Abstract The Weather Surveillance Radar–1988 Dopplers (WSR‐88Ds) are an operational network of S‐band, dual‐polarization radars in the United States. Currently, replacement options are being considered for the next generation of weather radars. One option is dual‐polarization phased array radar (PAR), which employs electronic beam steering to provide faster volumetric updates. The Advanced Technology Demonstrator (ATD) is a research PAR that is being used to evaluate the feasibility of PAR technology. Presumably, the improved temporal resolution of the PARs will lead to more accurate radar‐based quantitative precipitation estimation (QPE). However, PARs require more complex calibration, especially for dual‐polarization observations. To test this hypothesis, the ATD and KOUN are used. From radar data over rain gauge locations, rainfall accumulations are calculated and compared to rain gauge observations. We show that, overall, PAR‐based QPE can perform 16.1% better than current WSR‐88Ds, in part due to the <60‐s low‐level sampling rate of the ATD.

Abstract This study evaluates the effectiveness of the Bi‐directional Optical Flow (BIO) method in capturing accumulated precipitation intensity and structure by China New Generation Weather Radar (CINRAD) operated in volume coverage pattern 21 (VCP21) scan mode, aiming to improve the accuracy of radar quantitative precipitation estimation (QPE). The BIO method was analyzed across 13 precipitation events, including three types of precipitation events: convective, typhoon, and stratiform precipitation events. High‐resolution X‐band phased array radar QPE data and rain gauge observations were used to validate the precipitation intensity and structure obtained using the BIO method. The results of elliptical parameter analysis demonstrated that the BIO method's precipitation showed high positional accuracy, precise alignment of precipitation orientation, and spatial distribution closely consistent with original X‐band phased array radar precipitation data. Statistical comparisons further highlighted that accumulated precipitation intensity of the BIO method (BIOQPE) significantly reduced errors compared to intensity from VCP21 mode, with an average decrease of 11.2% in RMSE, 14.0% in RMAE, and an average increase of 4.7% in CC across the 13 events. Particularly, BIOQPE provided superior accuracy and continuity in intensity and structure for convective and typhoon precipitation events. For stratiform precipitation, BIOQPE improved spatial continuity and demonstrated greater accuracy than original radar data. The results reveal that the BIO method is effective in describing precipitation intensity and structure within observations, and significantly improves the accuracy of accumulated precipitation, especially under limited temporal resolution conditions.

Abstract The study evaluates radar-based quantitative precipitation estimates (QPE) for ten extreme rain events that occurred between 2013 and 2019 in the Kansas City Metropolitan area, United States. These precipitation estimates were derived at hourly and approximately 0.5 km scales using two polarimetric QPE algorithms—one based on specific attenuation ( A ) and the other on specific differential phase ( K DP )—for the study area covered by two overlapping radars in Topeka, Kansas and Kansas City, Missouri. The polarimetric QPE assessment for extreme rain events was motivated by improved flood forecasting and precipitation frequency analysis. The analysis utilizes ground reference observations from a dense network of about 170 rain gauges over the study area to quantitatively assess the accuracy of these polarimetric rainfall ( R ) estimates. The comparison of R ( A ) and R ( K DP ) with the conventional algorithm based on radar reflectivity observations reveals that the two polarimetric algorithms outperform the reflectivity-based approach. While R ( K DP ) shows a systematic conditional feature (i.e., underestimation at high rain rates) with reduced scatter, R ( A ) appears to be less biased but with relatively large scatter. R ( A )’s significant overestimation for one of the extreme events was attributed to the misestimation of its key parameter (α), which resulted from hail contaminated data samples. To examine the observed underestimation tendency of R ( K DP ), we characterized the magnitude of underestimation (bias) with rainfall spatial variability as this variability may account for different rainfall regimes or the smoothing effect of K DP to reduce its inherent noisiness. Our result demonstrates that the underestimation tendency of R ( K DP ) becomes more pronounced as rainfall spatial variability increases.

Abstract Multi-Radar Multi-Sensor (MRMS) system precipitation products are relied upon for a wide range of important downstream applications relating to hydrometeorological impacts over different temporal and spatial scales. There are several challenges inherent to physically based precipitation estimation in complex terrain areas: observational siting issues, radar beam blockage effects, and highly variable rain-rate relationships in orographic forcing scenarios. A prior paper from the MRMS research group detailed the initial development and testing of a deep learning convolutional neural network (CNN) model to augment current precipitation estimation capabilities over the western United States. This study describes recent modifications made to improve the CNN model consistency of performance across event types. Additional numerical weather precipitation (NWP) model and terrain-related input variables have been added to the original 13 radar input variables, allowing the model to better capture warm rain processes and evaporative effects. Also, a custom loss function is applied to alleviate an overestimation bias seen in lighter precipitation events related to the use of the mean absolute error (MAE) loss metric in the initial model version. The updated model was tested over the year of 2021 along with a 15-day period with several high-impact atmospheric river events from the winter of 2022–23. Gauge-based evaluation of the 24-h quantitative precipitation estimation (QPE) fields shows that the new CNN model consistently outperforms both the MRMS radar-based QPE and the original version of the CNN model, building confidence in the eventual adoption of the CNN QPE within the MRMS operational framework. Significance Statement In this work, we look to refine a deep learning model for precipitation estimation in complex terrain areas with a focus on the western United States where observational limitations caused by the higher terrain inhibit physically based approaches. The initial version of the model showed improved accuracy relative to conventional methods. In the latest version of the model presented here, modifications were made to the input fields and loss function resulting in better consistency of performance compared to the initial version of the model. These results indicate that deep learning can likely provide critical improvements to the fidelity of estimated precipitation fields in areas with challenging terrain.

Abstract. The eastern Tibetan Plateau has a high elevation, with a cold and dry atmospheric background. The features of the raindrop size distributions (DSD) in this region have notable differences from those in the plains. The general empirical relationships for retrieving parameters of precipitation from radar observations are not applicable in the eastern Tibetan Plateau. In this study, we developed a new method based on optimal estimation theory to retrieve the vertical profiles of N0 and Dm from a Ka-band zenith-pointing Doppler radar. Validation by a field campaign during the summer of 2024 indicates that the mean bias in the log 10(N0) and Dm derived from the PARSIVEL2 disdrometer and the retrieved values are 0.12 and −0.1 mm respectively, demonstrating the effectiveness of the retrieved DSD parameters in this region. Based on the retrieved vertical profiles of DSD parameters, some unique characteristics are found. The heavy precipitation (the maximum value in the reflectivity profile exceeding 30 dBZ) exhibits a higher particle number concentration above 2 km and larger raindrop size in the bottom of the rainfall on average. The mean values of Dm above 2 km are approximately 0.5 mm, for heavy precipitation, the value increase as the raindrops fall, reaching a peak at around 0.5 km. Precipitation that occurs after the nighttime cooling usually has higher particle concentrations and smaller particle sizes. Based on the above research, empirical relationships for the quantitative precipitation estimates (QPE) and attenuation correction using Ka-band radar in the eastern Tibetan Plateau are established.

This study, conducted in the framework of the LIAISE field campaign in NE Spain (May–September 2021), investigates how near-surface relative humidity influences early-stage rainfall characteristics when precipitation is most affected by temperature and relative humidity before rainfall onset. Two instrumented sites were examined, using disdrometers, Micro Rain Radar (MRR), C-band weather radar data, and automatic weather stations. Rainfall events were first classified as stratiform or convective using weather radar data based on a texture analysis of the reflectivity field. Then, only stratiform events were selected and further classified into dry and moist categories according to the upper and lower terciles of near-surface (2 m) relative humidity at the rainfall onset (dry < 54%; moist > 72%). Results show that during dry events, the time delay between the detection of precipitation at ~750 m above ground level (AGL) (by MRR or C-band radar) and its arrival at the surface (measured by the disdrometer) is consistently longer than during moist events, indicating possible evaporation of raindrops during their descent. Surface drop size distributions also differ: dry cases have generally fewer small drops (with diameters < 0.8 mm) but relatively more large drops, leading to higher radar reflectivity values despite similar surface rainfall amounts. However, reflectivity observed aloft by C-band radar and MRR does not present the dependence on relative humidity found at ground level. Findings reported here increase our understanding of the impact of low-level conditions on precipitation characteristics and microphysical associated processes and may contribute to improve correction schemes in operational weather radar quantitative precipitation estimates.

Abstract Classification of snowflakes based on their geometric shape, degree of riming, and melt/dry state can improve understanding, characterization, and quantification of other geometrical, microphysical, and scattering properties of ice particles. For example, classification provides essential ground-truth data for interpreting polarimetric radar signatures of snow while validating and advancing radar-based quantitative precipitation estimation. High-resolution photographs of snowflakes obtained by emerging multicamera instruments are well suited for snowflake classification, which, coupled with recent machine learning techniques based on convolutional neural networks (CNNs), enable methods for accurate and fast automatic classification of snowflakes using images. Given that the appearance of a snowflake generally changes significantly with viewing angle, this work proposes and presents a novel multiview snowflake classification methodology based on the high-resolution photographs of frozen hydrometeors in free-fall from multiple views collected by the multicamera instruments. The approach employs machine/deep learning algorithms leveraging multiangle camera systems and enhanced supervised CNN-based techniques to achieve precise classification of snowflakes based on their geometrical categories and accurate and reliable estimates of specific snowflake properties, such as riming degree and melt/dry state. This represents the first multiview snowflake classification framework that takes full advantage of multiview camera systems. Presented multiview classification results show record accuracies of 98.57%, 98.22%, and 95.83% for geometric classes, riming degree, and melt/dry state, respectively. Significance Statement This work proposes and presents a novel multiview snowflake classification methodology leveraging recent developments in machine learning, multiview classification, and multiangle multicamera instruments for acquiring high-resolution photographs of frozen hydrometeors in free-fall from multiple views. The results for multiview classification show record accuracies for snowflake geometric classification, riming degree estimation, and melt/dry state estimation, respectively, significantly outperforming other classification models in each of the same categories. Automatic multiview machine learning–based winter hydrometeor classification enhances understanding, characterization, and quantification of geometrical, microphysical, and scattering properties of ice and snow hydrometeors. These improvements are essential for quantitative precipitation estimation algorithms and for microphysical parameterizations employed in numerical winter-weather forecast models and regional climate projections, with impacts on economy, safety, and everyday life.

Abstract Microphysical characteristics of warm‐rain precipitation that occurred in Fuzhou region during warm seasons of 2022 and 2023 have been investigated by using polarimetric radar data. Results of a modified warm‐rain identification algorithm indicate positive Z DR variation in the liquid layer should be added as a criterion to prevent events dominated by breakup‐coalescence balance being mistakenly classified as warm‐rain events, causing the inaccuracy of quantitative precipitation estimation (QPE). Comparative analysis suggests that stratiform, convective and warm‐rain precipitation are distinguishable in Cao and Zhang parameter space due to the nature of clustering within concentrated ranges. Convection during certain life stage exhibits similar feature as warm‐rain precipitation in Kumjian and Ryzhkov (KR) parameter space, whereas initial Z DR and vertical variations of Z H and Z DR could be useful to separate these two precipitation types. Vertical profiles of polarimetric variables ( Z H , Z DR , K DP ) in warm‐rain precipitation all increase toward the ground, which is associated with lower echo‐top and storm‐top freezing levels than convective precipitation. Microphsysical processes above the melting layer significantly influence the precipitation growth processes below according to analysis of two typhoon‐related cases. Some insights are gained to the development of a warm‐rain identification algorithm, such as monotonically increase of Z H and Z DR in the liquid layer and suitable range of initial Z DR , in addition, synoptic environmental conditions, e.g., vertical velocity, lifting condensation level and moisture flux, could serve as auxiliary conditions to accurately identify warm‐rain processes, but further research is needed to determine how to utilize them specifically.

• The GB radar QPE product suffers from underestimation and overestimation errors. • Underestimation errors are more frequent, especially for higher hourly rainfalls. • We propose a new QC framework to address errors in the GB radar QPE. • In comparison to gauge data, the QC reduces RMSE by 29 % and improves CC by 31 %. • The QC retains true extreme rainfalls within the GB radar QPE. High-quality gridded precipitation datasets are essential for climate and flood risk research. Quantitative Precipitation Estimation (QPE) from weather radars can provide high resolution gridded rainfall products. However, even after quality control, these datasets still suffer from underestimation (beam blockage and signal attenuation), and overestimation (radar malfunction, ground clutter, and electronic noise) errors, because most QC methods focus solely on radar reflectivity and lack a systematic and holistic approach. By comparing the radar QPE product for Great Britain (GB) (hourly, 1 km resolution, 2006 ∼ 2018), with hourly rain gauge records (∼1300), we find that radar QPE errors increase with elevation, distance from radar, and rainfall intensity. Radar QPE often underestimates high-intensity rainfalls and fails to detect many high-intensity rainfall events (≥40 mm h −1 ). Underestimation occur at 1.71 times as frequently as overestimation in radar QPE (≥0.2 mm h −1 ). We thus propose a QC framework to detect and correct beam blockages, to identify ‘normal’ and ’suspect’ rainfall fields, and to capture bad rainfalls within the radar QPE. We then use a Gaussian interpolation method to replace these bad rainfalls. Our results reveal that all GB radars suffer from beam blockage. Our novel QC framework reduces the RMSE of radar QPE compared to gauge observations from 0.546 to 0.386 (29 % reduction), increases the correlation coefficient from 0.552 to 0.725 (31 % increase), and retains real extreme rainfalls found in gauge observations. The QC framework requires minimal geographical and meteorological knowledge and can be implemented in various radar network settings, making it applicable for other regions globally.

Abstract When extreme hydrometeorological events threaten the United States, quantitative precipitation forecasts (QPFs) from the National Centers for Environmental Prediction’s Weather Prediction Center (WPC) provide critical decision support to help mitigate risks to life and property. This study builds on previous work by providing an updated benchmark of WPC QPF skill for extreme precipitation events from 2012 to 2024, through evaluating the 6- and 24-h accumulations for the day 1–3 lead times. Extreme precipitation is defined using three thresholds: the 99th and 99.9th percentile precipitation values of all wet-site days from 2012 to 2024 for each National Oceanic and Atmospheric Administration (NOAA) River Forecast Center (RFC) region and the 2-yr average recurrence interval from the NOAA Atlas 14. WPC forecasts are verified against Stage IV quantitative precipitation estimates. Forecast skill is assessed seasonally, regionally, and temporally for the 1200 UTC forecast cycle. Results show that WPC forecasts of extreme precipitation have improved over time. The highest skill is observed along the West Coast, where events are dominated by atmospheric rivers from late fall through early spring. Forecasts in these regions often retain skill through day 3. The East Coast shows the second-highest skill, particularly in fall and winter, when extreme precipitation occurs from tropical and synoptic systems. Skill is lowest in inland areas, especially in the summer months when precipitation systems are weakly forced, smaller in scale, and often underforecasted by WPC. These events rarely show skill after a day 1 lead time. Significance Statement Extreme precipitation remains a primary driver of billion-dollar disasters in the United States, underscoring the growing need for accurate quantitative precipitation forecasts. By benchmarking the skill of National Oceanic and Atmospheric Administration’s (NOAA) Weather Prediction Center forecasts of extreme precipitation, we can understand where regional and seasonal weaknesses exist and evaluate the current boundaries of precipitation predictability. These insights are essential for guiding future advancements in operational forecasting and model development.

The accurate retrieval of the raindrop size distribution (DSD) is a longstanding objective in meteorology because it underpins reliable quantitative precipitation estimation. Among remote sensors, weather radars are the primary tool for mapping DSD over wide areas, and phased-array systems in particular have demonstrated unique advantages owing to their high temporal and spatial resolution together with agile beam steering. Exploiting the underused high-resolution capability of an X-band phased-array radar, this study induced a Rainfall Regression Model (RRM). The RRM assumes a normalized gamma DSD model and retrieves its three parameters. It was then applied to a rain event influenced by the remnant circulation of Typhoon Haikui that affected Guangzhou on 8 September 2023. First, collocated disdrometer observations and T-matrix scattering simulations are used to build polynomial regressions between DSD parameters (D0, Nw, μ) and the polarimetric variables. Validation against independent disdrometer samples yields Nash–Sutcliffe efficiencies of 0.93 for D0 and 0.91 for log10Nw. The RRM is then applied to the full volumetric radar data. Horizontal maps reveal that the surface elevation angle consistently exhibited the largest standard deviation for all three parameters. A vertical profile analysis shows that large-drop cores (D0 > 2 mm) can reside above 2 km and that iso-value contours tilt rather than align vertically, implying an appreciable horizontal drift of raindrops within the complex remnant typhoon–monsoon wind field. By demonstrating the ability of X-band phased-array radar to resolve the three-dimensional microphysical structure of remnant typhoon precipitation, this study advances our understanding of the vertical characteristics of raindrops and provides high-resolution DSD information that can be directly ingested into severe weather monitoring and nowcasting systems.

Rainfall extremes observed by a weather radar in the northern tropical Andes

Radar constant miscalibration is one of the major sources of uncertainty in the radar-derived rainfall products. The traditional bias correction techniques often struggle to account for the non-stationary and nonlinear nature of bias. This study proposes a novel machine learning-based framework using the XGBoost algorithm to model reflectivity bias as a function of ground radar (GR) and space radar (SR) reflectivity differences, along with radar geometrical parameters (range, azimuth, and elevation). Furthermore, a strategy for near real-time bias correction is proposed based on an ensemble approach that combines an offline-pretrained model with an adaptive online learning component, incrementally updating the output as new data becomes available. This allows the model to adapt to evolving bias patterns over time. The results indicate that the proposed technique consistently outperforms the traditional iterative method. The results point towards its potential in reducing bias in near real-time for improved quantitative precipitation estimation (QPE) and other applications.

Abstract The upper Colorado River basin is the primary source of water for 40 million people. With declining snowpack in the basin, forecasting hydrological budgets in the Southwest United States is more important than ever. However, due in part, to a lack of reliable observations of precipitation in complex terrain, hydrological models struggle to assess and forecast snowpack snow water equivalent (SWE) in the upper Colorado River basin (UCRB). Therefore, the need for more reliable SWE forecasts in the UCRB motivated the U.S. Department of Energy Atmospheric Radiation Measurement Facility’s Surface Atmospheric Integrated Field Laboratory (SAIL) that occurred from June 2021 to June 2023. During SAIL, the X-band precipitation radar from Colorado State University conducted volume scans sampling the precipitation properties over the UCRB. The ARM facility developed a gridded Surface Quantitative Precipitation Estimates (SQUIRE) product from the radar observations. To do this, various daily SWE estimates from radar using the radar reflectivity factor Z e and specific differential phase K dp were compared against ground-based precipitation gauges. SWE in precipitation calculated from Wolfe and Snider’s S – Z e estimator was in best agreement with the rain gauges for the days when SWE < 12 mm. For days with SWE > 12 mm, the WSR-88D Intermountain West relationship had the best agreement with the precipitation gauges. Airborne snow depth observations show that SQUIRE captures regions of orographic enhancement in the mountains to the west and northwest of the SAIL study area, indicating that the scientific community should focus on understanding and ultimately simulating orographic atmospheric precipitation processes to improve UCRB snowpack SWE assessment and forecasting. Significance Statement This paper looks at snowfall estimates from data collected during a field experiment in the upper Colorado River basin, a critical source of water for 40 million people in the United States. Here, our typical radars do not detect precipitation in this crucial region, so we detail the methods we used to estimate snowfall rates from these new data that will help scientists better predict the amount of water that will be available for people living in the Southwest United States.

Thailand lies in a tropical monsoon region and is frequently affected by the decay of tropical storms during the rainy season, with 2-5 storms typically occurring each year.Accurate quantitative precipitation estimates (QPE) are therefore essential for assessing storm-related rainfall and associated flood risks.Radar-based rainfall estimation is particularly suitable but is prone to systematic bias arising from the radar reflectivity-rainfall (Z-R) relationship.This study develops a Geographic Information System (GIS)-based analytical approach to evaluate and compare Z-R relationships and to reduce bias between radar-estimated and gauge-observed rainfall.The analysis was conducted across lowland and mountainous areas in northern and central Thailand using data from the Phitsanulok Cband weather radar and 89 rain gauge stations during Tropical Storm Son-Tinh ( 2018).This study integrates radar data with Geographic Information Systems (GIS) to systematically compare multiple Z-R relationships alongside spatial bias correction, and to evaluate differences in rainfall estimation accuracy between lowland and mountainous areas.Three Z-R relationships Marshall-Palmer (MP), Rosenfeld Tropical (RF), and Summer Deep Convection (SD) were employed to generate event-based radar rainfall estimates.Spatial bias correction was conducted using the Inverse Distance Weighting (IDW) method, and accuracy was assessed through five-fold cross-validation.The results indicate that uncorrected radar rainfall estimates generally underestimate actual precipitation, whereas the IDWbased correction significantly reduces the Mean Field Bias (MFB) and improves estimation accuracy across diverse terrains.Among the three Z-R relationships, the Marshall-Palmer equation yielded the lowest errors, with a root mean square error (RMSE) of 17.517 mm and a mean absolute error (MAE) of 13.405 mm.The event-based spatial adjustment demonstrates that integrating an appropriate Z-R relationship with GIS-based bias correction substantially enhances radar QPE reliability, particularly in regions with complex topography.This framework offers practical value for hydrological applications and flood risk management in tropical monsoon regions.

Abstract Exploring lightning patterns alongside meteorological variables and fuel information helps diagnose the conditions under which lightning ignites fires. This study describes distributions of lightning, land surface, and meteorological conditions associated with known lightning ignitions in the western United States during 2020–22. Although most fire ignition lightning is classified as cloud-to-ground (89%), lightning misclassified as intracloud ignites 11% of fires. Findings indicate clear differences between the ignition and nonignition (null) lightning, and the fires detected on the day of versus those that holdover. Both the National Lightning Detection Network (NLDN) and Geostationary Lightning Mapper (GLM) observations help identify lightning more likely to ignite wildfires. On average, positive and negative polarity ignition flashes are 3.9 and 4.8 kA stronger than null flashes. Both GLMs indicate that ignition flashes are larger and 3–4 times brighter than the domainwide values. Many additional variables reveal a clear distinction between the ignition and null storm environments. The average hourly (daily) quantitative precipitation estimate is 7.08 (12.69) mm for the null flashes and only 2.76 (5.16) mm for ignitions. The Multi-Radar Multi-Sensor variables indicate weaker storms surrounding the ignitions, and several different relative humidity measures indicate ∼10% drier environments. Each variable shown indicates that fires reported on the day of ignition occur in the most fire-prone environments, and null cases represent the least conducive environments, with holdover fires somewhere in between. Our analysis shows that distributions of lightning, land surface, and meteorological conditions can help identify environments most vulnerable to lightning ignitions. Significance Statement This study describes the development and application of a database of lightning known to have ignited fires in the western United States during 2020–22. Our findings indicate clear differences between the ignition and nonignition (null) lightning, and the fires detected on the day of versus those that holdover. Each of the variables shown indicates that fires detected on the day of ignition occur in the most fire-prone environments, the null cases represent the least conducive environments, and the fires that holdover fall somewhere in between. Our analysis indicates that distributions of lightning, land surface, and meteorological conditions can help identify environments most vulnerable to lightning ignitions. This knowledge will help guide the development of automated applications for identifying the most impactful lightning collocated with conditions conducive to wildfire ignition.

Abstract Assessing the performance of a limited-area model (LAM) based on an emerging dynamical core such as the Finite-Volume Cubed-Sphere Dynamical Core (FV3) in regional forecasting is essential to advancing existing operational systems. In this study, we configure an FV3-LAM using the Short-Range Weather Application and evaluate its performance in forecasting extreme precipitation over Taiwan, referred to as the SRW-TW experiment. We conduct the simulation of SRW-TW over 10 selected cases of different rainfall types between 2019 and 2022 and then compare the simulated rainfall with the quantitative precipitation estimation (QPE) and segregation using multiple sensors (QPESUMS) data and WRF-based forecasts as the ground reference and baseline product, respectively. By examining the simulated rainfall patterns and various performance metrics, we find that SRW-TW exhibits slightly higher correlation coefficient and critical success index values relative to the WRF baseline, although the differences are not statistically significant. However, SRW-TW performs slightly worse than the WRF baseline in terms of bias and frequency bias metrics. Our findings indicate that SRW-TW generates more realistic rainfall patterns but tends to underestimate precipitation, particularly in southwesterly flow-related events. Additionally, forecasting typhoon rainfall and hourly scale precipitation remains challenging, requiring further model refinement and the potential development of a bias-correction scheme.

Abstract Reliable rainfall estimation is essential for hydrological modeling, particularly as climate change intensifies rainfall extremes and challenges water resource management. In many intertropical basins, sparse observation networks limit quantitative precipitation estimation, making satellite precipitation products (SPPs) a key alternative. However, SPP performance varies geographically and must be assessed against high-resolution reference data. This study first evaluates five state-of-the-art SPPs at their native resolution (0.1°, 30 min) against high-resolution weather radar observations in French Guiana. Second, it introduces a correction framework that combines image classification with tailored bias adjustment. Rain fields are first grouped into clusters using k -means applied to their spatial features, distinguishing different rainfall structures. Within each cluster, a quantile matching by parts (QMP) correction scheme trained on weather radar data is applied. The correction scheme is first adjusted on a training dataset consisting of 4000 coincident radar and satellite rain maps. Then, it can be applied to other satellite rain maps and outside the radar coverage. The main findings are as follows: 1) Among the evaluated products, Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (GPM) (IMERG) Final shows the best performance, but IMERG Late is selected for application due to its shorter latency. 2) The cluster-specific QMP correction reduces SPP bias from −20% to +4%, improves the rainfall intensity distribution, and improves the spatial variability of the rain fields and the diurnal cycle. 3) In hydrological simulations of the Mana River basin, the corrected product improves the Kling–Gupta efficiency from 0.36 to 0.83 compared to benchmark simulations using radar data. Overall, this relatively simple and computationally light method improves satellite-based rainfall estimation for various potential applications in data-scarce tropical regions, offering a scalable solution as radar networks expand globally.

Quantitative Precipitation Estimation Based on S-Band Dual-Polarized Radar Observations over Coastal Eastern China during the Meiyu Season

Abstract The surface rainfall regions of convective systems do not always align spatially with the overlying cold‐cloud shields, posing challenges to satellite quantitative precipitation estimates. A novel index of convective rainfall deviation is defined in this study to quantify the spatial discrepancy between surface rainfall and cold‐cloud shields, utilizing the FY‐4A brightness temperature data and surface rainfall observation data from across China. The results show that systems with larger deviations are more prevalent in southern China. For small‐deviation systems, surface rainfall closely coincides with the coldest center of the cloud shields. However, in medium‐deviation systems, rainfall occurs southwest of the coldest center, with relatively expanded but weakened rainfall beneath the cloud shields. Retaining a pronounced southwestern deviation signature, large‐deviation systems also present a secondary northeastern rainfall center. This complex spatial pattern is associated with increased cloud tilting driven by the configuration of environmental updrafts and low‐level southwesterly wind shears.