Published in Scientific Papers. Series "Management, Economic Engineering in Agriculture and rural development", Vol. 21 ISSUE 4
Written by Cristian CONSTANTINESCU, Florin SALA

The study aimed to estimate wheat production based on aerial images taken with drone. The wheat crop, Alex cultivar, was fertilized with variable doses of nitrogen, in the range 0 - 250 kg ha-1 N active substance (a.s.). During the vegetation, drone images were taken on the experimental variants, between April and July 2018. The digital images, jpeg format, were analyzed and the values of the RGB parameters were obtained (R-red, B-blue, G-green; RGB colour system). At the time of biological maturity, wheat production was harvested, which recorded values between 1,896.64 kg ha^-1 (V1-control), and 4,787.50 ka ha^-1 (V9). Regression analysis was used to estimate production based on RGB parameters obtained from digital images, taken at four different times. Production prediction (Y_P) was possible in statistical safety conditions (R² = 0.997, p <0.001, images from April 29; R² = 0.993, p <0.001, images from May 13; R² = 0.990, p <0.001, images from 28 May; R2² = 0.968, p <0.001, images from 1 July). 3D and isoquants models were obtained, which expressed the variation of production according to the R and G parameters. RMSEP, as a prediction safety parameter and the F-test showed different levels of accuracy in predicting wheat production based on parameters R and G (RMSEP = 183.5859 for April 29; RMSEP = 330.3418 for May 13; RMSEP = 386.3834 for May 28; RMSEP = 703.9887 for July 1). The use of drones to obtain information about agricultural land is very useful at farm level, and the study can be adapted to different crops.

[Read full article] [Citation]

SPUPOPUPNO1