Flow recession analysis, relating discharge Q and its time rate of change -dQ/dt, has been widely used to understand catchment scale flow dynamics. However, data points in the recession plot, the plot of -dQ/dt versus Q, typically form a wide point cloud due to noise and hysteresis in the storage-discharge relationship, and it is still unclear what information we can extract from the plot and how to understand the information. There seem to be two contrasting approaches to interpret the plot. One emphasizes the importance of the ensembles of many recessions (i.e., the lower envelope or a measure of central tendency), and the other highlights the importance of the event scale analysis and questions the meaning of the ensemble characteristics. In this study, we examine if those approaches can be reconciled. We utilize a machine learning tool to capture the point cloud using the past trajectory of discharge. Our results show that most of the data points can be captured using 5 days of past discharge. We show that we can learn the catchment scale flow recession dynamics from what the machine learned. We analyze patterns learned by the machine and explain and hypothesize why the machine learned those characteristics. The hysteresis in the plot mainly occurs during the early time dynamics, and the flow recession dynamics eventually converge to an attractor in the plot, which represents the master recession curve. We also illustrate that a hysteretic storage-discharge relationship can be estimated based on the attractor.

Process-based modeling of soil water movement with the Richards equation requires the description of soil hydraulic material properties, which are highly uncertain and heterogeneous at all scales. This limits the applicability of Richards equation at larger scales beyond the patch scale. The experimental capabilities of the three hillslopes of the Landscape Evolution Observatory (LEO) at Biosphere 2 provide a unique opportunity to observe the heterogeneity of hydraulic material properties at the hillslope scale. We performed a gravity flow experiment where through constant irrigation the water content increases until the hydraulic conductivity matches the irrigation flux above. The dense water content sensor network at LEO then allows to map the heterogeneity of hydraulic conductivity at a meter scale resolution. The experiment revealed spatial structures within the hillslopes, mainly a vertical trend with the lowest hydraulic conductivity close to the surface. However, the variation between neighbouring sensors is high, showing that the heterogeneity cannot be fully resolved even at LEO. By representing the heterogeneity in models through Miller scaling we showed the impact on hillslope discharge. For the hillslope with the smallest heterogeneity, representing the dominant structures was sufficient. However, for the two hillslopes with the larger overall heterogeneity, adding further details of the local heterogeneity did impact the discharge further. This highlights the limitations of Richards equation, which requires the heterogeneous field of material properties, at the hillslope scale and shows the relevance to improve our understanding of effective parameters to be able to apply the process-based model to larger scales.