Making nitrogen contents model using hyperspectral remote sensing and estimation nitrogens content by nitrogen content model
2005
Ryu, C.(Kyoto Univ. (Japan)) | Suguri, M. | Nishiike, Y. | Umeda, M.
In this research, the effectiveness of hyperspectral remote sensing (AISA+, 400nm-1000nm, 68 bands) is investigated in order to estimate the nitrogen content in the panicle initiation stage, which is necessary to calculate the optimum amount of nitrogen fertilizer for the topdressing. The experimental field is virtually divided by two parts, which are supplied uniformly as 3kgN/10a and variably as 0 kgN/10a-7 kgN/10a with a 1 kgN/10a difference at basal dressing, respectively. The hyperspectral reflectance at the panicle initiation stage was compared with various field data, such as the amount of nitrogen fertilizer, dry weight, and leaf area index (LAI), and nitrogen contents. The correlation between the reflectance and field data at variably supplied parts, dry weight parameters such as plant length, the number of tillers, LAI and dry weight of leaf and stem, have strong relationships with each other. The patterns are also very similar in the whole spectral region except for the concentration group which are SPAD value, with a nitrogen percentage of leaf and stem. It was possible to make a nitrogen content estimation model to predict the nitrogen contents of 15 plots which were supplied with 8 degrees of nitrogen fertilizer doses using multi regression analysis with a forward stepwise regression method. It was also possible to estimate the nitrogen contents of rice plants using the nitrogen content estimation model (R2=0.85, 5% of a significant level), even if there was a low spatial variability of nitrogen content which was supplied with 3 kgN/10a nitrogen fertilizer at basal dressing. The maximum error is 0.97 g/m2.
Mostrar más [+] Menos [-]Palabras clave de AGROVOC
Información bibliográfica
Este registro bibliográfico ha sido proporcionado por Agriculture, Forestry and Fisheries Research Information Technology Center