Early forecasting corn yield using ground truth data and vegetation health indices in Bulgaria

1 Institute of Soil Science Agrotechnology and Plant Protection “N. Poushkarov”, Sofia, Bulgaria,
2 Center for Satelite Application and Research, NESDIS, NOAA, Washington DC, USA
3 Chapman University, Schmid College of Science&Technology, Orange, California, USA


Kogan F., Z. Popova, R. Singh and P. Alexandrova, 2018. Early forecasting corn yield using ground truth data and vegetation health indices in Bulgaria, Bulg. J. Agric Sci., 24 (Suppl. 2): 57-67

Weather-related maize crop yield losses due to the transition from a planned state to market economy and the increasing climate uncertainties and drought aggravation have been a concern for farmers and policy-makers in Bulgaria since 1990. This paper discusses the possibilities to use operational satellite-based vegetation health (VH) indices for modelling maize crop yield relative to semi-early A1 and late A2 cultivar technology for early warning of drought-related grain losses. The indices were tested in Pleven oblast (Gorni Dabnik) and Burgas oblast (Sadievo) that represent main grain productive regions of North-West and South-East Bulgaria. Correlation and regression analysis were applied to model maize gain yield observed in the experimental fields of Gorni Dabnik and Sadievo from VH indices during 1982-1991. Strong correlations between Pleven maize grain yield relative to semi-early A1 maize varieties and VH indices were found during the critical period of maize development, which starts in May (week 16) and ends in June (week 23) for technology A1B1. For the late cultivar technology A2B1, the critical period of maize starts in June (week 22) and ends much latter in August-Sept (weeks 32 and 41). Relative to Burgas, for corn late cultivar A2, strong correlations of yield deviations from the trend produced by the A2B1 technology dYi with VH indices occur during week 27 and week 28 (July). Several models were constructed where VH indices could serve as independent variables (predictors). Thus, drought-related corn yield losses relative to semi-early and late cultivars could be predicted in Pleven oblast and Burgas oblast in advance of harvest and official grain production statistic is released.

Keywords: Corn yield forecasting, long-term field experiments, satellite – base vegetation health indices, Correlation and Regression analysis

See the article as a PDF.