An investigation of the spatial and temporal characteristics of extreme dry and wet events across NLDAS-2 models

Extreme hydrological events (including droughts and floods) produce severe social and economic impacts. Monitoring hydrological processes from remote sensing is necessary to improve understanding and preparedness for these events, with current missions focusing on a range of hydrological variables (i.e. SWOT, SMAP, GRACE). This study uses output from three state-of-the-art land surface assimilation models and an event clustering algorithm to identify the characteristic spatial and temporal scales of large-scale extreme dry and wet events in the contiguous United States for three major hydrological processes: precipitation, runoff and soil moisture. We also examine the sensitivity of extreme event characteristics to model resolution, and assess inter-model differences. We find that models generally agree in terms of the mean characteristics of events: precipitation dry events are shorter duration compared to soil moisture and runoff, and more intense events tend to be smaller in area. We also find that mean spatial and temporal characteristics are highly dependent on model resolution; important in the context of detecting and monitoring these events. Results from this study can be used to inform land surface model development, extreme hydrology event detection, and sampling requirements of upcoming remote sensing missions in hydrology.