Full text loading...
-
Evaluation of yield forecast indicators of agricultural crops using aerospace imagery
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 16th International Conference on Geoinformatics - Theoretical and Applied Aspects, May 2017, Volume 2017, p.1 - 5
Abstract
A methodology for construction models of crop yield evaluation using a set of different-time aerospace survey data is proposed. Forecasting is carried out on the basis of constructing linear and nonlinear pair and multiple regression dependences of the yield indicators on spectral characteristics of vegetation and soil – vegetation indices. A peculiarity of the methodology is creation of predictive models for homogeneous subsamples of reference agricultural lands, preliminary formed using Data Mining methods. The use of homogeneous subsamples improves forecast accuracy on average by 5–11 %. Information models are constructed based on the proposed methodology. They made it possible to predict the yield of agricultural lands in the Myronivsky district of the Kyiv region (Ukraine) with accuracy up to 4 % according to 2013 data.