H14B-01 INVITED [WITHDRAWN]
How Social Science is Informing Our Understanding of Flash Flooding: The Need for "Hydro-Socio-Meteo- rologists"
Hydrologists and meteorologists are teaming up with social scientists in flash flood studies that offer new methods, partnerships, and greater insights. New initiatives that are bringing social scientists together with meteorologists and hydrologists including WAS * IS (Weather and Society * Integrated Studies) and SSWIM (Social Science Woven into Meteorology) will be explained. This presentation will summarize some of these new studies and offer suggestions for promising new "hydro- meteo-socio" hybrid work that offer more potential than conventional uni-disciplinary research. Examples include a summary of YOUTube videos of drivers who post themselves driving through flooded roads and metrics that highlight when fatalities occur along a time line from issued watches and warnings. The presentation also offers a rationale for more sophisticated databases that can incorporate diverse streams of observations from social and physical science to increase our understanding of flash floods and reduce societal impacts.
A New Algorithm for Blending Multiple Satellite Precipitation Estimates With In-situ Gauge Precipitation Measurements Over Canada
This study proposes a new algorithm for blending satellite precipitation estimates with in-situ gauge precipitation measurements over Canada, which is to be used to produce a blended monthly total precipitation data on a 2.5-by-2.5 lat-long grid over Canada. The input satellite data used in this study include SSMI GPROF and SSMI UMORA (from 1987 to present), TVOS (from 1979 to 2002) and AIRS (from 2003 to present). The field of satellite precipitation estimates was adjusted against the field of gauge data addressing the biases in its mean and standard deviation before being used in the blending analysis. The blending (merging) technique is based on the ordinary kriging and conditional simulation technique with a combination of multiple satellite fields as the simulation field. A preliminary assessment of the performance of this algorithm is presented in this study.
Scale by Scale Statistical Evaluation of Satellite Precipitation Products
Analysis of about 1000 orbits of the TRMM VIRS, TMI and PR instruments over the range of 4 - 20,000 km shows that they are quite accurately scaling over most of the range, i.e. that their statistical properties depend on their resolution in a strong, power law manner. This opens up new possibilities for evaluating satellite rain products : a necessary condition is that their statistical properties be the same as those of rain at all scales, not only at the calibration scale (usually taken as the finest available). We present a scale by scale evaluation of three TRMM precipitation products : the 2B31, (rain rate from radar reflectivity and passive microwave with relatively realistic statistics), the 2A12 product (essentially a pure passive microwave based product) and the operational 3B42 product which uses many passive microwave sensors as well as infra red data. We show that although the data can have reasonable good statistics at the calibration scale, and - even if it has an accurate climatological mean - that its large scale statistics can still be poorly determined. This means that both the large scale statistics of the rain/ no rain regions, but also the large scale statistics of the extremes can be poorly estimated. We discuss the implications of this and the possibility of developing algorithms statistically accurate at all scales.
Statistical Framework for Evaluating Satellite Rainfall Products
The lack of uncertainty measures in operational satellite rainfall (SR) products leads to a situation where users of the SR products know that there are significant errors in the products, but they have no quantitative information about the distribution of these errors. The authors propose a semiparametric model to characterize the conditional distribution of true rainfall (TR) given SR products. The model consists of two components: a conditional density given each SR, and a smooth functional relationship between the density parameters and SR. The model is developed for daily, 0.25 degrees, satellite rainfall products over the US. The results are more informative than deterministic SR products since the whole conditional distribution enables users to take appropriate actions according to their own risk assessments and cost/benefit analyses.
Rainfall Intensities of Extreme Precipitation Events
Effort of evaluating very high-resolution precipitation products (instantaneous, 5 km) from satellite observation during extreme precipitation events is presented. The rainfall rate fields are based on the TRMM spaceborne radar observations. These fields are compared to those based on the new NOAA Next Generation QPE high- resolution national mosaic product (Q2, instantaneous, 1 km). For this study we have adjusted the Q2 instantaneous radar products to the gauges. Among the TRMM overpasses, two occurred during the time Tropical Storm Fay was crossing central Florida (22-23 Aug 2008). This allowed us to compare the satellite estimates not only to the Q2 products but also to the NASA TRMM ground validation radar products available for central Florida, which are also adjusted to rain gauge measurements. Other extreme events that are addressed include Hurricanes (Humberto, Gustav, Ike) and tornadic thunderstorms (e.g., the Feb. 6, 2008 deadliest tornado outbreak associated with 84 tornadoes; and the May 11, 2008 tornadic thunderstorms in which a tornado was reported at the exact time of the TRMM overpass directly at nadir).
Evaluation of Satellite Rainfall Products in a Complex Terrain and Humid Region in Ethiopia
We evaluated the performance of very high-resolution satellite products, particularly CMORPH and PERSIANN- CCS, in a humid and highly complex terrain in Ethiopia, using an event rain network that we established in the region in June 2008. The comparison period (July 2 - August 9, 2008) spans a major rainy period in the region. Our results revealed that both CMORPH and PERSIANN-CCS tend to underestimate heavy rainfall events by 20% to 80%, and PERSIANN-CCS misses nearly half of the light daily rainfall events. We also evaluated the accuracy of the two products in terms of their value as inputs to a distributed hydrologic model called MIKE-SHE.
Remotely-Sensed Rainfall for the Wettest Season in Oklahoma on Record
In the summer of 2007 Oklahoma experienced the wettest June on record, Oklahoma City had 20 consecutive days of reported rainfall (also a record), and damaging flash floods occurred on 15 days. This study analyzes the spatial patterns, temporal variability, and magnitudes of remotely-sensed rainfall from TRMM satellite, PERSIANN-CCS, and the operational rainfall product in the US National Weather Service (NWS) that relies on radar data with adjustments from rain gauges and human quality control. Conclusions drawn from this part of the study will help guide future steps toward integrated, multisensor precipitation estimation as applied to a season of extreme rainfall. The second part of the study applies the rainfall estimates under evaluation to an extreme flash flood case over the heavily instrumented Ft. Cobb basin in Oklahoma. Discharge is simulated and compared to observed streamflow on three subbasins using the NWS's distributed hydrologic model. Results will help determine if satellite-based rainfall estimates can be used, given proper downscaling, as inputs to hydrologic prediction models for extreme, small-scale hydrometeorological events.
Satellite Remote Sensing and Hydrological Modeling for Real-time Flood Inundation Mapping: A Case Study in Nzoia Basin, Lake Victoria, Africa
Floods are among the most recurrent natural disasters around the globe impacting human being. Its trend has
increased over the last decades. Getting satisfactory ground data for setting-up a flood prediction system had
been a major constraint in the past despite the availability of numerous hydrological models. In the regions
where installation of the ground equipments is limited by the available resources and rugged terrain, remotely
sensed information with the global coverage have become an alternative and supplement to the ground-based
observations in order to implement a cost-effective flood prediction system in many under-gauged regions.
First, this presentation will demonstrate the applicability of integrating NASA's standard satellite precipitation
products with a flood prediction model for disaster management in Nzoia, sub-basin of Lake Victoria, Africa. A
high-resolution distributed hydrologic model has been calibrated for the period of 1985-2006 and benchmark
streamflow simulations is produced with the calibrated hydrological model using the rain gauge data and
observed discharges. Afterward, continuous discharge predictions forced by real-time satellite precipitation
data was made. Acceptable results are obtained in comparison with the benchmark performance. Moreover, it
is identified that the flood prediction results are improved with systematically bias-corrected satellite rainfall
data. Finally, a space-based flood monitoring technique is also utilized to compare the flood inundation maps
with the hydrologically modeled surface flooding map in terms of probability of detection of flooding areas and
temporal correlation. The ultimate goal of the project is to build up disaster management capacity in East Africa
by providing local governmental officials and international aid organizations a practical decision-support tool in
order to better assess emerging flood impacts and to quantify spatial extent of flood risk, as well as to respond
to such flood emergencies more accurately.