研究目的
To predict sugarcane yield based on NDVI and concentration of leaf-tissue nutrients under different straw removal rates, and to compare the efficiency of NDVI data from satellite images versus hyperspectral sensors.
研究成果
NDVI, particularly derived from satellite images, is effective for predicting sugarcane yield under straw removal management. Leaf-tissue N and P concentrations are key parameters in the prediction models. The approach can aid in monitoring yield changes in bioenergy production areas, but further validation with long-term experiments is needed.
研究不足
The study is limited to two specific sites in Brazil with particular soil and climate conditions, which may not be generalizable. The exact straw removal proportions were hard to achieve precisely in the field. Long-term effects of straw removal were not assessed, and the models require validation with long-term data. Atmospheric interference and soil influence on satellite-derived NDVI could affect accuracy.
1：Experimental Design and Method Selection:
A randomized block design with four straw removal rates (0%, 25%, 50%, 100%) and four replications per treatment was used. NDVI was calculated from satellite images (CBERS-4) and hyperspectral sensor (FieldSpec Spectroradiometer) data. Leaf-tissue nutrient concentrations (N, P, K, Ca, S) and sugarcane yield were measured. Stepwise multiple regression was applied to develop prediction models.
2：Sample Selection and Data Sources:
Two sites in S?o Paulo, Brazil (Area 1: Bom Retiro mill, Area 2: Univalem mill) with different soil types and climates were selected. Sugarcane varieties CTC 14 and RB 867515 were planted. Data included NDVI values, nutrient concentrations, and yield measurements from field experiments.
3：List of Experimental Equipment and Materials:
Hyperspectral sensor (FieldSpec Spectroradiometer, ASD–Analytical Spectral Devices Inc.), satellite images (CBERS-4), standard Lambertian plate for calibration, mechanical harvester for straw removal, balance for yield measurement, and laboratory equipment for nutrient analysis.
4：Experimental Procedures and Operational Workflow:
Straw removal treatments were mechanically applied during harvesting. NDVI data were collected using the hyperspectral sensor positioned 1m above canopy on sunny days, and satellite images were obtained and atmospherically corrected. Leaf samples were collected for nutrient analysis, and yield was measured at harvest. Data were analyzed using SAS software for stepwise regression and validation.
5：Data Analysis Methods:
Stepwise multiple regression was used to select variables for yield prediction. Models were evaluated based on coefficient of determination, residual standard error, and Durbin-Watson test. Validation was done with 30% of data using root mean square error (RMSE) and Pearson correlation.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
