Doctor of Philosophy (PhD)
Plant, Environmental Management and Soil Sciences
Nitrogen (N) is one of the most important and limiting nutrients in crop production. The best management practices for N fertilization is always challenging due to its dynamic system in the nature. Remote sensing has emerged as one of the most useful technologies in modern agriculture for non-invasive monitoring of plant N status. The objectives of this research were to 1) determine the effect of water background turbidity and depth on red and red-edge reflectance based prediction models for biomass and grain yield in rice, 2) evaluate agronomic parameters of different sugar cane varieties in response to variable levels of nitrogen supply, and 3) determine the effect of sugarcane varieties on the relationships between spectral reflectance and agronomic parameters. Rice experiments were variety (CL152 and CL261) x N trial established in Crowley, LA in 2011 and 2012. Sugarcane experiments were variety (L 99-226, L 01-283, and HoCP 96-540) x N trial established in St. Gabriel and Jeanerette, LA from 2010 through 2012. Spectral reflectance and agronomic parameters were collected each week for three consecutive weeks beginning two weeks before panicle differentiation in rice and for four consecutive weeks beginning three weeks after N fertilization in sugarcane. There was no significant effect of water background (turbid or clear) on the spectral reflectance at panicle differentiation, one week after panicle differentiation, and at 50 % heading (p <0.05). Water depth slightly influenced the reflectance at red waveband but this effect was not carried over when vegetation indices were computed. Use of red-edge based vegetation indices improved the estimation of biomass and grain yield in rice. The effect of variety on the accuracy of the yield prediction model varied depending on the transformation of reflectance within the red-edge and near infrared bands i.e., into normalized (NDVI) and simple ratio (SR) forms of vegetation indices. This result was associated with the behavior of near infrared wavebands on the geometrical structure of the plant canopy. There were no significant effects of variety on grain yield prediction models using derivative based red-edge indices. Our findings showed that red-edge based NDVI and SR are better predictors of rice grain yield than red-based NDVI and SR. Red-edge based NDVI or SR indices both have potential to predict rice grain yield and rice responsiveness to N fertilization. In sugarcane, the measured agronomic variables at early growth stage, i.e. biomass, tiller number, N content, height and FAI of three sugarcane varieties and their responses to N fertilizer were highly variable across year. The sugar yield response to N determined at harvest had stronger linear relationships with N response of biomass and N content at 4 to 5 weeks after N fertilization compared with N response of height and FAI. There were no differences in leaf spectral reflectance among varieties. In canopy level-spectral reflectance, wavebands at 450-500, 650-700, and 780-830 nm showed high correlation coefficient with agronomic parameters. The vegetation indices which have the potential for predicting biomass N uptake were red and red-edge based simple ratio and normalized difference vegetation index. Varietal effect on the models for estimating biomass and N uptake was significant only when red-based vegetation indices were used (p<0.05). Addition of plant height in the model substantially improved biomass and N uptake estimation while diminishing the effect of variety. Remote sensing technology can be a potential tool to estimate biomass and N uptake in rice and sugarcane. The delivered information from this technology is useful to improve mid-season N management.
Document Availability at the Time of Submission
Release the entire work immediately for access worldwide.
Kanke, Yumiko, "Optimizing yield and crop nitrogen response characterization by integrating spectral reflectance and agronomic properties in sugarcane and rice" (2013). LSU Doctoral Dissertations. 954.