Using Unmanned Aerial Vehicle (UAV) imagery to assess rice leaf area index (LAI) under a system of rice intensification (SRI) cultivation in Southern Taiwan.

• Main Authors: Massiafa SAGNON, 馬席法
• Other Authors: Yu-Min Wang
• Format: Others
• Language: en_US
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/4mng67

碩士 === 國立屏東科技大學 === 土壤與水工程國際碩士學位學程 === 107 === The system of rice intensification (SRI), as several agronomic management practices, use less water and enhances rice biophysical parameters such as grain yield and leaf area index (LAI) using terrestrial method to evaluate different amounts of water irrigation supplied. That method is more expensive compared to the aerospace one from remote sensing (RS). Many studies have evaluated the indirect method using satellites, but few have mentioned the use of unmanned aerial vehicles (UAV) which is known to be more effective and inexpensive methods for crop growth status monitoring. The study aims to estimate grain yield and LAI of rice crop (Oryza sativa; variety: Tainan 11) in different water depth during crop growing season under the evaluated plant biophysical parameters, and to appreciate canopy status through the multispectral RS measurements. UAV data were obtained over 20 plots with five treatments of four water depth: 2, 3, 4, and 5 cm based on observed soil cracking and 3 cm per week, and named T2, T3, T4, T5 and T3’, respectively, with four replicates. Three Vegetation Indices (VIs) were analyzed: Green Vegetation Index (GVI), and Normalized Difference Vegetation Index (NDVI) applied to estimate the grain yield, the photosynthetic material quality, and the LAI; whereas the Simple Ratio (SR) were required to inform about LAI density. Results from agronomic and radiometric measurements, indicated that during all crop development, the best correlations between biophysical variables and spectral variables were observed on treatment T3, with R-square = 0.90. In addition, near-infrared (NIR), visible (Red), Green and Red-Edge (RE) wavelength bands were used to perform UAV imageries, and generate VIs indicated. Agronomic parameters and LAI estimated from the whole 120 sample hills of 20 plots were used as the main inputs to generate an efficient statistical SRI yield estimated model, while 24 sample hills from 4 plots of T5 were used for validation. The research challenge was to develop model that can be inverted to extract relevant and reliable information from RS data, providing to users near-real-time information about the SRI.