Suchergebnisse

Abstract. Rainfall observed on the ground is dependent on the four dimensional structure of precipitation aloft. Scanning radars can observe the four dimensional structure of precipitation. Neural network is a nonparametric method to represent the nonlinear relationship between radar measurements and rainfall rate. The relationship is derived directly from a dataset consisting of radar measurements and rain gauge measurements. The performance of neural network based rainfall estimation is subject to many factors, such as the representativeness and sufficiency of the training dataset, the generalization capability of the network to new data, seasonal changes, and regional changes. Improving the performance of the neural network for real time applications is of great interest. The goal of this paper is to investigate the performance of rainfall estimation based on Radial Basis Function (RBF) neural networks using radar reflectivity as input and rain gauge as the target. Data from Melbourne, Florida NEXRAD (Next Generation Weather Radar) ground radar (KMLB) over different years along with rain gauge measurements are used to conduct various investigations related to this problem. A direct gauge comparison study is done to demonstrate the improvement brought in by the neural networks and to show the feasibility of this system. The principal components analysis (PCA) technique is also used to reduce the dimensionality of the training dataset. Reducing the dimensionality of the input training data will reduce the training time as well as reduce the network complexity which will also avoid over fitting.

Abstract. Flooding is one of the most common natural hazards that produce substantial loss of life and property. The QPE products that are derived at high spatiotemporal resolution, which is enabled by the deployment of a dense radar network, have the potential to improve the prediction of flash-flooding threats when coupled with hydrological models. The US National Science Foundation Engineering Research Center for Collaborative Adaptive Sensing of the Atmosphere (CASA) is dedicated to revolutionizing our ability to observe, understand, predict, and respond to hazardous weather events, especially in the lower atmosphere. CASA's technology enables precipitation observation close to the ground and QPE is one of the important products generated by the system. This paper describes the CASA QPE system built on the various underlying technologies of networked X-band radar systems providing high-resolution (in space and time) measurements, using the rainfall products from the radar. Evaluation of the networked rainfall product using 5 yr of data from the CASA IP-1 test bed is presented. Cross validation of the product using 5 yr of data with a gauge network is also provided. The validation shows the excellent performance of the CASA QPE system with a standard error of 25% and a low bias of 3.7%. Examples of various CASA rainfall products including instantaneous and hourly rainfall accumulations are shown.

Abstract. Weather radar systems are very suitable tools for the monitoring of extreme rainfall events providing measurements with high spatial and temporal resolution over a wide geographical area. Nevertheless, radar rainfall retrieval at C-band is prone to several error sources, such as rain path attenuation which affects the accuracy of inversion algorithms. In this paper, the so-called rain profiling techniques (namely the surface reference method FV and the polarimetric method ZPHI) are applied to correct rain path attenuation and a new neural network algorithm is proposed to estimate the rain rate from the corrected measurements of reflectivity and differential reflectivity. A stochastic model, based on disdrometer measurements, is used to generate realistic range profiles of raindrop size distribution parameters while a T-matrix solution technique is adopted to compute the corresponding polarimetric variables. A sensitivity analysis is performed in order to evaluate the expected errors of these methods. It has been found that the ZPHI method is more reliable than FV, being less sensitive to calibration errors. Moreover, the proposed neural network algorithm has shown more accurate rain rate estimates than the corresponding parametric algorithm, especially in presence of calibration errors.