**4. Results**

### **4.1 Morphological signatures and role of internal organization**

The spatial behaviors simulated at the outlet of the five basins are first studied. Several points of measurements have been added within the biggest basins, to further understand the genesis of the simulated sum of surfaces. In the Montecito Creek basin (**Figure 6**), the surfaces located upstream the points 2 and 3 support together the *SMax* measured at the final outlet (**Table 3**), around the 100th iteration. This catchment presents a high IC value at the final outlet (55.74), due to the hierarchical organization of networks. Such efficiency is more effective in the San Ysidro Creek (**Table 3**): many surfaces flow since the upstream part of the point 7, and numerous surfaces are drained thanks to the well-structured networks. In Hot Spring or Buena Vista Creek, the cascading surface flow system is slightly less inefficient as the contributions of sub-basins are shifted in space. As a consequence, values for *Smax* are lower. In Romero Creek, upstream areas of the point 10 present an efficient organization, but their contribution is not combined with other sub-basins at the final outlet, explaining the small value for *SMax* at the final outlet (**Figure 6**). A similar discrepancy induces in Hot Spring Creek a long out-flow. Obviously, the *SMax* value in upstream point 1 in the Montecito Creek is close to those obtained in upstream points 10 and 3 (**Figure 6**), but as they are not spatially combined, no major peak of surfaces can appear.


**Table 3.**

*Outputs collected at the final outlet of the five studied catchments.*

This first analysis then confirms that the global catchment scale is not relevant to address the effects of morphological conditions on hydrological responses, especially for the five studied catchments: for example, the upstream part of point 7 explains 100% of *Smax* estimated at the final outlet (point 6), while its surfaces only represent 64% of the global basin size (11.63 km2 ). So, we need to track concentration within networks at a finer scale, hence the interest in going down to the cellular scale.

#### **4.2 The "cauliflower effects"**

Maps indicating values of Smax at the cellular scale confirm that strong high values (IC >55) exist within the catchments (**Figure 7**). A very important value (IC = 81.1) is simulated within the San Ysidro upstream part (Smax: 22.58 ha; A: 7.74 km**2**), and this record has never been observed elsewhere and in previous studies. Indeed, even if RuiCells © has been applied on more than 450 catchments in France [49], the older maximum value was estimated in Saint-Martin-de-Boscherville (in France), with IC equal 71.6. In this study, for San Ysidro Creek, the internal efficiency was already suggested (**Figure 6**), but the "cauliflower effect" is remarkable: the contribution of three well-structured sub-basins suddenly increases the IC values, as they contribute together and surfaces arrive at the same moment in upstream of the point 7. IC values remain higher (>55) until the final outlet is reached (**Figure 8**), which indicates that surface flow is efficient during a distance of around 1.125 km. The number of branches and their similar distance to the outlet aggravate hydrological responses and support current solid debris content, especially during the postfire conditions.

In Romero Creek, another IC high value is clearly detected (IC = 76.9; Smax: 12.40; A: 2.65 km**2**), while IC was weak at the outlet (IC = 46.50). Here, the rest of the catchment does not play a role in the surface response. In Montecito Creek, several IC values appear, and they exceed the threshold value of 55 (57.8 at the point 2 and 58.4 at the point 3). One homothetic behavior is observed: efficient concentration areas emerge in different confluences in the river system, and this explains why a distance of 450 m (red-colored) upstream of point 0 still has a morphological efficiency. And finally, morphology in the Buena Vista and the Hot Spring Creeks seem to be less effective. In fact, the two basins present an internal concentration, but values are weaker (53.1 and 52.1) in comparison with others, so their morphological efficiencies are hidden by other extreme values.

As a consequence, the "cauliflower effect" is detected, and it gives new patterns to the relations between networks and areas (**Figure 8**), completing the previously

#### **Figure 8.**

*The "cauliflower effect" in upstream morphological areas.*

known forms (**Figure 5**). More networks and surfaces are numerous and equidistant from the outlet point, more this "cauliflower effect" takes sense as shown by the form around the new record point. The morphological areas often do not correspond to the global catchment scale, that is why the global scale is not relevant to address the morphological influence on surface flow dynamics. And it also explains with morphometric parameters (calculated for a given form or size) are really insufficient to detect their influence in case of violent and sudden events like debris flows.
