programming model was developed that includes a large amount of biological, physical and economic detail. The models of crop growth (grass, grass/clover, swedes, potatoes, barley, vining peas, oilseed rape and wheat) account for variations in weather conditions, soil types, weeds, applications of nitrogen and pesticides, different planting dates and cropping rotations. The livestock operations that are modelled include sheep, dairy and beef fattening enterprises. In addition a considerable amount of effort was devoted to modelling variations in machinery requirements (and the sensitivity of these operations to climate) labour, buildings and finance. The inclusion of this information has allowed the types of adjustments that farmers may implement to be considered. The model is structured as a series of linked sub-problems where the most basic units are farms. In turn the interactions between farmers in terms of flows of intermediate goods, land and labour are considered at the regional and national level. While further developments to the model would allow the status of Scottish agriculture to be more accurately modelled the model in its current stage of development has allowed a realistic evaluation of the effects of climate change to be performed. The results of this study suggest that climate change will have a detrimental effect on Scottish agriculture, however, the effects of climate are likely to vary between the different farm types and regions. In general, cropping farms are likely to benefit from a change while the profitability of livestock farms, and sheep farms in particular, will decline.

Natural resource use planning in a area such as the 'Cerrado' region in the Midwest of Brazil is a challenge. Any planning framework for analysing investment opportunities in such a region must emphasize sustainable development. This implies a shift from the traditional single agricultural expansion objective approach, to a multiple objective approach such as that described in this thesis. The main objective of this study was to develop an appropriate planning framework for quantifying the effects of new crop farm development proposals for the 'Cerrado' region, from a multiple objective perspective. Its practical implementation depends on the following: the generation of specific experimental data on rice, maize, wheat and soybean crops and reliable climate and soils data to calibrate corresponding crop models that are incorporated into a planning framework; the development of experiences concerning environmental quality and social impact analysis by Brazilian researchers; and training of planning framework users for the enlightenment of policy makers. The study was carried out from the point of view of systems theory, and the chosen research approach integrates a set of different modelling techniques within the developed planning framework. The application of a such planning framework for the Paracatu 'Planicie' area was orientated to the final selection (among nine alternatives) of the most appropriate farming system for adoption. However, by manipulating the models which were integrated into the planning framework, and parameters and data related to prices, crop yields, farm sizes, interest rates, credit and others, a range of policy types can be represented and their effects assessed. The results presented illustrate the potential of the planning framework developed here, as a policy analysis tool for farm development. In the conclusions, some insight is provided into aspects concerning the incorporation of this planning framework into a comprehensive decision support system.