**3. Results**

By presenting the results, this section provides insights into where the model/satellite systematically differ and discusses the possible causes.

Spatial distribution of precipitation accumulation during the flood days (15 March to 2 April 2019) corresponding to the in situ observations, satellite, and three NWP models are shown in Figure 2. It is clear that high precipitation areas extended from the northeast along the Alborz Mountains to the southeastern Caspian Sea; and from the west to southwest along the Zagros Mountains. Overall, the precipitation spatial patterns in all products were correctly captured, although there were differences in the precipitation magnitude among the products. In more specific terms, IMERG precipitation product (IMERG-RT V05) reasonably captured the precipitation distribution for most parts of the country when compared with the in situ observations. Moreover, the in situ observations highlighted four precipitation hotspots along the Zagros Mountain in the west of Iran with the largest precipitation amount. IMERG outperformed the other products to capture these hotspots. The remarkable precipitation gradients were well-captured by IMERG, possibly due to its native higher spatial IMERG resolution (~10 km × ~10 km) when compared with those of the NWP models (50 km × 50 km). Among the NWP models, the NCEP model forecasted much lower precipitation, while

UKMO generally overestimated the in situ observations. In general, most of the precipitation products could capture the basic precipitation characteristics in terms of the observed spatial pattern, however, it is unknown as to whether the overestimation/underestimation was really derived by the product structure or if it was due to lack of dense in situ observations. It should be noted that the mean of the ensemble NWP forecasts was used in Figure 2.

**Figure 2.** Spatial distribution of accumulated precipitation (mm) from the observational, ECMWF, NCEP, UKMO, and IMERG satellite for 19 precipitation days over Iran.
