*4.5. Spatial Distributions of Monthly Precipitation*

Using July 2016 (the month with the maximum precipitation) and September 2017 (the month with the minimum precipitation) as examples (Figures 9–12), the spatial distributions of the OBS, OIMERG, CIMERG, DS\_OIMERG, DS\_Spline, and DS\_CIMERG monthly precipitation data are shown in Figure 9; Figure 11. Furthermore, Figure 10; Figure 12 show the detailed distributions of the precipitation datasets in the Ili Valley after image enlargement. Figure 9a shows the observed precipitation distribution in July 2016. The Ili Valley was the region with the most abundant precipitation. The other five gridded estimated precipitation datasets (OIMERG (Figure 9b), CIMERG (Figure 9c), DS\_Spline (Figure 9d), DS\_CIMERG (Figure 9e), and DS\_OIMERG (Figure 9f)) all exhibited the same spatial distribution pattern as that of the OBS dataset. However, OIMERG underestimated the observed precipitation, particularly in the Ili Valley area. In contrast, CIMERG, obtained by combining OIMERG and ground observations using OI, showed a more reasonable precipitation distribution. Figure 10 shows a clearer view of the patterns in this area. Since DS\_Spline and DS\_OIMERG were generated by downscaling the original IMERG data, their performance was limited by the IMERG precipitation distribution, which underestimated the observed precipitation. By contrast, after OI-GWR processing, DS\_CIMERG not only showed noticeably improved accuracy relative to the initial IMERG precipitation data but also exhibited a precipitation distribution that was more detailed and consistent with the observed precipitation, thus demonstrating the improved capabilities of the proposed downscaling method. Figure 11 presents the spatial distributions of the precipitation data for September 2017 (the month with the lowest precipitation). It can be seen that, during this time period, the precipitation was weak throughout the Tianshan Mountain area (Figure 11a). Consistent with the results shown in Figure 9, the DS\_CIMERG dataset showed a distribution more similar to the observed precipitation distribution (Figure 11e), while the distributions of DS\_Spline (Figure 11d) and DS\_OIMERG (Figure 11f) were consistent with that of OIMERG.

**Figure 9.** The spatial distributions of the five estimated precipitation datasets and the observed precipitation in July 2016. (**a**) OBS; (**b**) OIMERG; (**c**) CIMERG; (**d**) DS\_Spline; (**e**) DS\_CIMERG; (**f**) DS\_OIMERG.

**Figure 10.** Enlarged views of the Ili Valley corresponding to Figure 9. (**a**) OBS; (**b**) OIMERG; (**c**) CIMERG; (**d**) DS\_Spline; (**e**) DS\_CIMERG; (**f**) DS\_OIMERG.

**Figure 11.** The spatial distributions of the five estimated precipitation datasets and the observed precipitation in September 2017. (**a**) OBS; (**b**) OIMERG; (**c**) CIMERG; (**d**) DS\_Spline; (**e**) DS\_CIMERG; (**f**) DS\_OIMERG.

**Figure 12.** Enlarged views of the Ili Valley corresponding to Figure 11. (**a**) OBS; (**b**) OIMERG; (**c**) CIMERG; (**d**) DS\_Spline; (**e**) DS\_CIMERG; (**f**) DS\_OIMERG.
