*4.2. Reasons Why the Combination of Information from Sentienel-1 and Sentinel-2 Enhances the Yield Estimation Model*

Previous studies, such as those published by Mercier et al. [100], have utilized data from S1 to predict the phenological stage of wheat. Other investigations have employed the combined information from S1 and S2 for the same purpose [101]. For example, Chaucha et al. [102] used the combined data from both satellites to determine wheat lodging in specific plots. Thus, there are previous studies in which the combined information from S1 and S2 has been utilized to estimate properties that can impact wheat yield or monitor crop development. However, to date, no studies have been identified in the literature that employ the combined information from both satellites to directly estimate wheat yield.

The findings of this study indicate that the utilization of data from both satellites improves the RMSE when compared to results obtained using only data from S2 (Figure 6). Establishing a relationship between wheat grain yield and S1 backscatter is not straightforward as the correlation is not linear, as shown by the performance of MLR (Figure 5). The backscatter is associated with crop canopy and soil roughness, which is related to crop development, LAI, biomass, and grain yield [103]. On the other hand, VIs derived from S2 data are relatively simple to calculate, are not computationally intensive, and are usually related to the biophysical properties of crops, such as greenness and health [104]. However, multicollinearity is a problem when using multiple VIs (Figure 3), as it reduces model accuracy [105]. The analysis of variable contribution showed that, among the top ten most representative variables, variables from both sources of information were present (Figure 8). Despite the unexpected nature of this finding, the variable that demonstrated the greatest contribution in the model was VV\_Day2. This is particularly surprising because VH polarization is usually more sensitive to crop changes than VV [98]. By using data from both S1 and S2 satellite sources together, a more comprehensive understanding of the crop can be obtained, which can lead to more accurate wheat yield predictions. This study demonstrates the potential of using combined S1 and S2 data for crop monitoring and yield prediction and highlights the importance of considering multiple data sources for more accurate crop assessment.
