Search

More than half of all streams globally are non-perennial, and thus dynamic due to their expanding and retracting nature. Field mapping in conjunction with observational data from gauges and/or in-situ loggers is a typical approach for studying non-perennial stream dynamics, but these approaches underrepresent their spatiotemporal variability. High-resolution distributed hydrological modeling promises to bridge this gap, thanks to advances in model physics, remote sensing, and computational power. As the first step towards this goal, we investigate the capability of distributed hydrologic modeling to capture stream dynamics in Upper Blue River Basin, OK. Coupled Routing and Excess STorage (CREST), a distributed hydrological model, is used to simulate spatiotemporally varied streamflow at 10-meter spatial resolution and daily time steps. USGS stream gauge data and in-situ state logger data are used to calibrate and validate the simulation at the watershed outlet and small headwater tributaries, respectively. Dynamic Surface Water Estimate (DSWE), a LANDSAT product is also compared with simulated water presence in high-order streams. Results show that the CREST model can capture low-moderate streamflow values at the watershed outlet with a log-NSE value over 0.7 in the validation period, while underestimating high flow values due to the daily time step. Also, the calibrated model can accurately estimate wet/dry status as monitored by in-situ state loggers in nine headwater catchments. The dynamic stream networks are mapped over 2510 stream segments using the CREST simulation. Non-perennial streams are the most dynamic in small headwater tributaries (contributing area

Precipitation is an essential climate and forcing variable for modeling the global water cycle. Particularly, the recent Integrated Multi-satellite Retrievals for GPM (IMERG) product retrospectively provides an unprecedented two decades of high-resolution satellite precipitation estimates. The primary goal of this study is to examine the similarities and differences between the two latest and also arguably the most popular, GPM IMERG Early and Final Run (ER and FR) products across the globe. The results reveal that: (1) ER systematically estimates 12.0% higher annual rainfall than FR, particularly over land (16.7%); (2) ER and FR show significant differences in instantaneous rates (Root Mean Squared Difference: RMSD = 2.38 mm h−1 and normalized RMSD: RMSD_norm = 1.09), especially in Africa (RMSD = 2.40 mm h−1) and hot, arid regions (RMSD_norm = 1.11), but less so in Europe (RMSD = 2.16 mm h−1) and cold areas (RMSD_norm = 0.87); and (3) ER measures 33.0% higher extreme rainfall rates than FR over the globe. The exploration of their similarities and differences provides a first-order global assessment of various hydrological utilities: FR is designed to be more suitable for retrospective hydroclimatology and water resource applications, while the earliest available ER product, though not bias-corrected by rain gauges, still shows potential utility for operational modeling of rainfall-triggered hazards. The findings of this study can provide an assessment to product developers and broader data users and practitioners to address the inherent issues in hardware limitations, retrieval algorithms, and uncertainty quantification for research and applications.