Rotten Hazelnuts Prediction via Simulation Modeling-A Case Study on the Turkish Hazelnut Sector.

The quality defects of hazelnut fruits comprise changes in morphology and taste, and their intensity mainly depends on seasonal environmental conditions. The strongest off-flavor of hazelnuts is known as rotten defect, whose candidate causal agents are a complex of fungal pathogens, with Diaporthe as the dominant genus. Timely indications on the expected incidence of rotten defect would be essential for buyers to identify areas where hazelnut quality will be superior, other than being useful for farmers to have the timely indications of the risk of pathogens infection. Here, we propose a rotten defect forecasting model, and we apply it in the seven main hazelnut producing municipalities in Turkey. We modulate plant susceptibility to fungal infection according to simulated hazelnut phenology, and we reproduce the key components of the Diaporthe spp. epidemiological cycle via a process-based simulation model. A model sensitivity analysis has been performed under contrasting weather conditions to select most relevant parameters for calibration, which relied on weekly phenological observations and the post-harvest analyses of rotten incidence in the period 2016-2019, conducted in 22 orchards. The rotten simulation model reproduced rotten incidence data in calibration and validation datasets with a mean absolute error below 1.8%. The dataset used for model validation (321 additional sampling locations) has been characterized by large variability of rotten incidence, in turn contributing to decrease the correlation between reference and simulated data (R 2 = 0.4 and 0.21 in West and East Black Sea region, respectively). This denotes the key effect of other environmental and agronomic factors on rotten incidence, which are not yet taken into account by the predictive workflow and will be considered in further improvements. When applied in spatially distributed simulations, the model differentiated the rotten incidence across municipalities, and reproduced the interannual variability of rotten incidence. Our results confirmed that the rotten defect is strictly dependent on precipitation amount and timing, and that plant susceptibility is crucial to trigger fungal infections. Future steps will envisage the application of the rotten simulation model to other hazelnut producing regions, before being operationally used for in-season forecasting activities.

The HADES Yield Prediction System - A Case Study on the Turkish Hazelnut Sector.

Crop yield forecasting activities are essential to support decision making of farmers, private companies and public entities. While standard systems use georeferenced agro-climatic data as input to process-based simulation models, new trends entail the application of machine learning for yield prediction. In this paper we present HADES (HAzelnut yielD forEcaSt), a hazelnut yield prediction system, in which process-based modeling and machine learning techniques are hybridized and applied in Turkey. Official yields in the top hazelnut producing municipalities in 2004-2019 are used as reference data, whereas ground observations of phenology and weather data represent the main HADES inputs. A statistical analysis allows inferring the occurrence and magnitude of biennial bearing in official yields and is used to aid the calibration of a process-based hazelnut simulation model. Then, a Random Forest algorithm is deployed in regression mode using the outputs of the process-based model as predictors, together with information on hazelnut varieties, the presence of alternate bearing in the yield series, and agro-meteorological indicators. HADES predictive ability in calibration and validation was balanced, with relative root mean square error below 20%, and R2 and Nash-Sutcliffe modeling efficiency above 0.7 considering all municipalities together. HADES paves the way for a next-generation yield prediction system, to deliver timely and robust information and enhance the sustainability of the hazelnut sector across the globe.

A process-based model to simulate sugarcane orange rust severity from weather data in Southern Brazil.

Forecasting the severity of plant diseases is an emerging need for farmers and companies to optimize management actions and to predict crop yields. Process-based models are viable tools for this purpose, thanks to their capability to reproduce pathogen epidemiological processes as a function of the variability of agro-environmental conditions. We formalized the key phases of the life cycle of Puccinia kuenhii (W. Krüger) EJ Butler, the causal agent of orange rust on sugarcane, into a new simulation model, called ARISE (Orange Rust Intensity Index). ARISE is composed of generic models of epidemiological processes modulated by partial components of host resistance and was parameterized according to P. kuenhii hydro-thermal requirements. After calibration and evaluation with field data, ARISE was executed on sugarcane areas in Brazil, India and Australia to assess the pathogen suitability in different environments. ARISE performed well in calibration and evaluation, where it accurately matched observations of orange rust severity. It also reproduced a large spatial and temporal variability in simulated areas, confirming that the pathogen suitability is strictly dependent on warm temperatures and high relative air humidity. Further improvements will entail coupling ARISE with a sugarcane growth model to assess yield losses, while further testing the model with field data, using input weather data at a finer resolution to develop a decision support system for sugarcane growers.