Rain Rate Retrieval Algorithms Research Articles

Characterization of Raindrop Size Distribution over Santa Clara Valley

This study presents a year-long (January 2019–April 2020) analysis of the Z–R relationship and drop size distribution (DSD) scaling parameters for size, concentration, and shape of rain events over Santa Clara Valley, CA. External influences were analyzed based on synoptic variability and seasons. For the former, 850 hPa winds were separated into groups based on direction and magnitude. Results show that greater drop size, lower concentration, and larger shape parameters for spring, while winter and fall showed smaller drop sizes, higher concentrations, and smaller shape parameters. For synoptic variability, southeasterly-to-southwesterly flow was associated with larger drop sizes, larger concentrations, and smaller shape parameters relative to northwesterly flow. Differences in the DSD scaling parameter values and Z–R relationship were also observed between strong and weak low-level flow. The results of this study suggest that it is beneficial to derive specific microphysical relationships based on seasons and different synoptic events to improve radar rain rate retrieval algorithms using the Z–R relationship.

Estimation of rain parameters for microwave backscattering model using PSO

The intention of the geophysical modelling of rain is to provide a better explanation for the effect of rainfall to the microwave backscattering and thus to interpret radar measurements. However, in the model, physical characteristic of raindrops should be estimated primarily and accurately by considering observation system, measurements and suitable rain rate retrieval algorithms to calculate backscattering coefficients from rainfall. In this study, a geophysical microwave backscattering model of rain type precipitation over sea surface is constructed by using Particle Swarm Optimization (PSO) algorithm in the multilayered Vector Radiative Transfer (VRT) model to estimate vertical profile of rain by using GPM DPR data. Rain column is partition into sublayers and for each sublayer, physical properties of raindrops such as drop radius, water volume fraction or layer thickness are estimated by using PSO to provide the best fit with measurements by searching within certain limits defined by rain rate. Backscattering coefficients from entire rain is provided by the solution of VRT equations via Matrix Doubling Method to consider multilayer effect. Results show that, vertical profile of rain parameters can be estimated accurately for moderate /high rain rates (up to 11–12 mm/h) by using presented model.

Rain Rate Retrieval Algorithm for Dual-Polarized Sentinel-1 SAR in Tropical Cyclone

Rain Rate Retrieval Algorithm for Dual-Polarized Sentinel-1 SAR in Tropical Cyclone

Advances on Rain Rate Retrieval from Satellite Platforms using Artificial Neural Networks

In the last two decades, great advances have been related with the development of rain rate retrieval algorithms using artificial neural networks, in order to exploit satellite data capabilities. The enhancement of computing processing capacity available from modern computers has impulsed a long number of researches aimed to generate more accurate and faster algorithms. This work deals with how the implementation of new trends in artificial neural networks and the spectral resolution improvement of spaceborne sensors have influenced in the design of retrieval algorithms to estimate rain rate from satellites using artificial neural networks. Recent results have shown an important increasing in accuracy and technical feasibility of implementation, however, the feasibility to use artificial neural networks to estimate rain rate in real time, using remote sensing techniques, is a research issue yet.

Rain Rate Retrieval Algorithm for Conical-Scanning Microwave Imagers Aided by Random Forest, RReliefF, and Multivariate Adaptive Regression Splines (RAMARS)

This paper proposes a rain rate retrieval algorithm for conical-scanning microwave imagers (RAMARS), as an alternative to the NASA Goddard profiling (GPROF) algorithm, that does not rely on any a priori information. The fundamental basis of the RAMARS follows the concept of the GPROF algorithm, which means, being consistent with the Tropical Rainfall Measuring Mission (TRMM) precipitation radar rain rate observations, but independent of any auxiliary information. The RAMARS is built upon the combination of state-of-the-art machine learning and regression techniques, comprising of random forest algorithm, RReliefF, and multivariate adaptive regression splines. The RAMARS is applicable to both over ocean and land as well as coast surface terrains. It has been demonstrated that, when comparing with the TRMM Precipitation Radar observations, the performance of the RAMARS algorithm is comparable with the 2A12 GPROF algorithm. Furthermore, the RAMARS has been applied to two cyclonic cases, hurricane Sandy in 2012, and cyclone Mahasen in 2013, showing a very good capability to reproduce the structure and intensity of the cyclone fields. The RAMARS is highly flexible, because of its four processing components, making it extremely suitable for use to other passive microwave imagers in the global precipitation measurement (GPM) constellation.

Application of lightning to passive microwave convective and stratiform partitioning in passive microwave rainfall retrieval algorithm over land from TRMM

This study analyzes relationships between lightning flash rate, radar reflectivity factor (reflectivity), and passive microwave brightness temperature (Tb) for convective and stratiform precipitation over land using multiyear data from the Tropical Rainfall Measuring Mission (TRMM) satellite. A new convective and stratiform index (CSI (an estimate of convective areal fraction)) for the TRMM Microwave Imager (TMI) is developed from the analysis. Four years of TRMM TMI, Lightning Imaging Sensor (LIS), and Precipitation Radar (PR) data (2002–2005) are colocated and remapped to 0.1 and 0.05 degree grids for analysis. The scientific objective of this study is to understand the relationship between lightning and active and passive microwave precipitation observations and explore ways of using lightning information to enhance the discrimination between convective and stratiform precipitation in TMI rain rate retrieval algorithm. PR provides the reference convective and stratiform classification and is coincident with LIS which reports lightning parameters such as the occurrence (yes or no) and flash rates. Analysis of ∼14 million coincident precipitating TRMM measurements over land (i.e., excluding oceans and coasts) reveals that 6% of rain data have lightning flash rates greater than zero. For all lightning data, 60% have 0–1 fl min−1, 28% have 1–2 fl min−1, and 12% have flash rates greater than 2 fl min−1. Overall, 86.5% (13.5%) of lightning occurred in convective (stratiform) precipitation. In other words, stratiform rainfall is predominant when LIS detects no lightning, and the convective rain probability increases with increasing lightning frequency. For example, 34% of rainfall is convective for low flash rates (0–1 fl/min), whereas the convective probability increases to 99.7% for high flash rates (>=2 fl/min). This study develops a simple method that incorporates lightning into the CSI to test if lightning can help passive microwave (PM) delineate convective and stratiform precipitation. LIS lightning occurrence and flash rates (i.e., no flash, 0–1 fl/min, 1–2 fl/min, and >2 fl/min) are used to preclassify TMI Tbs into four groups of increasing convective probability. Multivariable linear regression is then applied to each group to derive the CSI. Results reveal that lightning information primarily improves the identification of highly convective rainfall by correctly shifting microwave observations previously identified as moderate to highly convective. Alternatively, the absence of lightning also helps PM to identify likely stratiform. Overall, including lightning information results in a decrease of bias error of 6% and a small increase in RMS error of 4.5% on the entire range of rainfall rates.

Applicability of the Iterative Backward Retrieval Method for the GPM Dual-Frequency Precipitation Radar

The Dual-frequency Precipitation Radar (DPR) on the core satellite of the Global Precipitation Measurement mission will measure the radar reflectivity factor in the Ku- and Ka-bands. A rain-rate retrieval algorithm that does not require a surface reference was developed (called the MA04 method). However, MA04 cannot give the true solution in some cases of heavy rainfall. MA04 is a simplified version of the iterative backward retrieval method (IBRM), and the IBRM is equivalent to the forward retrieval method with a constraint. The purpose of this study is to clarify the essential conditions under which the IBRM and MA04 can give the true solution (the conditions are referred to as “the applicability” as in the title). For the purpose, DPR measurements are simulated under simplified assumptions. The applicability of the IBRM and MA04 is closely related to the magnitude of internal attenuation. The upper limit of rain rate for which the IBRM can obtain a true solution is 10 to 20 mm · h <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-1</sup> if the internal attenuation occurs between the top and middle of the target range bin and the vertical resolution is 0.25 km. The upper limit of rain rate for which MA04 can obtain a true solution is dependent on the number of range bins, and it is 24 to 36 mm · h <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-1</sup> in the case of n = 12. MA04 can apply a wider range of rainfall than the IBRM because MA04 tends to select the solution with the smallest attenuation among possible solutions.

A Rain-Rate Retrieval Algorithm for Attenuated Radar Measurements

Abstract A dynamic regularization scheme for rain-rate retrievals from attenuated radar measurements is presented. Most regularization techniques, including the optimal estimation method, use the state-space parameters to regularize the problem, which will always lead to a bias in the solution. To avoid this problem the authors introduce an evolutionary regularization technique, which is based on the spatial derivative of the measured reflectivity profile and allows for a bias-free global solution. The regularization strength is determined by the quadratic eigenvalue solution using the regularized total least squares method. With the new method, the authors perform a retrieval of rain-rate profiles from simulated measurements of a nadir-pointing W-band (94 GHz) radar, in a configuration similar to the cloud radar employed on CloudSat. The simulations assume that multiple scattering is negligible and only liquid hydrometeors are taken into account. The authors compare the results of this method with the outcome of an optimal estimation method and demonstrate that their method is superior in terms of reliability, correlation coefficient, and dispersion to the optimal estimation method for layers experiencing high values of attenuation; therefore, the a priori bias typical for optimal estimation solutions is avoided.

Rainfall retrieval over Indian land and oceanic regions from SSM/I microwave data

Present study focuses on the estimation of rainfall over Indian land and oceanic regions from the Special Sensor Microwave/Imager (SSM/I) on the Defense Meteorological Satellite Program (DMSP) F-13. Based on the measurements at 19.35, 22.235 and 85.5 GHz channels of SSM/I Satellite, scattering index (SI) has been developed for the Indian land and oceanic regions separately. These scattering indices were co-located against rainfall from Precipitation Radar (PR) onboard Tropical Rainfall Measuring Mission (TRMM) to develop a new regional relationship between the SI and the rain rate for the Indian land and oceanic regions. A non-linear fit between the rain rate and the SI is established for rain measurement. In order to have confidence in our method, we have also estimated rainfall using the global rainfall and scattering index relationship developed by Ferraro and Marks [Ferraro, R.R., Marks, G.F. The development of SSM/I rain rate retrieval algorithms using ground based radar measurements. J. Atmos. Ocean. Technol. 12, 755–770, 1995]. The validation with the rain-gauge shows that the present scheme is able to retrieve rainfall with better accuracy than that of Ferraro and Marks (1995). Further intercomparison with TRMM-2A12 and validation with rain-gauges rainfall showed that the present algorithm is able to retrieve the rainfall with reasonably good accuracy.

An Algorithm for Real-Time Rainfall Rate Estimation by Using Polarimetric Radar: RIME

Abstract Polarimetric radars provide measurements that describe the shape and dimensions of hydrometeors and are unaffected by calibration, attenuation, and the presence of ice. These measurements can potentially lead to a more detailed description of hydrometeors and to an improvement in quantitative rainfall rate estimation. The authors present an algorithm that exploits polarimetric measures for rain-rate estimation and investigate its application in a real-time framework by using measurements from the C-band polarimetric radar at Monte Settepani in Savona, Italy. It is based on a flowchart decision tree that allows the use of the best rain-rate retrieval algorithm, depending on the value of polarimetric variables. The methodology was applied to a real-time framework for more than a year, and the results were presented for all the significant events observed during the test period. To evaluate the performance of the algorithm, a comparison was made with rain gauge observation from a dense regional network. The performances of the algorithm were compared with those obtained by standard operational Z–R formulations to evaluate the benefit of this approach for operational applications.

Rainfall-Induced Changes in Actual Surface Backscattering Cross Sections and Effects on Rain-Rate Estimates by Spaceborne Precipitation Radar

Abstract In this study, the authors used Tropical Rainfall Measuring Mission precipitation radar (TRMM PR) data to investigate changes in the actual (attenuation corrected) surface backscattering cross section (σ0e) due to changes in surface conditions induced by rainfall, the effects of changes in σ0e on the path integrated attenuation (PIA) estimates by surface reference techniques (SRTs), and the effects on rain-rate estimates by the TRMM PR standard rain-rate retrieval algorithm. Over land, σ0e is statistically higher under rainfall than under no rainfall conditions (soil moisture effect) unless the land surface is densely covered by vegetation. Over ocean, the dependence of σ0e on the incident angle differs under rainfall and no-rainfall conditions (wind speed effect). The alongtrack spatial reference (ATSR) method, one of the SRTs used in the standard algorithm, partially considers these effects, while the temporal reference (TR) method, another SRT, never involves these effects; its PIA estimates thus have negative biases over land. In the hybrid spatial reference (HSR) method used over ocean, different incident angles create different biases in PIA estimates. If the TR method is replaced by the ATSR method, the monthly rainfall amount in July 2001 all over the land within the TRMM coverage increases by 0.70%. The bias in the HSR method over ocean can be mitigated by fitting a σ0–θ curve separately to smaller incident angles and to larger incident angles. This improvement increases or decreases the monthly rainfall amounts in individual incident angle regions by up to 10%.

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.

Rain/No-Rain Classification Methods for Microwave Radiometer Observations over Land Using Statistical Information for Brightness Temperatures under No-Rain Conditions

Abstract One of the goals of the Global Precipitation Measurement project, the successor to the Tropical Rainfall Measuring Mission (TRMM), is to produce a 3-hourly global rainfall map using several spaceborne microwave radiometers. It is important, although often difficult, to classify radiometer observations over land as either “rain” or “no rain” because background land surface conditions change significantly with time and location. In this study, a no-rain brightness temperature database was created to infer land surface conditions using simultaneous observations by TRMM Microwave Imager (TMI) and precipitation radar (PR) with a resolution of 1 month and 1° latitude × 1° longitude. This paper proposes new rain/no-rain classification (RNC) methods that use the database to determine the background brightness temperature. The proposed RNC methods and the RNC method developed for the Goddard profiling algorithm (GPROF; the standard rain-rate retrieval algorithm for TMI) are applied to all TMI observations for the entire year of 2000, and the results are evaluated against the RNC made by PR as the “truth.” The first method (M1) simply uses the average brightness temperature at 85-GHz vertical polarization [denoted as TB (85 V)] under no-rain conditions as the background brightness temperature at 85-GHz vertical polarization [denoted as TBe (85 V)]. The second method (M2) uses a regression equation between TB (85 V) and TB (22 V) under no-rain conditions from the database. Here, TBe (85 V) is calculated by substituting the observed TB (22 V) into the regression equation. The ratio of accurate rain detection by GPROF to all rain occurrences detected by PR was 59%. This ratio was 57% for M1 and 63% for M2. The ratio with the weight of the rain rate was 81% for M1 and 86% for M2; it was 80% for GPROF. These comparisons were made by setting a threshold using a constant coefficient k0 to make the ratio of false rain detection to all no-rain occurrences detected by PR almost the same (approximately 0.85%) for all three methods. Further comparisons among the methods are made, and the reasons for the differences are investigated herein.

Oceanic rain rate estimates from the QuikSCAT Radiometer: A Global Precipitation Mission pathfinder

The SeaWinds scatterometer, launched onboard the QuikSCAT satellite in 1999, measures global ocean vector winds. In addition to measuring radar backscatter, SeaWinds simultaneously measures the microwave brightness temperature of the atmosphere/surface, and this passive microwave measurement capability is known as the QuikSCAT Radiometer (QRad). This paper presents a QRad retrieval algorithm used to infer instantaneous oceanic rain rates. This statistical algorithm is trained using near‐simultaneous observations of major rain events by QRad and the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI). Rain rate retrieval algorithm validation is presented through comparisons with independent rain measurements from the TMI 2A12 surface rain rates and the TRMM 3B42RT composite microwave and visible and infrared near‐real time data product. Results demonstrate that QRad rain rate measurements are in good agreement with these independent microwave rain observations and superior to the visible/infrared rain estimates. Thus the QRad rain measurement time series is a valuable addition to the oceanic precipitation climatology that can be used to improve the diurnal estimation of the global rainfall, which is a goal for the future Global Precipitation Mission program. Moreover, the availability of QRad data will provide GPM users early access to learn to use less‐precise rain measurements that will occur in the GPM era with the use of less‐capable constellation satellites. Finally, these QRad rain estimates will be available in the planned data reprocessing (FY 2006) of QuikSCAT winds to improve the rain flagging of rain‐contaminated oceanic wind vector retrievals.

Quantitative estimation of Tropical Rainfall Mapping Mission precipitation radar signals from ground‐based polarimetric radar observations

The Tropical Rainfall Mapping Mission (TRMM) is the first mission dedicated to measuring rainfall from space using radar. The precipitation radar (PR) is one of several instruments aboard the TRMM satellite that is operating in a nearly circular orbit with nominal altitude of 350 km, inclination of 35°, and period of 91.5 min. The PR is a single‐frequency Ku‐band instrument that is designed to yield information about the vertical storm structure so as to gain insight into the intensity and distribution of rainfall. Attenuation effects on PR measurements, however, can be significant and as high as 10–15 dB. This can seriously impair the accuracy of rain rate retrieval algorithms derived from PR signal returns. Quantitative estimation of PR attenuation is made along the PR beam via ground‐based polarimetric observations to validate attenuation correction procedures used by the PR. The reflectivity (Zh) at horizontal polarization and specific differential phase (Kdp) are found along the beam from S‐band ground radar measurements, and theoretical modeling is used to determine the expected specific attenuation (k) along the space‐Earth path at Ku‐band frequency from these measurements. A theoretical k‐Kdp relationship is determined for rain when Kdp ≥ 0.5°/km, and a power law relationship, k = a Zhb, is determined for light rain and other types of hydrometers encountered along the path. After alignment and resolution volume matching is made between ground and PR measurements, the two‐way path‐integrated attenuation (PIA) is calculated along the PR propagation path by integrating the specific attenuation along the path. The PR reflectivity derived after removing the PIA is also compared against ground radar observations.

Rainfall Retrieval and Flooding Monitoring in China using TRMM Microwave Imager(TMI).

The ability of the Tropical Rainfall Measuring Mission Microwave Imager (TRMM/TMI) for rainfall and flooding detection has been demonstrated in this study. A dynamic cluster K-mean method was presented to classify surface types and rain areas.Based on the TMI measurements and surface observations during the Huaihe River Basin Energy and Water Cycle Experiment, the scattering index over land (SIL) was developed and a rain rate (RR) retrieval algorithm of TMI was established, in which the RR is the function of SIL and PCT85 (polarization corrected temperature at 85 GHz). Compared with the rainfall derived from the weather radar and the products of TMI(2A12) and TRMM/PR(2A25), the given surface classification method and the rain rate retrieval algorithm can provide reasonable estimates of rain areas and rain rates.On the basis of TMI measurements, for methods, the classification method, the soil wetness index (SWI) method, the polarization difference index (PDI) method, and the polarization ratio index (PRI) method, were utilized to monitor flooding and study soil wetness in the Yangtze River basin and the Huaihe River basin during summer 1998. Compared with the images provided by L-band Synthetic Aperture Radar (L-SAR) and Radar Satellite (Radarsat) and the figures derived from daily rain gauge data based on the Z-index method, the four methods of flooding detection were proved to be successful, among which the PRI method was best.

Mountain reference technique: Use of mountain returns to calibrate weather radars operating at attenuating wavelengths

The Mountain Reference Technique (MRT) was proposed as a means to perform a self‐calibration of a weather radar system operating at an attenuating wavelength in a mountainous environment. Two convective rain events observed during the Grenoble 97–98 Experiment are selected here for an illustration and a further verification of the method at X band: the June 16, 1997, event is a medium event with maximum path‐integrated attenuations (PIAs) of about 15 dB over a 9‐km path, while the July 3, 1998, event is quite extraordinary with (1) a maximum PIA of 50 dB over the same distance and (2) the temporary presence of hail. An improved scheme is proposed for the MRT parameter estimation procedure with a more satisfactory treatment of such high‐attenuation effects. It is shown that the optimal calibration factors obtained for the two rain events are almost equal to each other, a comforting result with respect to the radar equipment stability during the corresponding 1‐year period. Although the MRT is based on reflectivity and attenuation measurements only, validation of the rain rate retrieval algorithms with respect to rain gage data for the June 16, 1997, rain event showed that this technique is relevant in terms of rain rate estimation. In particular, the MRT‐calibrated Hitschfeld‐Bordan algorithm provides satisfactory results for this medium‐attenuation event. However, the July 3 case clearly demonstrates that this algorithm cannot correct for such high‐attenuation effects because of its inherent instability. For both rain events the Marzoug‐Amayenc algorithm, originally proposed for spaceborne configurations, is found to be stable and efficient in terms of rain rate estimation. These interesting features are counterbalanced by the fact that the algorithm implementation is limited to directions for which PIA measurements are actually available.

The Contribution to Tropical Rainfall with respect to Convective System Type, Size, and Intensity Estimated from the 85-GHz Ice-Scattering Signature

This study compiled a database of precipitating cloud clusters from 85-GHz data in 10 regions of the wet Tropics for a calendar year (November 1992–October 1993). The cloud clusters were grouped into four classes of basic system types, based on size (closed 250 K contour greater or less than 2000 km2) and minimum enclosed 85-GHz brightness temperature (greater or less than 225 K) to indicate the presence or absence of large areas of active, deep convection. For each cloud cluster, instantaneous volumetric rain rates (mm km2 h−1) were calculated using an 85-GHz ice-scattering-based rain-rate retrieval algorithm. Because the ice-scattering signature is linearly related to but does not directly measure rain rate, the methodology was appropriate for estimating relative contributions rather than quantifying tropical rainfall. For the 3-month wet season of each study region, the rainfall contributions with respect to system type, size, and intensity were calculated. Regional differences were small among the contributions with respect to system type and to precipitating area. Although mesoscale convective systems constituted 10%–20% of the regional populations, they contributed 70%–80% of the rainfall. With respect to cloud cluster area, the top 10% of cloud cluster areas contributed more than 70% of the rainfall, and the top 1% (greater than 20 000 km2) contributed about 35% of the total rainfall. Regional differences were apparent in the distributions of rainfall contribution with respect to minimum brightness temperature. The Amazon’s distribution more closely resembled the oceanic distributions than the continental distributions. The distributions of the oceanic regions peaked at 200 K, and over half of the rain in the oceanic regions was contributed by the fewer than 20% of the cloud clusters colder than 210 K. Distributions in the continental regions peaked at 175 K. A total of 70%–80% of the rain was contributed by the 20%–30% of continental cloud clusters colder than 200 K, with nonnegligible contributions from a small number of cloud clusters colder than 120 K. Sub-Saharan Africa had the largest contribution from cloud clusters colder than 120 K.

Comparison of Radar Rainfall Retrieval Algorithms in Convective Rain during TOGA COARE

The authors compare deterministic and stochastic rain-rate retrieval algorithms by applying them to 14-GHz nadir-looking airborne radar reflectivity profiles acquired in tropical convective rain during the Tropical Ocean Global Atmosphere Coupled Ocean-Atmosphere Response Experiment, The deterministic algorithms both use the path-integrated attenuation (PIA), measured by the surface reference technique, as a constraint. One deterministic algorithm corrects the k-R relation, while the second corrects the Z-R relation. The stochastic algorithms are based on applying an extended Kalman filter to the reflectivity profile. One employs radar reflectivity only; the other additionally uses the PIA. The authors find that the stochastic algorithm, which uses the PIA, is the most robust algorithm with regard to incorrect assumptions about the drop size distribution (DSD). The deterministic algorithm that uses the PIA to adjust the Z-R relation is also fairly robust and produces rain rates similar to the stochastic algorithm that uses the PIA, The deterministic algorithm that adjusts only the k-R relation and the stochastic radar-only algorithm are more sensitive to assumptions about the DSD. It is likely that they underestimate convective rainfall, especially if the DSD is erroneously assumed to be appropriate for stratiform rain conditions.

An Automated Neural Network Cloud Classifier for Use over Land and Ocean Surfaces

Abstract An automated neural network cloud classifier that functions over both land and ocean backgrounds is presented. Motivated by the development of a combined visible, infrared, and microwave rain-rate retrieval algorithm for use with data from the 1997 Tropical Rainfall Measuring Mission (TRMM), an automated cloud classification technique is sought to discern different types of clouds and, hence, different types of precipitating systems from Advanced Very High Resolution Radiometer (AVHRR) type imagery. When this technique is applied to TRMM visible–infrared imagery, it will allow the choice of a passive microwave rain-rate algorithm, which performs well for the observed precipitation type, theoretically increasing accuracy at the instantaneous level when compared with the use of any single microwave algorithm. A neural network classifier, selected because of the strengths of neural networks with respect to within-class variability and nonnormal cluster distributions, is developed, trained, and teste...