Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2013, Vol. 10, No. 4, pp. 245-256
Yield forecasting on the basis of long-term space-based observations of the dynamics of vegetation
A.B. Murynin
1, V.G. Bondur
2, V.Y. Ignatiev
3, K.Yu. Gorokhovskiy
21 Institute for Scientific Research of Aerospace Monitoring “AEROCOSMOS”; Dorodnicyn Computing Centre of RAS, Moscow, Russia; Moscow, Russia
2 Institute for Scientific Research of Aerospace Monitoring “AEROCOSMOS”, Moscow, Russia
3 Institute for Scientific Research of Aerospace Monitoring “AEROCOSMOS”; Moscow Institute of Physics and Technology, Moscow, Russia; Moscow Region, Russia
The method of yield forecasting on the basis of multispectral satellite images is proposed in this article. The results of computer simulation to predict crop yields for different crops of different areas are presented and analyzed. The study is conducted to determine the most accurate model of soil fertility, which results in a model that takes values of vegetation indices for 5 months, starting in March and ending in July months of this year. Based on the analysis concludes that the accuracy of forecasting and ongoing possibility of extending the scope of the resulting model for predicting crop yields.
Keywords: прогнозирование урожайности, множественная линейная регрессия, множественная нелинейная регрессия, вегетационный индекс, yield forecasting, multiple linear regression, multiple nonlinear regression, vegetation indices
Full text