Graduate Exam Abstract
Amin AlqudahPh.D. Preliminary
November 21, 2008, 10:30 AM
Rainfall Estimation from Spaceborne and Ground Based Radars Using Neural Networks
Abstract: Rainfall observed on the ground is dependent on the four dimensional distribution of precipitation aloft. In principle, the functional relation between rain rate on the ground and the four-dimensional radar observations aloft can be obtained. However it is difficult to express this in a simple form. The key challenge in radar rainfall estimation is the space-time variability in precipitation microphysics, such as Drop Size Distribution (DSD) and drop shapes. A simple Z-R relation is not sufficient to capture the variability and has large uncertainty and it needs to be adaptively adjusted based on validation. Prior research has shown that neural networks can be used to estimate ground rainfall from radar measurements. Neural network is a nonparametric method to represent the relationship between radar measurements and rainfall rate. The relationship is derived directly from a data set consisting of radar measurements and rain gauge measurements. The usefulness of the rainfall estimation using neural networks is subject to many factors such as the representativeness and sufficiency of the training data set, the generalization capability of the network to new data, seasonal change, regional change, and so on.
Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) is known to be the first observation platform for mapping precipitation over the tropics. TRMM is a joint mission between of National Aeronautic and Space Administration (NASA) of USA and Japanese Aerospace Exploration Agency (JAXA) of Japan. TRMM measured rainfall is important in order to study the precipitation distribution all over the globe in the tropics. TRMM ground validation is a critical important component in TRMM system to ensure the measurement accuracy and its successful application using certain ground based weather radars and rain gauge networks of stable and sufficient quality. However, these ground sensing systems have quite different characterizations from TRMM in terms of resolution, scale, viewing aspect, and uncertainties in the sensing environments. This makes the use of ground radar rainfall information in order to get TRMM point of view a very challenging task. In addition, the ground validation systems themselves get validated often with rain gauges.
In this proposal, rainfall estimation using neural networks is investigated in order to improve rainfall estimation using neural networks technique based on measurements taken by ground radars and TRMM PR radar. Ground Radar measurements will be used to estimate rainfall using adaptive neural network techniques. Improvements are also suggested and performed including the use of Principal Components Analysis. For TRMM-PR purposes a single neural network is not efficient to extract the relation between TRMM-PR measurements and the rain gauges; this is because of the resolution differences between TRMM-PR profile and the rain gauges and the low number of TRMM overpasses over these gauges which will make the training data set to have less number of profiles and not be able to generalize. Therefore, a novel hybrid Neural Network model is presented to train ground radars for rainfall estimate using rain gauge data and subsequently the trained ground radar estimates rain to train TRMM PR based Neural Networks for rainfall estimation. This hybrid neural network model will derive the relation between rain gauges and ground radar measurements, and transfer this relation to adaptive rainfall estimation for TRMM precipitation radar.
Adviser: Prof. V Chandrasekar
Non-ECE Member: Prof. Paul W. Mielke, Jr., Department of statistics
Member 3: Prof. V N. Bringi, Electrical and Computer Engineering Department
Addional Members: NA
Program of Study: