The prediction of iron contents in orchards using VNIR spectroscopy

Abstract

This study investigated the possibility of determining iron deficiency in orchard leaves using visible/near-infrared methods. Sampling was performed in 24 different healthy orchards (apple, cherry, and peach trees) grouped depending on the severity of iron deficiency. The study was conducted on a total of 120 plant samples including 40 apple, 40 cherry, and 40 peach trees. Spectral reflectance of leaves was measured with a FieldSpec HandHeld spectroradiometer (ASD Inc.) using a plant probe and a leaf clip. Total and active iron contents of leaves in the same samples were determined. Derivative graphs were drawn for the measured spectral curves. Stepwise multiple linear regression analysis was used to model the total and active iron levels selected from spectral reflectance values and derivative curves. Mathematical estimation models with the highest relationship were established. Wavelengths between 540 nm and 560 nm (visible green) and 990 nm and 1010 nm (near infrared) were found to indicate the active iron levels in apple, cherry, and peach trees. The coefficients of accuracy for active iron content were found to be as follows: apple 76.70%, cherry 75.28%, and peach 78.69%. The total iron content was found to be as follows: cherry 63.25%, peach 59.65%, and apple 75.08%. The selected wavelengths produced higher estimation values for the determination of active iron than those for total iron because total iron content affected different regions of the electromagnetic spectrum.