Modeling corn (Zea mays L.) productivity under variable irrigation and nitrogen regimes using NDVI

Abstract

Abstract Climate change results in rainfall distribution anomalies that trigger drought, resulting in reduced crop yields. Additionally, the scarcity of nitrogen fertilizer is an obstacle in increasing production yields. Therefore, an intensification approach is needed in corn cultivation. In corn cultivation, quick and accurate decision-making is essential, which necessitates the use of technological inputs. One technology that can be utilized is the UAV (Unmanned Aerial Vehicle). UAV can be used to obtain vegetation indices such as NDVI (Normalized Difference Vegetation Index) and canopy cover density (CCD). NDVI development has been extensively studied, but NDVI applications in Indonesia remain limited due to differences in agroecosystems and lack of localized calibration. As a result, NDVI has not yet been utilized in Indonesia. Thus, the development of NDVI is necessary. The objective of this study was to develop a predictive model for corn productivity using NDVI and agronomic traits under varying irrigation intervals and nitrogen doses. This research was conducted from July to October 2024 at the Bajeng Balitsereal Experimental Farm, Pabentengan Village, Bajeng Subdistrict, Gowa District, South Sulawesi. This study was designed using a split-plot design, where irrigation intervals were the main plots with three irrigation intervals (5 days, 10 days, and 15 days), and nitrogen doses were the subplots with five dose levels (0, 100, 150, 200, and 250 N kg/ha). Each treatment combination was replicated three times, resulting in 45 experimental units. Each experimental plot was 16 m 2 in size. Drone imagery was taken at 09:00. Agronomic data will be analyzed using analysis of variance, correlation analysis, and path analysis. After identifying potential agronomic characteristics, these characteristics were analyzed using linear regression and multiple regression on NDVI and canopy cover density based on pixels. Regression results were validated using the coefficient of determination ( R 2 ) and Root Mean Square Error (RMSE). Based on multivariate analysis results, plant height, male flowering age, female flowering age, peel cob weight, and number of seeds per row were identified as the agronomic traits with the greatest influence on productivity. These agronomic traits were then analyzed using linear and multiple regression against NDVI and CCD traits. The productivity (ton/ha) model formulation based on the linear regression approach combined with NDVI (45.55 (NDVI) – 5.15) was considered more effective than other model formulations because it has a high and stable R 2 (train: 0.8555, and test: 0.8543).

Access to Document

Other files and links

• Link to publication in Scopus
• Open Access Version Available

Fingerprint