Abstract

Hitherto, it is essential to ensure sustainable crop production to carry out adequate control of agricultural yields. Remote sensing plays a pivotal role in this context, as it not only enhances the monitoring of crop growth but also optimizes the accuracy of yield predictions. For this reason, in this study has been used remote sensing and GIS techniques to predict red globe variety yield on four field plot trials, carried out on Lambayeque department, (three in Pátapo district and one in Chongoyape district) with drip irrigation in an agricultural area located in the Chacay-Lambayeque Valley at Northern of Peru. To carry out this work, a total of 423 satellite imageries (138 Landsat 8 OLI, 218 Sentinel 2 L2A and 67 Landsat 7 ETM+) were acquired during the study period (from November 1st, 2018 up to December 10th, 2021). During the harvest dates, December and June of each year, the yield data (kg ha−1) per plot were correlated with the dry matter production data (kg ha−1), resulting in a KNN based algorithm capable of predicting yield (r = 0.91; R2 = 0.969; p ≤ 0.05). In order to check the accuracy of red globe grape variety yield estimates, authors analyzed the data collected by satellite imageries and finding that the difference between the predicted and actual yield values was between 0.07 and 6.03%. Results of this study revealed a considerable variation in field productivity, between December (mean value of 10,250 kg/ha) and June (mean value of 8500 kg/ha), as a consequence of the El Niño phenomenon. It was also observed that the relationship between grape yield and the assessed indices (GNDVI, HDMI, and dry matter) exhibited a weak correlation for HDMI, high for GNDVI, and very high for dry matter. In conclusion, it has been demostrated that combining KNN and Correction Algorithm (CA) resulted in more precise yield estimation.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call