Makara Seri Sains (Nov 2009)
Assessing The Field Hyperspectral Remote Sensing Data To Diagnose Crop Variables In Tropical Irrigated Wetland Rice
Abstract
Measurement of canopy spectral (400 - 980 nm) using ground-based hyperspectral device and the crop variables of ricesuch as leaf area index (LAI), leaf dry weight (LDW) and SPAD values were measured periodically during growthseason with involving three rice cultivars (Pandanwangi, Ciherang and IR Jumbo) and four nitrogen (N) applicationlevels (N0, N80, N92 and N103 kg/ha). This study explored all possible combination wavebands tested inhyperspectral-based vegetation indices (HVIs) and to develop the relationships model between HVIs with cropvariables. Several HVIs used are NDVI (Normalized Difference Vegetation Index), RDVI (Renormalized DifferenceVegetation Index), RVI (Ratio Vegetation Index) and SAVI (Soil Adjusted Vegetation Index). Analysis of pairingwaveband (λ2 > λ1) used in HVIs was investigated with 6,786 combinations to gain optimal waveband. NDVI shownthe highest R2 values for LAI were found in band combinations from green to red region (500 nm to 730 nm).Validation model using FDR implied better accuracy to estimate LAI using whole season data (R2 = 0.856), however,inversely for LDW and SPAD values, validation using reflectance data indicated better accuracy to predict LDW andSPAD values. Model using SAVI denoted the highest values (R2 = 0.852) for predicting LAI Validation of model usingRVI implied the highest values (R2 = 0.797) for predicting LDW. Testing model using SAVI indicated the highest value(R2 = 0.658) for predicting SPAD values. RVI has the best accuracy to validate the model of LAI than that of LDW orSPAD values.
