*3.1. Spatiotemporal Dynamics Evaluation over the Mekong Delta by CyGNSS GNSS-R Measurements*

The annual/seasonal dynamics of Γ (i.e., high values in the rainy season from June to October, and low values in the dry season from February to May) could clearly be visualized in the CyGNSS GNSS-R product, as shown in Figure 6. By improving the change detection algorithm by considering the difference in the local incidence angle among each grid cell (i.e., from Γ-normalization to Γ(*θ*)-normalization), two peaks with high Γ values could be detected annually.

**Figure 6.** Temporal dynamics of the Lv. 2 daily product with a 30-day moving average ((**a**): Γnormalized and (**b**): Γ(*θ*)normalized) and the Lv. 3 15-day-cycle Kalman smoother product [(**c**): Γnormalized (500 m resolution with the precision index), (**d**): Γ (3000 km resolution without the precision index) and e: Γ (500 m resolution without the precision index)]. Each line/plot denotes spatially averaged values over the delta.

For the Lv. 2 product, a moving average was required to see the seasonal dynamics. However, this seasonal pattern was clearly illustrated in the Lv. 3 product even if the change detection algorithm was not applied (Figure 6e). Particularly for high incidence angles (>55◦), the non-normalized Γ series shows a wide distribution among relatively high dB values (−20>) during the rainy season. This indicated that our proposed precision index worked adequately as a Kalman smoother weight-mean processing tool and enabled robust spatiotemporal comparisons. Compared with the result that was obtained from the 3000 m grid spacing rasterization result without the precision index (Figure 6d), the 500 m grid spacing rasterization that was enabled by using the precision index displayed the seasonal contrast more clearly (Figure 6e).

The Γ normalization step applied to each incidence angle [i.e., Γ(*θ*)normalized] significantly improved the sensitivity of the results to the temporal dynamics of the incidence angle by increasing the dynamic range [0.2–0.4 for Γ and 0.1–0.7 for Γ(*θ*)]. Γ(*θ*) values with lower incidence angles tended to show a greater dynamic range than values with higher incidence angles (Figure 6b).

The Lv. 3 product's spatial distribution snapshot maps showed relatively strong Γ values in the northwest triple-rice-cropping region (a.k.a., Dong Thap and An Giang provinces, Figure 7). Irrespective of seasonal differences, the northeastern non-rice-cropping upland zone showed low Γ reflectivity (Figure 7). These results were consistent with the Lband SAR data-based rice paddy distribution map and rice floodability map (Figure 8 [21]). The southwestern coastal wetland zone (comprising mangrove forests, fishponds and peatlands) showed continuously high Γ values throughout the year. High Γ (dB) noise occasionally remained in the specular data in the fine-spatial-resolution (i.e., low effective scattering area) Lv. 3 product (Figure 5). However, the noise was accompanied by high DDM 3D skewness/kurtosis values because the noise was derived from the locally high land surface roughness.
