The significant potential of portfolio data warehouses in revolutionizing agricultural innovation
Using knowledge-bases to improve agricultural project design
In a recent Decision Analysis Initiative1 workshop three topics were identified as areas where there is a need for the dedication of more intellectual effort to improve project design based on the accumulation of relevant and comparable data within a multi-project portfolio. A proposed solution is a portfolio data warehouse (PDW). Although it was emphasized that for PDWs to provide useful knowledge there is a need for a more refined techniques to identify appropriate portfolio datasets, this article will simply describe improvements in useful knowledge that can be achieved with a well-designed portfolio data warehouse. The details concerning design will be covered in a sequel.
Portfolio data warehouses and data warehouses, an important distinction
SEEL concluded as a result of reviews of studies as well as direct staff experience, that attempts to use data warehouses for agricultural policy and project development have not been particularly successful. This is because most such initiatives have been policy-maker led so access to administrative and regulatory data is emphasized and facilitated. However, in practical terms production response data and farm accountancy data is not used enough. Production response data and detailed farm accountancy including gross margin estimates and determinants provide the core information of decision analysis in policy design. Too much of the data in data warehouses is administrative and regulatory which has been collected for diverse objectives. In many countries the very nature of the administrative and policy environment causes what is recorded under administrative and regulatory regimes to be unreliable. Combining this data often compounds errors to a degree of rendering the output of calculations, at most, indicative.
A portfolio data warehouse is based solely on the data collected as a project by project data series. The data is standardised by basing it's specifications on a due diligence design procedure which completes the following analyses:
Locational state elements
Within the constraints analysis section of due diligence procedures, there should be a provision for the determination GPS coordinates for any number of points within a project area during the design phase. Subsequently, dates/times of monitoring observations during implementation phases are recorded. This provides extremely useful information on the project environment and its relationship to project production operations.
This 3D model illustrates the diminishing marginal returns of biomass production to an increase in value of any one factor when the others are fixed. When all factors increase, the biomass production increases to a greater extent. The utility of this information depends upon the physical relationships between GPS data and how location in space-time determines the specific coordinates of EW&T values, thereby determining the expected biomass production.
Starting at the global level the longitude and latitude data define the location on the earth's surface and depending upon whether this location is in the northern or southern hemisphere, the data and time data will provide an indication of the local "season" in terms of spring, summer, autumn or winter as well as the time when any observations and recorded measurements are taken during a specific day.
The observed temperature will depend upon the altitude of the location and time of day. In general terms temperatures fall by around 0.6oC for every 100 metres gain in altitude. Because of this relationship it is possible to use topographic maps (contour maps) to interpolate temperatures from single readings from a meteorological station across a terrain. The diagram below left shows the notional relationship between altitude and relative temperatures of locations at different altitudes.
Water availability (W) is a function of rainfall, evapotranspiration and soil texture which will influence the availability of water remaining in the soil, to plants. The diagrams below provide a short explanation of this impact of soil texture on water availability to plants.
As explained in the box on the right, the availability of water to plants depends on the soil texture. The more water is accessible the easier it is for the plant to absorb nutrients which are essential for transpiration, growth and development of the plant.
The measurement of water availability to plants is calculated on the basis of the difference between evapotranspiration and precipitation as water deficit. The water holding capacity of the soil has a direct impact on water deficit and as explained the holding capacity depends upon the soil texture. The overall impact of this information is that although in the 3D diagram water (W) and fertility (E) are separate axes, water ends up as a composite measure of water deficit, which takes into account soil texture, and fertility is function of:
In terms of sustainable production, rotations and the inclusion of nutrient fixation, for example nitrogen fixation by leguminous crops can be adopted. In many situations in tropical production systems the tradition clearance and production (slash and burn) tends to result in the virgin natural fertility declining as a result of mineral mining until productivity is so low that production moves on to a new area.
Using this information within context of a portfolio data warehouse
Because of the registration of GPS data, a portfolio's data, which combines different projects, is a geographic information system. A portfolio contains records of project designs, project implementation and records of projects that have been completed; all of this information is useful providing a knowledge base whose value increases with time. Different crop genotypes tend to perform differently according to the EWT complex. As data is collected over several production seasons the information on adaptability of genotypes to specific types of EWT complex become clear 3. This information qualifies a vast array of performance reference data which can be organized as performance benchmarks based on actual performance. As a result, the benchmarks used in project design and simulation exercises can become more realistic by making use of already-established evidence and its use as a performance benchmark. Over time the concentration of data will increase leading to an increasingly refined understanding of variance in production by crop type, genotypes and according to seasonal and EWT conditions.
Different executing agencies often have projects in the same geographic zones so the ability to share productivity information associated with EWT data and the locational state indicators can help improve the understanding of the specific local conditions that impact project performance. For example the diagram below shows the locations of projects managed by three different executing agencies. The last synoptic image, on the right, shows all of this data combined in a data warehouse configuration that receives data from each executing agency portfolio database.
The significance of this sort of data combination is that it becomes possible to undertake gradient and transition analyses with greater precision. Gradient and transition analysis involves comparing different project data values such as yields (production) along a gradient of altitudes and temperatures, soil conditions and water deficits. Where projects are in very close proximity comparative analysis is a way to assess accuracy of records bases on proximity of results, where projects are at a distance, the impacts of gradients can be evaluated. These analyses can be used to refine benchmarks so as to improve the knowledge base for project design.
Locational state equivalence
Locational state theory explains the relationships between object properties in the space-time dimension. Data remains directly comparable even although observations are made at different times, sometimes separated by years. This carries an important message that old data is not something to be archived and forgotten about but rather should be used effectively to improve current project design activities. This is why the combination of data sets is so useful.
In order to take advantage of information in data from different executing agency databases it is evident that they need to apply the same due diligence design procedures in order to ensure that all data remains comparable.
Agro-ecological zoning is one of the most practical examples of locational state theory in practice. Agro-ecological zoning has the purpose of assessing the suitability of different locations for different types of agricultural production. Clearly such information is important to ensure that projects do not embark upon the production of a specific crop or cropping systems in a location where conditions are not suitable. There are many excellent examples of the application of agro-ecological zoning in supporting the rationalization of project designs under incentive schemes included in policies. In particular, many good examples have been produced in Brazil to good effect with significant benefits foe the cost of incentive policies. Some examples are provided in the table below.
Standardized datasets and the application of applied locational state theory can have a revolutionary impact of the value of knowledge stored in a portfolio data warehouse. Some of the examples of the economic benefits of agro-ecological zoning provided in the table below serve to emphasize the practicality of the concept.
It is very apparent that there is a considerable amount of change necessary in project cycle and portfolio management services supporting international funding operations of agricultural production, research and innovation initiatives. The returns on investment are potentially high as a result of better project identification and design and the likelihood of financial success of projects as a result of reduced exposure to risks resulting from this approach. This approach has a significant role to play in food security operations and yet there are few logical and technical impediments to introducing these changes, it just requires the will to do it.
Reference 1: McNeill, H.W., "The state of the art and the future of decision analysis", DAI Workshop 5-6 May, Sustainable Agricultural Economic Development Session, SEEL, Portsmouth, 2018.
Reference 2: McNeill, H. W., "Some principles of nature", First Decision Analysis Initiative Workshop, August, 2010, GBF, London.
Reference 3: McNeill, H. W., "The Role of Micro-Bio-Climatic Zoning & Genotypic Mapping" - Agricultural Research, Development & Dissemination Series, SEEL-Systems Engineering Economics Lab, Portsmouth, 2009
Reference 4: McNeill, H. W., various articles in BAC-Brazilian Agriculure & Commodities, Hambrook Publishing Company, 1979-1982.