The goal of this research is to develop a framework to improve satellite precipitation applications
in hydrologic modeling, and is carried out in three parts. First, the multi-scale variability of
streamflow in the Connecticut River Basin (CRB) (i.e. study area) is captured using the Coupled
Routing and Excess STorage (CREST) model. This analysis presents a good opportunity to
enhance and refine the estimation of hydrological processes via the CREST model and
understand its calibration and validation procedures over northern-latitude basins. It introduces a
large time frame and focuses on different time resolutions and multiple basin scales (191–25,000
km2). Available observations are used to assess the accuracy of simulations, determine the spatial
variability of calibration parameters, and understand the applicability of calibrated parameter
values applied to interior catchments in order to predict flows at ungauged locations.
In the second research dimension, multiple high-resolution precipitation products from
currently available quasi-global satellite algorithms (TRMM-3B42V7, CMORPH, and
PERSIANN) are investigated in terms of their hydrologic applicability relative to the National
Weather Service (Stage IV) gauge-adjusted radar-rainfall dataset, which is a novel analysis in
terms of satellite error propagation issues. Basin-average satellite precipitation and resulting
streamflow simulations are evaluated for a period of 9 years (2002 through 2010) using as
reference the Stage IV rainfall and associated CREST-based flow simulations. Error metrics in
basin-average rainfall and runoff simulations are determined for high values (>90th percentile)
and overall (>20th percentile), warm vs. cold periods and two temporal aggregations (3 hourly
and daily). Results show that the TRMM-3B42V7 product, which includes gauge adjustment, is
the most accurate in terms of capturing the high percentile values of precipitation and runoff over
the study basins. The satellite-alone CMORPH and PERSIANN precipitation products
consistently underestimated the radar precipitation and reference-driven streamflow simulations.
The study demonstrated a basin scale dependency of the satellite-precipitation error metrics and
the precipitation-to-runoff simulation propagation properties of these errors.
The third part was motivated by the above findings to evaluate the accuracy of return
period flow estimations using a hydrologic model driven with high-resolution satellite
precipitation data. Four quasi-global satellite products (TRMM-3B42V7, TRMM-3B42RT,
CMORPH, and PERSIANN) at 3-hourly/0.25-deg resolution and the National Weather Service
(Stage IV) gauge-adjusted radar rainfall dataset (representing the reference rainfall) are
integrated in this study with the CREST model to simulate 3-hourly and daily temporal
resolution annual peak flows during warm season (May-November) months in CRB. The log-
Pearson type III frequency distribution applied on an 11 year (2002 to 2012) record of annual
peak flow data is used to derive 1.01-yr, 5-yr, 10-yr, 20-yr and 50-yr return period flows.
Evaluation against the Stage IV driven simulations shows that the TRMM-3B42V7 product has
the highest correlation and a slight overestimation in terms of the derived annual maxima flows
compared to the other satellite products. Each basin exhibits distinct flood frequency curves
derived from reference precipitation data, which are well captured by the flood frequency
analyses of the various satellite products driven simulations. With the increasing record length of
high-resolution satellite products, results from this study can motivate future studies on flood
frequency analysis of ungauged basins driven by remotely sensed precipitation. |