(2)UTAD/CITAB (Universidade de Trás-os-Montes e Alto Douro & Centro de Investigação e Tecnologias Agroambientais e Biológicas)
The “decision support” concept refers to “organised efforts to produce, disseminate and facilitate the use of data and information” to improve decision making. It includes processes, decision support tools and services. Some examples include methods to assess trade-offs between options, future scenarios used to explore the impact of alternative decisions, vulnerability and impact assessments, as well as tools that help people find, organise, and display data in new ways. The results of effective “decision support” processes include building relationships and trust that can support long-term problem-solving capacity between knowledge producers and consumers.
“Decision support” activities that facilitate well-structured decision processes can result from a consensus on the definition of the problems to be addressed, objectives and options to be considered, criteria for evaluation, opportunities, potential consequences and options.
In the specific case of agriculture, the ongoing farm management is currently supported by information on crops, growing conditions, seasonal progress, as well as on the impact of climate change, among others. The quality and timeliness of the information available is a critical factor that determines the quality of the decisions taken and, therefore, the quality of the results - of the crops produced and, ultimately, of the economic profitability. As most of the information is linked to site location, spatial (geographical) relevance is a key feature of the data that can be made available. The relevance and applicability of spatially related and analysed information is one of the main motivations for its use in a decision support system, as many variables that affect crop quality are, inherently, of spatial nature.
Additionally, there are still insuficient technological tools based on spatial data infrastructures in the market, which, in an integrated and standardised way, provide a collection of maps with different information layers (e.g., land drainage, soil, irrigation) comprising a spatial database (Figure 1).
A system of this type allows the exploration of spatial correlations, associating attributes and functionalities that support the analysis of patterns and processes in a given farm. Another capability that stems from the use of a spatial data infrastructure applied to agriculture is the analysis of spatial extents with common characteristics, phenomena or similarities (or specific differences). This includes proximity to other plant species, agricultural zoning, fauna, or other integrated sets of information. A thematic Spatial Data Infrastructure in agriculture promotes and develops automated data analysis with the production of aggregated digital maps (Figure 2), or images, presenting specific reports to the farmer or producer for appropriate decisions and actions. As an illustration, identifying potential agricultural plots of superior quality based on spatial patterns identified from certain areas that have orographic, soil, or other specificities, allows understanding and confirming the contributing factors to enhance quality.
The consolidation of all this information, presented graphically, intuitively and in real-time, will always be very relevant to decision support in mitigating problems.
One of the most critical problems today is the impact of climate change on agriculture. Climate projections should be based on a set of anthropogenic greenhouse gas emission scenarios. These scenarios allow the incorporation of uncertainties related to the different pathways of global socio-economic development in the upcoming decades. Moreover, these scenarios must also be based on different simulations, generated by different climate models (multi-model ensembles), thus allowing the assessment of the uncertainty regarding physical modelling, model initialisations and parameterisations. The accurate assessment and integration of uncertainties are essential for effective decision support. Furthermore, the spatial resolution of the global models (about 100 km) is insufficient to the useful application in agriculture. Therefore, the development of downscaling methodologies is key. These methodologies can be dynamic, by coupling regional climate models (spatial scale of about 10 km) into global models, in both cases physical-mathematical models. Subsequently, this information can be complemented with geostatistical methodologies that allow the reduction of the spatial scale to values of the order of 1 km. When networks of sensors are installed on farms, it is even possible to further reduce the information to lower scales (meters), allowing to addressthe microclimatic patterns of a given plot. In these cases, hourly data or even lower timescales can still be solved.
The use of diverse and extensive geographic information existing at the territory level, supported by the collection of agronomic information -namely on the phenological evolution, levels of water stress that the cultures are subject, degree of maturation and productivity - allow, on the one hand, the adoption of practices that mitigate the effects of climate change in the short term (e.g., application of deficit irrigation), and on the other hand the adoption of long-term mitigation measures (e.g., selecting more climate-resilient crops). Nonetheless, the collection of this type of agronomic and climatic information, at the level of each agricultural plot, located in different microclimatic and edaphic conditions, is not feasible to be carried out by each farmer. Besides, since this information can be extrapolated to a regional level (e.g., for a certain region), being most appropriately used by a certain sector, it can and should be complemented with climate information adapted/modelled to the territory (e.g., incorporating the impact of orography, solar exposure or geological information). The need for more and better information, namely in real-time, using networks of sensors in the field, collecting spatial-temporal data in a wide and integrated way, becomes essential to support effective decision making.