H32A-01 INVITED
On Streamflow and Water Balance Modeling at Regional and Aquifer Scales: Validation of Remotely Sensed Radar Precipitation Through Continuous Simulation
The hydrologic water balance, governing the amount of water that an aquifer receives, is important for estimating the amount of water that can be safely extracted without diminishing the water resources of a region. The change in storage of water in a stream-aquifer system for a specific time period is affected by precipitation, surface runoff, ground water recharge/discharge/pumping, evaporation, and transpiration. Sufficiently accurate and detailed precipitation measurements are needed over regional, river basin, and aquifer recharge areas. The accuracy and spatial sampling density of rainfall observation systems affect the accuracy of hydrologic predictions, and can be a major limitation to understanding the water fluxes across surface and subsurface boundaries. The study area includes the Arbuckle-Simpson aquifer located in South Central Oklahoma and underlies an approximate area of 1295 km2. The Arbuckle-Simpson aquifer provides water to streams and rivers as baseflow, including the 1200 km2 Blue River. This study is motivated by research concerning recharge and the hydrologic water balance over the Arbuckle- Simpson aquifer region. The objective of this study is to identify components of the hydrologic water balance, especially streamflow and recharge. The influence of spatially and temporally variable precipitation on the surface runoff, and uncertainty associated with radar and gauge sensor systems is evaluated over a period of fourteen years at hourly timesteps within a distributed hydrologic modeling context. To accomplish this objective, a distributed hydrologic model of the surface drainage systems including the Blue River, and components of the Arbuckle-Simpson aquifer recharge area are simulated using spatially variable rainfall derived from three rainfall products: 1) Gauge only products derived from surrounding Mesonet gauges, 2) National Weather Service Stage III/MPE radar rainfall (ABRFC) with no adjustment or quality control, and 3) Mean Field Bias adjustment of the ABRFC product with Mesonet gauges and gap filling. Validation of these rainfall products using observed streamflow is assessed through continuous simulation of direct runoff. The uncertainty and accuracy of streamflow within a distributed modeling context is accomplished using Vflo setup at various resolutions including 200- and 500-m grids over an area of 1590 km2, and 13240 km2, respectively. Precipitation derived from rain gauge is consistent with radar estimation at annual timescales, however, significant differences in streamflow simulated with these products result.
H32A-02 INVITED
Rainfall Measurement, Estimation, and Validation: A Practical Perspective
Measuring and estimating rainfall accurately continues to be a significant challenge. While rainfall is arguably the most dynamic water flux, thus crucial for predictive hydrologic models, monitoring its spatial and temporal variability is subject to observational difficulties. The authors discuss field activities focused on providing accurate data for validation studies of remote sensing based rainfall estimates. The activities include three operational rain gauge networks based in Iowa and Kansas. The authors present the instrumental and informational technologies used in operating these research networks. They also discuss evaluation efforts of optical and mechanical disdrometers. Drop size distribution data are crucial for the interpretation and improvements of satellite and radar remote sensing of rainfall. Quantities that are still lacking thorough understanding include the spatial variability of DSD characteristics at the scales smaller than those of the satellite and radar resolution, i.e. ~1-10 km. To obtain this information in the context relevant to hydrologic applications it is necessary to deploy a dense network of inexpensive but reliable disdrometers. The authors present comparison results of several collocated optical and mechanical disdrometers and tipping bucket rain gauges. They discuss the instruments' adequacy for such a deployment.
H32A-03
Multi-sensor Precipitation Reanalysis
The archive of the Next Generation Radar (NEXRAD) Level II and Level III data at the National Climatic Data Center (NCDC) provides a unique opportunity for developing a high-resolution rainfall climatology suitable for regional applications. The data from almost all of the Weather Surveillance Radars 1988-Dopplers (WSR-88D) are archived and can be accessed via the High Density Storage System (HDSS) at NCDC. The Level III Digital Precipitation Array (DPA) for several radars over North and South Carolina are used to develop a merged radar and rain gauge rainfall product that has hourly temporal resolution and 4x4 km2 spatial resolution. The operational multi-sensor precipitation estimation algorithm of the NWS river forecast centers is adapted to work in a reanalysis mode. Six radars over North and South Carolina are selected for development of the multi-sensor precipitation reanalysis. Each radar has been operational prior to 1996 and thus provides for a 10-year data set through 2005. Ultimately, data sets that are monthly, yearly, and seasonal averages and accumulations will be developed for studying the rainfall variability over a relatively long period (Approx. 10 years) and large spatial extent. This presentation describes the science issues, such as reduction of long term radar-to-radar biases, reduction of local small scale biases, parameter optimization via cross validation, and merging of multi-sensor data sets. Initial results are shared in this presentation. A beta version regional product is available via a THREDDS data server at NCDC.
H32A-04
Delineation of Aerial Extent of Precipitation Using Multi-Spectral Remotely Sensed Data
This study analyzes the importance of using multi-spectral data for precipitation area detection. Five image channels from GOES-12, including channel 1 (visible channel, 0.65μm), channel 2 (3.9μm), channel 3 (water vapor channel, 6.5μm), channel 4 (thermal channel, 10.7μm), and channel 6 (13.3μm), were evaluated. In part of the precipitation classification procedure, the self organizing feature map (SOFM) was used to classify multi-dimensional images into a number of clusters. The probability of precipitation (POP) for each cluster is then calculated based on NEXRAD precipitation observations. Experiments were set to the summer time period (Jun-Aug of 2006) over the continental United States. Different scenarios, with a combination of various image channels, were tested to find the best combination of channels for day-time and night-time precipitation detection. Using the pattern-matching technique of Lovejoy and Austin (1979) an optimum POP threshold (for 1D) or boundary (for 2D) is defined for each scenario such that satellite classes having a higher POP are treated as precipitation and those with lower POP as no-precipitation. Comparison of the scenarios is accomplished by using the equitable threat score (ETS) obtained from the contingency table. Some overall results are: 1) Including visible channel in the day-time results in a much better score than other channel combination without visible channel. 2) For night-time precipitation detection, a significant improvement in score is achieved using the combined channels (2-6) instead of one single channel alone. In the presentation, detailed ETS statistics will be provided. In addition, extension of the precipitation detection from five GOES channels to future GOES-R channels will be discussed.
H32A-05
An inter-comparison of five high resolution satellite precipitation estimates with three-hourly gauge data
The last several years have seen the development of a number of new satellite-derived, globally complete, high resolution precipitation products with a spatial resolution of at least 0.25° and a temporal resolution of at least three-hourly. These products generally merge geostationary infra-red data and polar-orbiting passive microwave data to take advantage of the high sampling of the infra-red and the superior quality of the microwave. The Project to Evaluate High Resolution Precipitation Products (PEHRPP) was established to evaluate and inter- compare these datasets at a variety of spatial and temporal resolutions with the intent of guiding dataset developers and informing the user-community as to the most useful products. As part of this project, we have performed a sub daily inter-comparison of five high resolution datasets (commonly known as CMORPH, TMPA, NRL-Blended, the Hydro-Estimator and PERSIANN) with existing sub-daily gauge data over the US and the Pacific Ocean. The results show that these datasets are surprisingly good at representing high resolution precipitation, with correlations against three-hourly gauge data as high 0.7 for some datasets. In general, CMORPH yields the highest correlations against the gauges used, although the TMPA yields the lowest biases due to the gauge correction applied as part of the algorithm.
H32A-06
Comparison of Four Global Satellite Rainfall Data Products over the United States against WSR-88D Radar Rainfall Data
We assessed the multi-dimensional uncertainty structure of four pseudo-real time and publicly available quasi- global satellite rainfall products overland at spatial scales of 0.04 and 0.25 degrees. These four products were: 1) NASA's Infrared Rainfall (IR) product 3B41RT; 2) NASA's Merged Microwave Rainfall product 3B42RT; 3) NOAA CPC Passive Microwave Rainfall product CMORPH, and 4) PERSIANN produced by University of California (Irvine). Each of these products adopts a distinct physical scheme (such as the use of Lagrangian or Eulerian frame of reference) and calibration strategy (such as neural networks or the use of TRMM-PR) in the rainfall estimation process. Analysis was performed over two regions in the US known to have a distinct hydro- climatology – i) Mid-western US (a semi-arid zone), and ii) Florida (sub-tropical zone modulated by coastal effects). Uncertainty was assessed using WSR-88D rainfall as reference for ground validation data. Nine hydrologically relevant error metrics were computed for each product according to a multi-dimensional error modeling scheme that decomposes the marginal error in three major dimensions: 1) temporal dimension (how does the error vary in time?); ii) spatial dimension (how does the error vary in space?) and iii) retrieval dimension (how ‘off' is each rainfall estimate from the true value over rainy areas?).
H32A-07
Hydrological evaluation of satellite precipitation products in the La Plata basin
A number of satellite precipitation products with high spatiotemporal resolutions and quasi-global coverages have become available to the research community. Precipitation is arguably the most important atmospheric driving force to the land surface hydrological system. A major science objective of the proposed Global Precipitation Measurement (GPM) mission is to improve prediction capabilities for floods, landslides, freshwater resources, and other hydrological applications. However, given the perceived retrieval errors in satellite precipitation estimates, satellite precipitation products are not well accepted by the hydrological community. In this study, we evaluate three existing satellite-based precipitation products (TMPA-3B42RT, PERSIANN, and CMORPH) in La Plata basin in South America. These products are compared among themselves and with in situ observations. To make the analysis period consistent for each data set, we selected the 2003-2006 period (all data are available at sub-daily time steps and 0.25° grid) for the investigation. Spatially-distributed and basin- wide precipitation estimates from the satellite precipitation products are compared with gauge-derived precipitation to quantify the bias and uncertainty in the remotely sensed precipitation. The simulated runoff fields and streamflow resulting from a semi-distributed precipitation-runoff model driven by the satellite precipitation products are compared with those driven by gridded in situ observations and with observed streamflow. The impacts of errors in the satellite precipitation estimates on hydrologic state variables (e.g., soil moisture) and fluxes (including river discharge) are investigated. This study helps clarify the level of the accuracy of the state-of- the-art satellite precipitation products for hydrological predictions and provides useful insights into the potential utility of the planned GPM mission.
H32A-08
Evaluating Satellite Rainfall Products and their Impacts in hydrologic Model simulations
Recent research works have produced satellite-based high-resolution rainfall products, such as, PERSIANN, CMORPH and TRMM's 3B42. These products have their own limitations in terms of estimation accuracy and space-time resolutions, and hence, their potential and utility for hydrological applications has yet to be assessed. In this study, we evaluate the accuracy of the rainfall products by comparing them with high-quality and dense ground-based rainfall observations, and assess their hydrological impacts by comparing observed streamflows with hydrologic model simulations forced by satellite rainfall products. We focus our work on the Little Washita watershed (610 km2) in central Oklahoma, which contains high-quality radar datasets, a dense network of rain gauges (42 rain gauges), five stream gauges, and three meteorological stations. In the first part of the study, we describe the accuracy of the satellite rainfall products on different space-time resolutions, using a variety of performance metrics. In the second part, we describe the performance of the hydrologic model forced by satellite rainfall products with respect to reproducing observed streamflow data. We used a physically-based, fully- distributed hydrologic model, known as the TIN-based Real-time Integrated Basin Simulator (tRIBS) for this investigation. The results of this study will shed light on the accuracy of satellite rainfall products, both from meteorological and hydrological perspectives.