Session
Title: Decision Support Systems
Date: Tue Aug 2, 2016
Time: 3:40 PM - 5:40 PM
Moderator: Bob Stewart
Closing Yield Gaps with GxExM and Precision Agriculture

There are many challenges to be faced by agriculture if the global population of nine billion people projected for 2050 is to be fed and clothed, especially given the effects of changing climate.  A focus on the interactions of genetics x environment x management (GxExM) offers potential for meeting the yield, and environment and economic sustainability goals that are integral to these challenges.  The yield gap –defined as the difference between current farmer yields and potential yields offered by advances of genetics and breeding, addresses all factors affecting yields and also when these factors affect yield during the growing season. A fundamental tenet of precision agriculture is that the dominant factors affecting yield gap are different for each growing region, landscape or within-field location and that there are temporal considerations when addressing these factors.  Precision agriculture is thus a way to close yield gaps using a GxExM approach.  However, understanding and quantifying yield gaps through a GxExM approach requires transdisciplinary teams of agronomists, engineers, soil scientists, geneticists, plant pathologists, entomologists, weed scientists, and human nutritionists to comprehensively evaluate all factors limiting production. Precision agriculture with a broadened participation by these and other disciplines will enable greater synthesis and integration of new knowledge into production systems that can be implemented via precision farming.  Further, aspirational production systems that include all we have learned about farming on the edge of the future, developed by transdisciplinary teams employing our best technology, may accelerate advances of agriculture to better meet the challenges of the future. 

Charles Walthall (speaker)
Length (approx): 20 min
 
EZZone - An Online Tool for Delineating Management Zones

Management zones are a pillar of Precision Agriculture research.  Spatial variability is apparent in all fields, and assessing this variability through measurement devices can lead to better management decisions.  The use of Geographic Information Systems for agricultural management is common, especially with management zones.  Although many algorithms have been produced in research settings, no online software for management zone delineation exists.  This research used a common grouping technique based on minimizing the sum of squares between groups to create an open source tool called EZZone for management zone delineation.  The tool is accessible by anyone at ezzone.pythonanywhere.com, and allows users to upload their data for delineation.  The tool is designed to be user friendly and easily integrate with other GIS software.  EZZone was applied and evaluated on data collected from small to large fields, as well as fields incorporating Variable Rate Technology.  Five fields were chosen from Georgia, USA, the UK and Greece.  The zones were constructed in these fields until the Goodness of Variance fit was above 0.8.  The results for each field are discussed and the strengths and weaknesses of the algorithm are examined.

Camden Lowrance (speaker)
George Vellidis
Univ of Saskatchewan
, 53
Spyros Fountas
Vasileios Liakos
Length (approx): 20 min
 
Delineation of Site-specific Management Zones Using Spatial Principal Components and Cluster Analysis

The delineation of site-specific management zones (MZs) can enable economic use of precision agriculture for more producers. In this process, many variables, including chemical and physical (besides yield data) variables, can be used. After selecting variables, a cluster algorithm like fuzzy c-means is usually applied to define the classes. Selection of variables comprise a difficult issue in cluster analysis because these will often influence cluster determination. The goal of this study was to assess the effectiveness of the variable selection techniques - spatial correlation analysis, principal component analysis (PCA) and multivariate spatial analysis based on Moran's index PCA (MULTISPATI-PCA) - when used with the fuzzy c-means algorithm to generate MZs. The data used in experiments were collected from 2012 to 2014 in two agricultural fields with corn and soybean crops, located in Brazil. The variables selected were used as input for the fuzzy c-means, generating two, three, and four classes. The performance of the three techniques was assessed by applying analysis of variance (ANOVA), variance reduction index, fuzziness performance index, and modified partition entropy index. The delineated MZs were different according to the variable selection approach used along with fuzzy c-means. For the two agricultural fields, it was possible to define two classes with potential yields that showed statistically significant differences. The MULTISPATI-PCA technique resulted in classes with higher internal homogeneity, better performance of the clustering algorithm, the best variance reduction values, ​​and the most viable MZs to be implemented in terms of field operations.

Eduardo Souza (speaker)
Alan Gavioli
Claudio Bazzi
Nelson Betzek
Length (approx): 20 min
 
Analysis of High Yield Condition Using a Rice Yield Predictive Model

Rice production in Japan is facing problems of yield and quality instability owing to recent climate changes and a decline in rice prices, and possible competition with foreign inexpensive rice. Thus, it is becoming more important to stably achieve high yield and quality, while reducing production costs. Various data, including crop growth, farmer’s management styles, yield and quality, has recently become accessible in actual fields using advanced information and communication technologies. Those data can be effectively used to aid farmer’s decision-making on their management. In this study, we built predictive models of brown rice yield (yield) using 85 data sets collected from 2010 to 2015 at 21 paddy fields in Itoshima city, Fukuoka prefecture, Japan. In the paddy fields, rice was cultivated under various environmental conditions and management styles. Support vector machine was applied to build the models to predict three yield classes (low, middle, high) that ranged from 2.98 to 6.17 t ha-1. The models were built using all the combinations of four explanatory variables: number of spikelets, inorganic nitrogen (N) supply from panicle initiation to mid-ripening dates, and average values of sunshine hours and air temperature from heading to mid-ripening dates. A yield predictive model had the highest classification accuracy of 79% when the model selected two variables: the number of spikelets and inorganic N supply. The model identified high-yield conditions ranging from 27,216 to 30,146 m-2 for number of spikelets and from 58.1 to 72.1 kg ha-1 for inorganic N supply. The results indicated that shortage of inorganic N supply was one of the major causes to lower yield. The results also suggested that applying second topdressing was necessary to achieve high yield in the target region.

Yasumaru Hirai (speaker)
Length (approx): 20 min
 
Considering Farmers' Situated Expertise in AgriDSS Development to Fostering Sustainable Farming Practices in Precision Agriculture

Agriculture is facing immense challenges and sustainable intensification has been presented as a way forward where precision agriculture (PA) plays an important role. More sustainable agriculture needs farmers who embrace situated expertise and can handle changing farming systems. Many agricultural decision support systems (AgriDSS) have been developed to support farm management, but the traditional approach to AgriDSS development is mostly based on knowledge transfer. This has resulted in technology being considered an isolated phenomenon, not adapted to farmers’ actual needs or their decision making in practice. This study examined farmers’ use of AgriDSS in relation to their situated expertise and how they manage their fields. The theoretical framework of distributed cognition (DCog) was applied in investigating and analyzing farmers' use of a software tool called CropSAT, developed for calculation of variable rate application (VRA) files for nitrogen (N) fertilization from satellite images. The results revealed that CropSAT could function as a tool supporting decision making and development of situated expertise among farmers, improving their care perspective.

Christina Lundström (speaker)
Jessica Lindblom
Length (approx): 20 min
 
Estimating Environmental Systems Using Iterated Sigma Point Techniques: a Biomass Substrate Hypothetical System

This paper addresses the problem of biomass substrate hypothetical system estimation using sigma points kalman filter (SPKF) methods. Various conventional and state-of-theart state estimation methods are compared for the estimation performance, namely the unscented Kalman filter(UKF), the central difference Kalman filter (CDKF), the square-root unscented Kalman filter (SRUKF), the square-root central difference Kalman filter (SRCDKF), the iterated unscented Kalman filter (IUKF), the iterated central difference Kalman filter (ICDKF), the iterated square root unscented Kalman filter (ISRUKF), the iterated square root central difference Kalman filter (ISRCDKF) through a biomass substrate hypothetical system with two comparative studies in terms of estimation accuracy, convergence and execution times and under constanttime and varying-time parameter constraints. In the first comparative study, the state variables are estimated from noisy measurements, and the various estimation techniques are compared by computing the estimation root mean square error (RMSE) with respect to the noise-free data. In the second comparative study the state variables as well as the model parameters are simultaneously estimated, and the impact of the practical challenges (measurement noise and number of estimated states/parameters) on the performances of the estimation techniques are investigated. The results of both comparative studies reveal that the ISRCDKF method provides a better estimation accuracy than the IUKF, ICDKF and ISRUKF methods; while the IUKF, ICDKF, ISRUKF and ISRCDKF methods provide improved accuracy over the UKF, CDKF, SRUKF and SRCDKF methods. The benefit of the ISRCDKF method lies in its ability to provide accuracy related advantages over other estimation methods since it re-linearizes the measurement equation by iterating an approximate maximum a posteriori estimate around the updated state, instead of relying on the predicted state. The results of the comparative studies show also that, for all the techniques, estimating more model parameters affects the estimation accuracy as well as the convergence of the estimated states and parameters. The ISRCDKF, however, still provides an improved state accuracies than the other techniques even with abrupt changes in estimated states.

Marie-France Destain (speaker)
Ahmed Ben Hamida
Length (approx): 20 min