Close Window

PC-09 Abstract

High Resolution, Multi-Spectral, Automatic Satellite Rainfall Estimation Over Amazonia in Real Time

Marcos Costa — UFV - Universidade Federal de Viçosa (SA-PI)
Gilberto A. Vicente — NOAA (US-PI)

Precipitation is an essential component

of the hydrological cycle and probably is the most important climatic variable

of the Amazonian water balance. Accurately estimating precipitation rates

in the Amazon Basin has been a significant challenge, because it has very

few rain gauges and receives large amounts of rainfall. Willmott et

al. (1994) estimates that, for the region’s gauge density (fewer than

20 stations per 106 km2), estimates of precipitation

can be in error by as much as 100 to 700 mm/yr. The low rain gauge density is also

reflected in the quality of studies that use rainfall as input data.

For example, in a simulation study of Amazonian river discharge as a function

of land cover and climatic conditions, Costa and Foley (1997) showed that

uncertainties in the rainfall data in western Amazon basin (where rain

gauge density is lower) was one of the most important factors that contributed

to errors in the simulated discharges, even at locations thousands of kilometers

downstream.




In order to minimize the gauge density

problem, satellite measurements and numerical models have been widely used

in recent years. The rainfall estimates provided by the assimilation of

the results of numerical weather prediction models present several problems,

including low resolution and undulations associated with the spectral representation

of topography in such models (Costa and Foley, 1998). On the other hand,

satellite estimates can have high resolution, and accuracy depends on

the technique and calibration used. The most common satellite rainfall

estimation techniques rely on cloud top temperature infrared (IR) measurements

only (Martin et al., 1990), microwave (MW) measurements (Negri et

al., 1994) or on the combination of IR and MW (Vicente, 1994). The

IR-only technique is the most indirect approach because it relies on the

statistical interpretation of cloud top temperature. However it provides

very high temporal resolution for the estimates since the temperature measurements

are made by a satellite on a geosynchronous orbit. The MW technique provides

the most accurate measurement of instantaneous rainfall rates, because

the precipitation amounts are physically related to the cloud MW radiation

emission and scattering. However, this technique can only provide estimates

a few times a day over a narrow area because it relies on measurements

made from a polar orbiting satellite. The techniques that use both IR and

MW take advantage of the strengths of both measurements.




In order to address the request of the

US National Weather Service (NWS) for an operational, automated and highly

accurate satellite rainfall estimation technique, G. Vicente and R. Scofield

(two of the authors of this proposal) have developed in recent years

at the National Oceanic and Atmospheric Administration (NOAA), a multi-spectral,

multi-channel satellite rainfall estimation technique (Vicente et al. 1998).

The technique evolved from a regression relationship between the GOES-8,

10.7 mm IR brightness temperature channel and surface rainfall rate derived

for the summer season on the US central Great Plains and the Gulf of Mexico

areas [Figure 1] (Vicente 1996). Further improvements led to the use

of model-generated relative humidity (RH) and precipitable water (PW) to

analyze the environmental moisture and to balance the rainfall amounts accordingly,

based on Scofield (1987). A refinement on the location of the precipitation

systems and screening of non-precipitating cloud is achieved by a detailed

analysis of spatial cloud top temperature gradient, cloud top growth, and

decaying rate, and by the use of the GOES-8 visible channel during daytime.

The technique has been running in real time for over three years and provides

1-hour and 3-hour total rainfall estimates every 30 minutes, 6-hour total

estimates at synoptic time and 24-hour estimates every 12:00 UTC. All estimates cover

the whole US territory with a spatial resolution of 4 by 4 km. The results

are available to Internet users through the NOAA Flash Flood Home Page

(URL: http://orbit-net.nesdis.noaa.gov/ora/ht/ff).

Statistical analyses made by the NOAA Satellite Analyses Branch (SAB)

have shown that the technique is accurate for operational use by the NWS

in 90% of the cases where the precipitation comes from cloud with warm

top and 70% of the cases where the precipitation comes from cold

cloud tops. Independent validation work by Vicente et al. (1998) has shown correlation coefficients varying from 0.52 to 0.76 for 1-hour

estimates at a 12 by 12 km grid size resolution. The results are better

when extended over longer time periods (3, 6 and 24 hours) and larger spatial

scales (48 by 48 km and 100 by 100 km grid size boxes).




Parallel to the work of developing an operational

satellite rainfall estimation technique for the US territory, Vicente and

Scofield (1996) have started a simplified version to compute rainfall rate

in real time over Brazil. The Brazilian version of the technique has been running

for about two and half years and has many of the features of the US version.

However, it has not been calibrated or validated. The results are also

available in real time as an experimental product on the NOAA Flash Flood

Home page mentioned above (see Precipitation over South America). In this

proposal, we intend to expand this version not only to assimilate most

of what has been learned from the US version, but especially to calibrate

the parameterizations to the precipitation systems characteristic of Amazonia.

The deliverable of this research will provide satellite rainfall estimates

at spatial resolution of 4 by 4 Km and at time resolution of 30 minutes.

These precipitation measurements will be initially produced for 3 years,

but can be extended to the whole LBA experiment (6 years) or even longer,

subject to availability of funds. The proposed technique has better temporal

and spatial coverage than the TRMM satellite estimates, and knowledge of

diurnal variations of precipitation can be important for other parts of

the LBA project, especially LBA-Ecology. Several other research groups in

the LBA-Hydrometeorology and LBA-Ecology program will benefit from this

information, especially the modeling groups. We believe that the results

will allow the building of an LBA-long dataset of all components of the

surface water cycle.




We will validate the technique on a daily

bases using gauge measurements provided by the LBA project, whenever it

is available and of good quality. Additional comparisons to the instantaneous

rainfall estimates provided by the TRMM will also be done for the times

and areas covered by the TRMM orbits. The statistics of the comparisons

on a region by region basis in Amazonia, as well as real-time estimates

for 1, 3, 6 and 24 hours will be available daily through a specially dedicated

Web page to internet users during the whole duration of the project. The

digital data in GrADS (Grid Analysis and Display System) and in McIDAS

(Man computer Interactive Data Access System) will be available through

anonymous FTP to all investigators supported by the NASA LBA program from

our centers both in the U.S. (NOAA/NESDIS location at the World Weather Building

in Camp Springs, MD, USA) and in Brazil (Federal University of Viçosa,

MG, Brazil). An additional copy of the archived precipitation measurements

can be stored at the CPTEC LBA data center, if requested.




Figure 1.











 



References 



Costa, M. H. and J. A. Foley, 1998:

A comparison of precipitation datasets for the Amazon basin. Geophysical

Research Letters, 25, 155-158.




Costa, M. H. and J. A. Foley, 1997:

The water balance of the Amazon basin: Dependence on vegetation cover and

canopy conductance, J. Geophys. Res.-Atmospheres, 102, 23973-23990.




Goodman, B., D. W. Martin, W. P.

Menzel and E. C. Cutrim, 1994: A non-linear algorithm for estimating 3-hourly

rain rates over Amazonia from GOES/VISSR observations. Remote Sensing

Reviews, 10, 169-177.




Martin, D. W., B. Goodman, T. J.

Schmit and E. C. Cutrim, 1990: Estimates of daily rainfall over the Amazon

Basin. J. Geophys. Res., 95, No. D10, 17043-17050.




Negri, J. A., R. F. Adler, E. J.

Nelkin, and G. J. Huffman, 1994. Regional Rainfall Climatologies Derived

from Special Sensor Microwave Images (SSM/I) Data. Bull. Amer. Meteor.

Soc., 75, 1165-1182.




Scofield, R. A., 1987: The NESDIS

operational convective precipitation technique. Mon. Wea.. Rev., 115, No. 8, pp 1773-1792.




Vicente, G. A., 1998: The operational

GOES infrared rainfall estimation technique. September issue of the Bulletin

of the American Meteorological Society.




Vicente, G. A., 1996: Algorithm

for rainfall rate estimation using a combination of GOES-8 11.0 and 3.9

mm measurements. Proceedings of the 8th Conference on Satellite Meteorology

and Oceanography, Atlanta, GA, USA, 5 pp.




Vicente, G. A. and Meiry S. Sakamoto, 1996: Access to real time rainfall estimation over Brazil and South America

through the WWB. Proceedings of the 9th Brazilian Meteorology Conference,

Belo Horizonte, MG, BRAZIL.




Vicente, G. A. and R. Scofield.  1996.  Experimental

GOES-8/9 derived estimates for flash flood and hydrological applications.  Proceedings

of the 1996 Meteorological Satellite Data User's Conference
, Vienna,

Austria.  8pp.




Vicente, G. A., 1994: Microwave

and Infrared Satellite Radiometric Measurements, Ph.D. Thesis. Department

of Atmospheric and Oceanic Sciences, University of Wisconsin - Madison,

WI, USA, 127 pp.




For more information e-mail to: gvicente@nesdis.noaa.gov

 







Close Window