Evan Ruzanski

The CASA Nowcasting System

AbstractShort-term prediction (nowcasting) of high-impact weather events can lead to significant improvement in warnings and advisories and is of great practical importance. Nowcasting using weather radar reflectivity data has been shown to be particularly useful and the Collaborative Adaptive Sensing of the Atmosphere (CASA) radar network provides high-resolution (0.5-km spatial and 1-min temporal resolution) reflectivity data that are amenable to producing valuable nowcasts. This paper describes the theory and implementation of a nowcasting system operating in the CASA Distributed Collaborative Adaptive Sensing network and shows that nowcasting can be reliably performed in such a distributed environment. In this context, nowcasting is used in a traditional sense to produce predictions of radar reflectivity fields up to 10 min into the future to support emergency manager decision making, and in a novel manner to support researchers and operational forecasters where 1–5-min nowcasts are used to steer the ...

An Investigation of the Short-Term Predictability of Precipitation Using High-Resolution Composite Radar Observations

AbstractThe short-term predictability of precipitation patterns observed by meteorological radar is an important concept as it establishes a means to characterize precipitation and provides an upper limit on the extent of useful nowcasting. Predictability also varies on the basis of spatial and temporal scales of the observed meteorological phenomena. This paper describes an investigation of the short-term predictability of precipitation patterns containing microalpha (0.2–2 km) to mesobeta (20–200 km) scales using high-resolution (0.5 km–1 min–1 dBZ) composite radar reflectivity data, extending the analysis presented in previous work to smaller space and time scales. An experimental approach is used in which continuous and categorical lifetimes of radar reflectivity fields in Eulerian and Lagrangian space are used to quantify short-term predictability. The space–time scale dependency of short-term predictability is analyzed, and a practical upper limit on the extent of Lagrangian persistence-based nowcas...

Evaluation of the Simultaneous Multiple Pulse Repetition Frequency Algorithm for Weather Radar

Abstract Performance of the simultaneous multiple pulse repetition frequency algorithm (SMPRF) for recovery of mean power and mean Doppler velocity is investigated using simulated weather radar data. Operation and functionality of the algorithm is described; methods to estimate mean power values using statistical inversion and to estimate mean velocity from unevenly spaced autocorrelation function samples are presented and analyzed. A simulation technique for constructing multiple pulse repetition interval data is described and the algorithm performance results are presented for an example SMPRF code using three weather profiles. This leads to the development of an error structure related to factors influencing moment recovery, including finite-length time series effects, the effects of overlaid echoes that create an effective signal-to-noise ratio that limits moment recovery performance, and the effects of spectrum width and radar frequency related to coherence time.

Scale Filtering for Improved Nowcasting Performance in a High-Resolution X-Band Radar Network

Precipitation patterns exist on a continuum of scales where generally larger scale features facilitate longer useful prediction times at the expense of coarser resolution. Favorable measurement range and resolution make weather radar observations an attractive choice for input to automated short-term weather prediction (nowcasting) systems. Previous research has shown that nowcasting performance can be improved by spatially filtering radar observations and considering only those precipitation scales that are most representative of pattern motion for prediction or filtering those scales from predicted fields deemed unpredictable by remaining past their lifetimes. It has been shown that an improvement in nowcasting performance can be obtained by first applying a nonlinear elliptical spatial filter to observed Weather Surveillance Radar 88 Doppler vertically integrated liquid water fields to predict motion of larger scale features believed to better represent the motion of the entire precipitation pattern for forecast lead times up to 1 h. It has also been shown in the literature that wavelet transform can be used to develop measures of predictability at each scale and adaptive wavelet filters can be designed to remove perishable scales from predicted continental-scale reflectivity fields according to prediction lead time. This paper investigates the adaptation of both of these approaches and Fourier filtering to evaluate the effects of scale filtering on nowcasting performance using a Fourier-based nowcasting method and high-resolution Engineering Research Center for Collaborative Adaptive Sensing of the Atmosphere radar reflectivity data. A maximum improvement of approximately 18% in terms of Critical Success Index was observed by applying Fourier filtering in the context of truncating Fourier coefficients within the prediction model to the observed sequence of reflectivity fields used for assimilation. In addition, applying Fourier filtering to the resulting predictions showed a maximum reduction in mean absolute error of approximately 14%.

Improved Liquid Water Equivalent Nowcasting Using the Weather Support to Deicing Decision Making System

AbstractShort-term automated forecasts (nowcasts) of liquid water equivalent (LWE) values can be used to assist aviation deicing decision-making activities. Such decisions can mitigate hazards that cause losses of life and property and increase costs because of travel delays. The Weather Support to Deicing Decision Making (WSDDM) system provides LWE nowcasts and is currently deployed at several major airports in the United States. WSDDM produces these nowcasts in two steps. First, an equation relating radar reflectivity to LWE rate is calibrated by correlating radar and surface observations. Then, nowcasts of reflectivity are converted to nowcasts of LWE using this calibrated equation. This paper shows that the incorporation of the Dynamic and Adaptive Radar Tracking of Storms (DARTS) radar–based nowcasting method into WSDDM can provide more accurate and efficient nowcasts of LWE relative to the correlation-based nowcasting method currently used. Results of an evaluation considering approximately 92 h of ...

Short-term predictability of weather radar quantities and lightning activity

Accurate, spatially specific, and temporally extended short-term automated forecasts (nowcasts) of lightning activity are of great interest to the preservation of life and resources for a multitude of applications. This paper presents a study to provide insight into the best manner and extent to nowcast lightning activity to a desired location. Radar quantities whose observations have previously been shown to be reliable precursors of lightning activity are nowcasted to introduce a concept of total lead time, whereby these nowcasts can potentially be used to nowcast lightning activity to a desired upstream site (e.g., airport, stadium, etc.). This study extends previous work by analyzing the correlation structure between nowcasts of radar and lightning quantities in Lagrangian space, giving insight into the manner and extent to which lightning activity at a desired location can be nowcasted.

Weather Radar Data Interpolation Using a Kernel-Based Lagrangian Nowcasting Technique

The Dynamic Radar Tracking of Storms (DARTS) model is a Lagrangian persistence-based nowcasting model that has previously shown utility in nowcasting a variety of weather radar data in severe weather and aviation decision support applications. DARTS is based on the discrete Fourier transform and thus provides an inherent means to perform interpolation. In this context, the model is modified such that interpolation can be accurately and efficiently performed by appropriately windowing the input data and evaluating an interpolating polynomial using the fast Fourier transform. The utility of this interpolation methodology for operational use is demonstrated, and its performance is compared with linear and cubic spline interpolation methods. The use of the original DARTS model to perform advection-based interpolation is also investigated. Rainfall rates derived from data collected by the Weather Service Radar-1988 Doppler S-band radar and the X-band radar at the Dallas-Fort Worth test bed were used for the analyses. The results show that the modified DARTS technique yielded normalized standard error values that were close to those of the forward-backward advection approach using the original DARTS model and ran about 2-4 orders of magnitude faster in terms of computation time. The error structure of the interpolation methods in the context of spatial variability and sampling of atmospheric scales represented by the data is also presented. In this sense, utility of the 1-2-km scales was shown, and the modified DARTS-based approach showed the ability to effectively utilize the value in these scales.

Recent updates to the CASA nowcasting system

The Collaborative Adaptive Sensing of the Atmosphere (CASA) nowcasting system currently provides 0-30 min automated forecasts (nowcasts) of precipitation to National Weather Service forecasters, emergency managers, and researchers using composite X-band weather radar data. Nowcasting is accomplished in two steps. First, the Fourier-based Dynamic and Adaptive Radar Tracking of Storms (DARTS) technique computes a motion vector field representing precipitation pattern motion using a recently observed sequence of radar reflectivity fields. Then, future reflectivity fields are estimated by recursively advecting the latest observed or predicted field according to this motion vector field using a sine kernel-based method. This paper presents potential upgrades to the CASA nowcasting system. The performance of the current sine kernel-based advection method is compared to that of a backward mapping technique in terms of categorical (rain/no rain) assessments of accuracy. Because computational efficiency is an important concern given the high-resolution (0.5 km/1 min) nature of the CASA data, the respective computational efficiencies are also compared. A technique to perform temporal interpolation within the DARTS model with the potential application to data fusion is also presented and assessed.

Two-year assessment of nowcasting performance in the CASA system

Nowcasting refers to short-term automated forecasting (0–6 hours or less) of high-impact weather events such as heavy rainfall that can produce severe flooding. Accurate and efficient nowcasting can be used to assist emergency managers in the decision-making process. This paper presents an evaluation of nowcasting performance within the Collaborative Adaptive Sensing of the Atmosphere (CASA) system from the 2009–2010 Integrative Project 1 (IP1) experiment. The nowcasting methodology consists of the Dynamic and Adaptive Radar Tracking of Storms (DARTS) method for motion estimation and a sinc kernel-based advection method. Radar reflectivity fields were predicted up to 10 min into the future. Previous analysis is extended to include data over a two year period using Critical Success Index (CSI), False Alarm Ratio (FAR), Probability of Detection (POD), and Mean Absolute Error (MAE) scores for evaluation. Analysis of the categorical scores (i.e., CSI, POD, FAR) relative to scoring threshold and neighborhood is also presented.

Nowcasting rainfall fields estimated from specific differential phase

This paper presents a preliminary evaluation of short-term prediction (nowcasting) of rainfall fields estimated from specific differential phase fields derived from Collaborative Adaptive Sensing of the Atmosphere X-band radar data. A Fourier-space, linear system-based nowcasting method used these rainfall fields as input to generate rainfall forecasts up to 20 min. The results show the extent to which specific differential phase-derived rainfall fields can be predicted and the utility of such predictions to be approximately 15 min.