Fig. 3 shows the linear regressions between the estimated and the observed LAI values using the calibration points from RADARSAT-2 and UAVSAR, for both corn and soybeans. The R, RMSE and MAE statistics are provided in Table 6 for all the inversion scenarios. Correlation coefficients of 0.48 (HH–VV), 0.92 (HH–HV), 0.93 (VV–HV) and 0.70 (HH–VV–HV) were obtained for RADARSAT-2 and corn. However, the R values were lower at 0.38 (HH–VV), 0.76 (HH–HV), 0.76 (VV–HV)
and 0.47 (HH–VV–HV) for RADARSAT-2 and soybeans. These results demonstrate that combinations of cross-polarizations and copolarizations (i.e. HH–HV and VV–HV) produce better LAI estimations when compared to the co-polarization option (i.e. HH–VV). This observation is partially explained by a better calibrated HV model when compared to models which use one or both of the co-polarizations (HH or VV (Fig. 2)). For UAVSAR and corn, strong correlations are reported between
estimated and observed LAI with R values of 0.81 (HH–VV), 0.87 (HH–HV), 0.76 (VV–HV) and 0.86 (HH–VV–HV). These results contrast sharply with the results reported for soybeans using L-band with R values of 0.15 (HH–VV), 0.03 (HH–HV), 0.33 (VV–HV), −0.17 (HH–VV–HV).
While L-band may be suited to estimate LAI for corn canopies, this frequency is unlikely to provide accurate estimates for lower biomass crops such as soybeans.
Fig. 3 shows the linear regressions between the estimated and the observed LAI values using the calibration points from RADARSAT-2 and UAVSAR, for both corn and soybeans. The R, RMSE and MAE statistics are provided in Table 6 for all the inversion scenarios. Correlation coefficients of 0.48 (HH–VV), 0.92 (HH–HV), 0.93 (VV–HV) and 0.70 (HH–VV–HV) were obtained for RADARSAT-2 and corn. However, the R values were lower at 0.38 (HH–VV), 0.76 (HH–HV), 0.76 (VV–HV)and 0.47 (HH–VV–HV) for RADARSAT-2 and soybeans. These results demonstrate that combinations of cross-polarizations and copolarizations (i.e. HH–HV and VV–HV) produce better LAI estimations when compared to the co-polarization option (i.e. HH–VV). This observation is partially explained by a better calibrated HV model when compared to models which use one or both of the co-polarizations (HH or VV (Fig. 2)). For UAVSAR and corn, strong correlations are reported betweenestimated and observed LAI with R values of 0.81 (HH–VV), 0.87 (HH–HV), 0.76 (VV–HV) and 0.86 (HH–VV–HV). These results contrast sharply with the results reported for soybeans using L-band with R values of 0.15 (HH–VV), 0.03 (HH–HV), 0.33 (VV–HV), −0.17 (HH–VV–HV). While L-band may be suited to estimate LAI for corn canopies, this frequency is unlikely to provide accurate estimates for lower biomass crops such as soybeans.
การแปล กรุณารอสักครู่..
