Rain Rate Estimation Research Articles

A REVIEW OF TRMM/PR RAIN RATE RETRIEVAL ALGORITHM BIASES CAUSED BY SURFACE REFERENCE TECHNIQUE

The standard rain rate retrieval algorithm for TRMM/PR applies hybrid method with surface reference technique (SRT) and HitschfeldBordan method. Path integrated attenuation (PIA) estimated by the SRT of the standard algorihtm is sometimes negatively biased because surface backscattering cross section increases during rainfall. This is called “soil moisture effect” in this paper and it is not explicitly considered in the SRT of the standard algorithm. Temporal reference method does not follow this effect at all and spatial reference method partly follows this effect, but fails in some area such as the Sahel. The negative bias in PIA leads to the negative bias in the rain rate estimates especially for heavy rain.

Seasonal and Diurnal Variations of Raindrop Size Distribution in Asian Monsoon Region

Diurnal and seasonal variations of raindrop size distribution (DSD) at Gadanki (GD), Singapore (SG) and Kototabang (KT) are studied to elucidate characteristics of DSD in the Asian monsoon region. It is found that DSDs are affected by diurnal convective cycles and seasonal variations in precipitation characteristics. GD has the most significant seasonal variation in DSD. Clear difference in rainfall characteristics between the Southwest and Northeast monsoon seasons is considered to be the main cause of such clear seasonal variation. KT has the most significant diurnal variation of DSD, which is probably caused by the fact that KT is greatly affected by ocean-land contrast and mountain effects to generate local convection in the afternoon. SG has less diurnal and seasonal variations compared with the other two locations, which is related to the fact that SG is affected both by land and oceanic rainfall. Z-R relations applicable to radar rainfall measurement in these areas are derived. It is shown that the use of the Marshall-Palmer Z-R relation (Z = 200R1.6) gives bias errors of about 1.5 dB or less in rain rate estimation except for the northeast monsoon season in GD, for 12∼18 local time during pre-southwest monsoon season in GD, and for 06∼12 local time during some monsoon seasons in KT.

Validation of Rain-Rate Estimation in Hurricanes from the Stepped Frequency Microwave Radiometer: Algorithm Correction and Error Analysis

Abstract Simultaneous observations by the lower fuselage (LF) radar, the tail (TA) radar, and the Stepped Frequency Microwave Radiometer (SFMR) on board the NOAA WP-3D aircraft are used to validate the rainfall rate estimates from microwave emission measurements of SFMR in tropical cyclones. Data collected in Hurricane Bonnie (1998) and Hurricane Humberto (2001) with a total of 820 paired samples are used in the comparisons. The SFMR 10-s path-integrated rain rates are found to have an overestimate in light rain and an underestimate in heavy rain relative to radar rainfall estimates. Examination of the existing SFMR algorithm shows that the coefficient should be changed in the attenuation—rain-rate relationship used in the inversion algorithm. After this correction, a linear regression result with a correlation coefficient of 0.8 and a slope close to 1 is obtained. But an overall high bias of 5 mm h−1 of the SFMR rainfall estimate relative to radar is also found. The error analysis shows that the bias is nearly independent of rain type, a result confirming Jorgensen and Willis’s conclusion that the drop size distributions between convective and stratiform rain in hurricanes are similar. It is also shown that the bias is a weak function of wind speed, as well as a weak inverse function of radial distance to the hurricane center. Temperature dependence has been ruled out as the main explanation. After doing sensitivity tests, the authors conclude that the bias results from a combination of two factors: an underestimate of the freezing-level height, and a downward increase of radar reflectivity in the high wind regions. If the true downward increase is 1–2 dBZ km−1, a 0.5-km underestimate of the freezing-level height could account for up to a 3–5 mm h−1 bias.

Effect of Temporal Sampling on Inferred Rainfall Spatial Statistics

Abstract On the basis of temporally sampled data obtained from satellites, spatial statistics of rainfall can be estimated. In this paper, the authors compare the estimated spatial statistics with their “true” or ensemble values calculated using 5 yr of 15-min radar-based rainfall data at a spatial domain of 512 km × 512 km in the central United States. The authors conducted a Monte Carlo sampling experiment to simulate different sampling scenarios for variable sampling intervals and rainfall averaging periods. The spatial statistics used are the moments of spatial distribution of rainfall, the spatial scaling exponents, and the spatial cross correlations between the sample and ensemble rainfall fields. The results demonstrated that the expected value of the relative error in the mean rain-rate estimate is zero for rainfall averaged over 5 days or longer, better temporal sampling produces average fields that are “less noisy” spatially, an increase in the sampling interval causes the sampled rainfall to be increasingly less correlated with the true rainfall map, and the spatial scaling exponent estimators could give a bias of 40% or less. The results of this study provide a basis for understanding the impact of temporal statistics on inferred spatial statistics.

Monitoring soil wetness variations by means of satellite passive microwave observations: the HYDROPTIMET study cases

Abstract. Soil moisture is an important component of the hydrological cycle. In the framework of modern flood warning systems, the knowledge of soil moisture is crucial, due to the influence on the soil response in terms of infiltration-runoff. Precipitation-runoff processes, in fact, are related to catchment's hydrological conditions before the precipitation. Thus, an estimation of these conditions is of significant importance to improve the reliability of flood warning systems. Combining such information with other weather-related satellite products (i.e. rain rate estimation) might represent a useful exercise in order to improve our capability to handle (and possibly mitigate or prevent) hydro-geological hazards. Remote sensing, in the last few years, has supported several techniques for soil moisture/wetness monitoring. Most of the satellite-based techniques use microwave data, thanks to the all-weather and all-time capability of these data, as well as to their high sensitivity to water content in the soil. On the other hand, microwave data are unfortunately highly affected by the presence of surface roughness or vegetation coverage within the instantaneous satellite field of view (IFOV). Those problems, consequently, strongly limit the efficiency and the reliability of traditional satellite techniques. Recently, using data coming from AMSU (Advanced Microwave Sounding Unit), flying aboard NOAA (National Oceanic and Atmospheric Administration) satellites, a new methodology for soil wetness estimation has been proposed. The proposed index, called Soil Wetness Variation Index (SWVI), developed by a multi-temporal analysis of AMSU records, seems able to reduce the problems related to vegetation and/or roughness effects. Such an approach has been tested, with promising results, on the analysis of some flooding events which occurred in Europe in the past. In this study, results achieved for the HYDROPTIMET test cases will be analysed and discussed in detail. This analysis allows us to evaluate the reliability and the efficiency of the proposed technique in identifying different amounts of soil wetness variations in different observational conditions. In particular, the proposed indicator was able to document the actual effects of meteorological events, in terms of space-time evolution of soil wetness changes, for all the analysed HYDROPTIMET test cases. Moreover, in some circumstances, the SWVI was able to identify the presence of a sort of "early" signal in terms of soil wetness variations, which may be regarded as a timely indication of an anomalous value of soil water content. This evidence suggests the opportunity to use such an index in the pre-operational phases of the modern flood warning systems, in order to improve their forecast capabilities and their reliability.

The Tropical Rainfall Potential (TRaP) Technique. Part I: Description and Examples

Abstract Inland flooding caused by heavy rainfall from landfalling tropical cyclones is a significant threat to life and property. The tropical rainfall potential (TRaP) technique, which couples satellite estimates of rain rate in tropical cyclones with track forecasts to produce a forecast of 24-h rainfall from a storm, was developed to better estimate the magnitude of this threat. This paper outlines the history of the TRaP technique, details its current algorithms, and offers examples of its use in forecasting. Part II of this paper covers verification of the technique.

Monitoring of rainfall by ground-based passive microwave systems: models, measurements and applications

Abstract. A large set of ground-based multi-frequency microwave radiometric simulations and measurements during different precipitation regimes are analysed. Simulations are performed for a set of frequencies from 22 to 60 GHz, representing the channels currently available on an operational ground-based radiometric system. Results are illustrated in terms of comparisons between measurements and model data in order to show that the observed radiometric signatures can be attributed to rainfall scattering and absorption. An inversion algorithm has been developed, basing on the simulated data, to retrieve rain rate from passive radiometric observations. As a validation of the approach, we have analyzed radiometric measurements during rain events occurred in Boulder, Colorado, and at the Atmospheric Radiation Measurement (ARM) Program's Southern Great Plains (SGP) site in Lamont, Oklahoma, USA, comparing rain rate estimates with available simultaneous rain gauge data.

Satellite radiometric remote sensing of rainfall fields: multi-sensor retrieval techniques at geostationary scale

Abstract. The Microwave Infrared Combined Rainfall Algorithm (MICRA) consists in a statistical integration method using the satellite microwave-based rain-rate estimates, assumed to be accurate enough, to calibrate spaceborne infrared measurements on limited sub-regions and time windows. Rainfall retrieval is pursued at the space-time scale of typical geostationary observations, that is at a spatial resolution of few kilometers and a repetition period of few tens of minutes. The actual implementation is explained, although the basic concepts of MICRA are very general and the method is easy to be extended for considering innovative statistical techniques or measurements from additional space-borne platforms. In order to demonstrate the potentiality of MICRA, case studies over central Italy are also discussed. Finally, preliminary results of MICRA validation by ground based remote and in situ measurements are shown and a comparison with a Neural Network (NN) based technique is briefly illustrated.

Modeling and measurement of rainfall by ground-based multispectral microwave radiometry

The potential of ground-based multispectral microwave radiometers in retrieving rainfall parameters is investigated by coupling physically oriented models and retrieval methods with a large set of experimental data. Measured data come from rain events that occurred in the USA at Boulder, Colorado, and at the Atmospheric Radiation Measurement (ARM) Program's Southern Great Plains (SGP) site in Lamont, OK. Rain cloud models are specified to characterize both nonraining clouds, stratiform and convective rainfall. Brightness temperature numerical simulations are performed for a set of frequencies from 20 to 60 GHz at zenith angle, representing the channels currently deployed on a commercially available ground-based radiometric system. Results are illustrated in terms of comparisons between measurements and model data in order to show that the observed radiometric signatures can be attributed to rainfall scattering and absorption. A new statistical inversion algorithm, trained by synthetic data and based on principal component analysis is also developed to classify the meteorological background, to identify the rain regime, and to retrieve rain rate from passive radiometric observations. Rain rate estimate comparisons with simultaneous rain gauge data and rain effect mitigation methods are also discussed.

On optimum rain rate estimation from a pulsed Doppler Weather Radar at Chennai

A 10 cm S-band Doppler Weather Radar (DWR) has been installed as a replacement of an outlived analogue S-band radar at Cyclone Detection Radar (CDR) station, Chennai during September-October, 2001. Technical specifications and capabilities of this DWR have been briefly mentioned in this paper. The digital data obtained from this DWR have been used for the period November-December, 2001 to estimate the rain rate based on the Marshall-Palmer relationship between the radar reflectivity factor (z) and the rain rate (R). The relation z = 267R1.345 estimates well the rain rate as measured by the self recording rain gauges located within 100 km radius from the DWR. This relationship has been tested for its operational applicability during March 2002 - December, 2003 and found that the accumulated precipitation from the radar estimation was within an error of 15% from the rain gauge measured values. Information on the twenty four hours accumulated areal distribution of precipitation can be used by the water managers and operational hydrologists for the effective water management over the catchments since the error in rain rate estimation over a wider area is relatively small in comparison to point rainfall estimation. 
 

'Down-to-Earth' modelling of equivalent surface precipitation using multisource data and radar

The estimation of surface rainfall from reflectivity data derived from weather radar has been much studied over many years. It is now clear that central to this problem is the adjustment of these data for the impacts of vertical variations in the reflectivity. In this paper a new procedure (known as Down-to-Earth, DTE) is proposed and tested for combining radar measurements aloft with information from a numerical weather-prediction (NWP) model and an analysis system. The procedure involves the exploitation of moist cloud physics in an attempt to account for physical processes impacting on precipitation during its descent from the height of radar echo measurements to the surface. The application of DTE leads to increased underestimation in the radar measurements compared to precipitation gauge observations at short and intermediate radar ranges (0-120 km), but is successful at reducing the bias at further ranges. However the application of DTE does not lead to significant decreases in the random error of the surface rain rate estimate. No improvement is made when attempting to account for the precipitation phase measured by radar. It is concluded that further work on radar data quality control, along with improvements to the NWP model, are essential to improve upon results using such a physically based procedure.

Variability of Drop Size Distributions: Time-Scale Dependence of the Variability and Its Effects on Rain Estimation

Abstract A systematic and intensive analysis is performed on 5 yr of reliable disdrometric data (over 20 000 one-minute drop size distributions, DSDs) to investigate the variability of DSDs in the Montreal, Quebec, Canada, area. The scale dependence (climatological scale, day to day, within a day, between physical processes, and within a physical process) of the DSD variability and its effect on rainfall intensity R estimation from radar reflectivity Z are explored in terms of bias and random errors. Detail error distributions are also provided. The use of a climatological R–Z relationship for rainfall—affected by all of the DSDs’ variability—leads on average to a random error of 41% in instantaneous rain-rate estimation. This error decreases with integration time, but the decrease becomes less pronounced for integration times longer than 2 h. Daily accumulations computed with the climatological R–Z relationship have a bias of 28% because of the day-to-day DSD variability. However, when daily R–Z relationships are used, a random error of 32% in instantaneous rain rate is still present because of the DSD variability within a day. This illustrates that most of the variability of DSDs has its origin within a storm or between storms within a day. Physical processes leading to the formation of DSDs are then classified according to the vertical structure of radar data as measured by a UHF profiler collocated with the disdrometer. The DSD variability among different physical processes is larger than the day-to-day variability. A bias of 41% in rain accumulations is due to the DSD variability between physical processes. Accurate rain-rate estimation (∼7%) can be achieved only after the proper underlying physical process is identified and the associated R–Z relationship is used.

A Model for Estimation of Rain Rate on Tropical Land from TRMM Microwave Imager Radiometer Observations

Over the tropical land regions observations of the 85 GHz brightness temperature (T(sub 85v)) made by the TRMM Microwave Imager (TMI) radiometer when analyzed with the help of rain rate (R(sub pR)) deduced from the TRMM Precipitation Radar (PR) indicate that there are two maxima in rain rate. One strong maximum occurs when T(sub 85) has a value of about 220 K and the other weaker one when T(sub 85v) is much colder approx. 150 K. Together with the help of earlier studies based on airborne Doppler Radar observations and radiative transfer theoretical simulations, we infer the maximum near 220 K is a result of relatively weak scattering due to super cooled rain drops and water coated ice hydrometeors associated with a developing thunderstorm (Cb) that has a strong updraft. The other maximum is associated with strong scattering due to ice particles that are formed when the updraft collapses and the rain from the Cb is transit2oning from convective type to stratiform type. Incorporating these ideas and with a view to improve the estimation of rain rate from existing operational method applicable to the tropical land areas, we have developed a rain retrieval model. This model utilizes two parameters, that have a horizontal scale of approx. 20km, deduced from the TMI measurements at 19, 21 and 37 GHz (T(sub 19v), T(sub 21v), T(sub 37v). The third parameter in the model, namely the horizontal gradient of brightness temperature within the 20 km scale, is deduced from TMI measurements at 85 GHz. Utilizing these parameters our retrieval model is formulated to yield instantaneous rain rate on a scale of 20 km and seasonal average on a mesoscale that agree well with that of the PR.

Method for estimation of rain rate with Rayleigh and Mie scattering assumptions on the Z– R relationship for different rainfall types

A comparison of the rain rate estimated with the assumptions of Rayleigh and Mie scattering is made. We analyzed the different relationships between the radar reflective factor and rain rate (so-called Z– R relationship) with both scattering models for different DSDs (droplet size distributions) and rainfall types as the wavelength is 2.2 cm which is in accord with the band of TRMM/PR. Meanwhile we introduced a discrete ordinates method to retrieve the Z– R relationship for Mie scattering assumption. It is found that the retrieval result can be represented as the sum of some simple Z– R relationships. By the analysis of the Z– R relationships estimated from Rayleigh and Mie scattering assumptions in the rain types, we found that the difference of Z– R relationships between Rayleigh and Mie scattering in the thunderstorm that represents the larger raindrop size is larger than that in the drizzle that represent the smaller raindrop size.

Semi‐operational rainfall observations with X‐band multi‐parameter radar

AbstractThree‐year semi‐operational observations of rainfall distributions with NIED X‐band multi‐parameter (or polarimetric) radar started in the Kanto area of Japan from July 2003. The purposes and outlines of the radar observations with networks of rain gauges and disdrometers for ground validations are described. Preliminary results of validation analysis of polarimetric rain rate estimators show the usefulness of X‐band multi‐parameter radar for hydrological and meteorological applications in a small area. Copyright © 2005 Royal Meteorological Society

Effect of Natural Variations in Rain Drop Size Distributions on Rain Rate Estimators of 3 cm Wavelength Polarimetric Radar

Statistical errors of rain rate estimators due to natural variations in raindrop size distribution (DSD) are studied for 3-cm wavelength polarimetric radar. Four types of estimators are examined: A classical estimator R(ZH), and three types of polarimetric radar estimators R(KDP), R(ZH, ZDR), and R(KDP, ZDR), where R is the rain rate, ZH is the reflectivity factor at horizontal polarization, KDP is the specific differential phase, and ZDR is the differential reflectivity. The T-matrix method is employed for the scattering calculations, and a total of 7,664 one-minute raindrop size spectra, measured with a Joss-Waldvogel type disdrometer are used.According to simulation results, the normalized errors (NEs) of R(ZH), R(KDP), R(KDP,ZDR), and R(ZH,ZDR) for all DSD samples are 25%, 14%, 9%, and 10%, respectively. The NEs of all estimators, except R(ZH), tend to decrease with increasing rain rate. For rain rates larger than 10 mmh−1, e.g., the average NEs of R(ZH), R(KDP), R(KDP, ZDR), and R(ZH,ZDR) are 25%, 9%, 5%, and 7%, respectively. The simulation results show that the classical estimator R(ZH) is the most sensitive to variations in DSD and the estimator R(KDP, ZDR) is the least sensitive.The lowest sensitivity of the rain estimator R(KDP, ZDR) to variations in DSD can be explained by the following facts. The difference in the forward-scattering amplitudes at horizontal and vertical polarizations, which contributes KDP, is proportional to the 4.78th power of the drop diameter. On the other hand, the exponent of the backscatter cross section, which contributes to ZH, is proportional to the 6.38th power of the drop diameter. Because the rain rate R is proportional to the 3.67th power of the drop diameter, KDP is less sensitive to DSD variations than ZH. However, DSD spectra with unusually large median volume diameter D0 can increase the estimation error of R(KDP). The differential reflectivity ZDR reduces the effect of unusual D0 and is useful for further improvement of the estimator R(KDP). This is due to the fact that ZDR itself is a good measure of D0.

Rainfall estimation from infrared data using an improved version of the Auto-Estimator Technique

In order to apply a suitable procedure for now-casting application in the case of heavy precipitation events, we discuss possible improvements to the Auto-Estimator Technique (AET). The AET rain estimation technique identifies precipitating clouds by comparing two successive infrared satellite images. If the pixel in the second image is colder, then upward motion is assumed and the convective system is considered to be a precipitating cloud. If the pixel in the second image is warmer, then the convective system is likely to be in a weakening phase and no precipitation is assigned. In the case of horizontal movement of the convective system, for instance approaching and weakening, the pixel in the second image can be colder without any upward motion or rain. We propose an improvement of AET (IAET) that determines the cloud top temperature variation and the corresponding rain rate estimation, taking into account the effect of horizontal motion of clouds, obtained by optical flow techniques. Precipitation events over Sardinia during November 1999, May 2000 and June 2000 were analysed. Three-hour cumulated precipitation on one pixel size area were determined by infrared Meteosat data, using both AET and IAET. A comparison between satellite estimation and 49 corresponding ground data stations was made. Contingency tables and relative statistical indices (Bias, False Alarm Ratio, Probability of Detection, Hit Rate, Critical Success Index), and errors on 3 h precipitation (mean absolute error (MAE) and Bias) were computed. When no horizontal motion of the convective system is present, AET and IAET have the same performances. In the events of May 2000, with fast movement of the precipitation system, AET significantly overestimates the rain rate while IAET gives much better performances.

A Hybrid Surface Reference Technique and Its Application to the TRMM Precipitation Radar

Abstract Satellite weather radars that operate at attenuating wavelengths require an estimate of path attenuation to reconstruct the range profile of rainfall. One such method is the surface reference technique (SRT), by which attenuation is estimated as the difference between the surface cross section outside the rain and the apparent surface cross section measured in rain. This and the Hitschfeld–Bordan method are used operationally to estimate rain rate using data from the precipitation radar (PR) aboard the Tropical Rainfall Measuring Mission (TRMM) satellite. To overcome some of the problems associated with the latest operational version of the SRT, a hybrid surface reference is defined that uses information from the along-track and cross-track variations of the surface cross sections in rain-free areas. Over ocean, this approach eliminates most of the discontinuities in the path-attenuation field. Self-consistency of the estimates is tested by processing the orbits backward as well as forward. Calculations from 2 weeks of PR data show that 90% of the rain events over ocean for which the SRT is classified as reliable or marginally reliable are such that the absolute difference between the forward and backward estimates is less than 1 dB.

Multivariate statistical integration of Satellite infrared and microwave radiometric measurements for rainfall retrieval at the geostationary scale

The objective of this paper is to investigate how the complementarity between low earth orbit (LEO) microwave (MW) and geostationary earth orbit (GEO) infrared (IR) radiometric measurements can be exploited for satellite rainfall detection and estimation. Rainfall retrieval is pursued at the space-time scale of typical geostationary observations, that is at a spatial resolution of few kilometers and a repetition period of few tens of minutes. The basic idea behind the investigated statistical integration methods follows an established approach consisting in using the satellite MW-based rain-rate estimates, assumed to be accurate enough, to calibrate spaceborne IR measurements on sufficiently limited subregions and time windows. The proposed methodologies are focused on new statistical approaches, namely the multivariate probability matching (MPM) and variance-constrained multiple regression (VMR). The MPM and VMR methods are rigorously formulated and systematically analyzed in terms of relative detection and estimation accuracy and computing efficiency. In order to demonstrate the potentiality of the proposed MW-IR combined rainfall algorithm (MICRA), three case studies are discussed, two on a global scale on November 1999 and 2000 and one over the Mediterranean area. A comprehensive set of statistical parameters for detection and estimation assessment is introduced to evaluate the error budget. For a comparative evaluation, the analysis of these case studies has been extended to similar techniques available in literature.

Uniting Space, Ground, and Underwater Measurements for Improved Estimates of Rain Rate

Global precipitation is monitored from a variety of platforms including spaceborne, ground-, and ocean-based platforms. Intercomparisons of these observations are crucial to validating the measurements and providing confidence for each measurement technique. Probability distribution functions of rain rates are used to compare satellite and ground-based radar observations. A preferred adjustment technique for improving rain rate distribution estimates is identified using measurements from ground-based radar and rain gauges within the coverage area of the radar. The underwater measurement of rainfall shows similarities to radar measurements, but with intermediate spatial resolution and high temporal resolution. Reconciling these different measurement techniques provides understanding and confidence for all of the methods.