Precipitation Retrieval Research Articles

An Analysis for the Applicability of Global Precipitation Measurement Mission (GPM) IMERG Precipitation Data in Typhoons

This study selected examples of 17 typhoons that landed in Fujian after passing through Taiwan. The study evaluated the precipitation in different time scales and the spatial distribution of daily precipitation of varying magnitudes in the southeastern coastal area by comparing satellite precipitation estimation products with meteorological observation station data. The evaluation used a correlation coefficient, mean relative error, relative bias, and graded assessment indexes (probability of detection, false alarm rate, and critical success index). Correlation coefficient analysis revealed that maximum daily precipitation performed best, followed by process total precipitation. The relative bias indicates that the precipitation estimated by the satellite is lower than the rainfall recorded by the automatic weather station. Mean relative error analysis showed that hourly precipitation had the highest error, followed by maximum daily precipitation. The GPM IMERG precipitation products’ retrieval of daily precipitation of varying magnitudes was assessed using three indicators. The assessment revealed that the satellite had a low under-reporting rate for light rain events but a high under-reporting rate for torrential rain events, especially extremely heavy rainstorm events, in terms of probability of detection. For the false alarm rate, the satellite had a small probability of false predictions for light rain events, while extremely heavy rainstorm events had the highest probability. For the critical success index, the satellite’s estimation of light rain events was basically consistent with reality; however, its ability to estimate precipitations above rainstorm levels was low. The results of the spatial assessment of heavy precipitation show that the satellite’s ability to detect heavy precipitation’s structure, intensity, and location is fair and has some reference value, especially for regions where conventional information is scarce.

A unified synergistic retrieval of clouds, aerosols, and precipitation from EarthCARE: the ACM-CAP product

Abstract. ACM-CAP provides a synergistic best-estimate retrieval of all clouds, aerosols, and precipitation detected by the atmospheric lidar (ATLID), cloud-profiling radar (CPR), and multi-spectral imager (MSI) aboard EarthCARE (Earth Cloud Aerosol and Radiation Explorer). While synergistic retrievals are now mature in many contexts, ACM-CAP is unique in that it provides a unified retrieval of all hydrometeors and aerosols. The Cloud, Aerosol, and Precipitation from mulTiple Instruments using a VAriational TEchnique (CAPTIVATE) algorithm allows for a robust accounting of observational and retrieval errors and the contributions of passive and integrated measurements and for enforcing physical relationships between components (e.g. the conservation of precipitating mass flux through the melting layer). We apply ACM-CAP to EarthCARE scenes simulated from numerical weather model forecasts and evaluate the retrievals against “true” quantities from the numerical model. The retrievals are well-constrained by observations from active and passive instruments and overall closely resemble the bulk quantities (e.g. cloud water content, precipitation mass flux, and aerosol extinction) and microphysical properties (e.g. cloud effective radius and median volume diameter) from the model fields. The retrieval performs best where the active instruments have a strong and unambiguous signal, namely in ice clouds and snow, which is observed by both ATLID and CPR, and in light to moderate rain, where the CPR signal is strong. In precipitation, CPR's Doppler capability permits enhanced retrievals of snow particle density and raindrop size. In complex and layered scenes where ATLID is obscured, we have shown that making a simple assumption about the presence and vertical distribution of liquid cloud in rain and mixed-phase clouds allows improved assimilation of MSI solar radiances. In combination with a constraint on the CPR path-integrated attenuation from the ocean surface, this leads to improved retrievals of both liquid cloud and rain in midlatitude stratiform precipitation. In the heaviest precipitation, both active instruments are attenuated and dominated by multiple scattering; in these situations, ACM-CAP provides a seamless retrieval of cloud and precipitation but is subject to a high degree of uncertainty. ACM-CAP's aerosol retrieval is performed in hydrometeor-free parts of the atmosphere and constrained by ATLID and MSI solar radiances. While the aerosol optical depth is well-constrained in the test scenes, there is a high degree of noise in the profiles of extinction. The use of numerical models to simulate test scenes has helped to showcase the capabilities of the ACM-CAP clouds, aerosols, and precipitation product ahead of the launch of EarthCARE.

Evaluation and Applicability Analysis of GPM Satellite Precipitation over Mainland China

This study aims to systematically evaluate the accuracy and applicability of GPM satellite precipitation products (IMERG-E, IMERG-L, and IMERG-F) with varying time lags at different spatial and temporal scales over mainland China. We use quantitative statistical indicators, including correlation coefficient (CC), root mean square error (RMSE), mean absolute error (MAE), mean daily precipitation, probability of detection (POD), false alarm rate (FAR), bias, and equitable threat score (ETS), based on observations from 2419 national gauge sites. The results show that GPM satellite precipitation products perform well in eastern and southern humid regions of China, with relatively poorer performance in western and northern regions in terms of spatial distribution. It reflects the sensitivity of GPM precipitation retrieval algorithm to climate and precipitation type, topography, density, and quality of ground observation across different latitudes. Despite the design of GPM for different forms of precipitation, IMERG products perform the best in summer and the worst in winter, indicating that estimating snowfalls via satellite is still challenging. In terms of precipitation intensity, IMERG products significantly improve performance for light and no rain (POD ≥ 0.7), but errors gradually increase for moderate, heavy, and torrential rain, due to the saturation tendency of satellite echoes. Overall, we comprehensively evaluate the IMERG products, revealing the distinct characteristics at various spatial–temporal scales focusing on rainfall accumulations over mainland China. This study provides an important reference for other similar satellite-based precipitation products. It also helps the parameter optimization of hydrological modelling, especially under extreme precipitation conditions, to enhance the accuracy of flood simulation.

Research on the Monitoring Ability of Fengyun-Based Quantitative Precipitation Estimates for Capturing Heavy Precipitation: A Case Study of the “7·20” Rainstorm in Henan Province, China

One of the important tasks of the Chinese geostationary and meteorological satellite Fengyun-2 (FY2) series is to provide quantitative precipitation estimates (QPE) with high spatiotemporal resolutions for East Asia. To analyze the monitoring capabilities of FY2-based QPEs in extreme rainfall events, this study comprehensively evaluated and compared the performances of FY-2G and FY-2H QPEs for the “7.20” rainstorm in Henan province, China from 17 July 2021 to 22 July 2021. Three continuous metrics and three categorical metrics were adopted to assess the accuracies of FY-2G and FY-2H QPEs, referenced by gauge observations from 116 meteorological stations. The results show that the FY-2G QPE has lower BIAS (−9.64% for FY-2G, −46.22% for FY-2H) and RMSE (5.83 mm/h for FY-2G, 8.4 mm/h for FY-2H) and higher CC (0.57 for FY-2G, 0.24 for FY-2H) than FY-2H QPE in this rainstorm event. Moreover, the FY-2G QPE is not only more consistent with the ground reference with respect to the rainfall amount, but also has higher detecting capability in the “7.20” rainstorm event when compared with the FY-2H QPE. The FY-2G QPE presented a higher capability to correctly capture the precipitation event for the “7.20” rainstorm because of higher POD (probability of detection) and CSI (critical success index) relative to FY-2H QPE, especially in complex topography. From the spatial distribution of precipitation amount, the FY-2G QPE captured the rainstorm center of extreme precipitation more accurately relative to the latest FY-2H product. On the other hand, the previous generation of FY-2G QPE was closer to the continuous rainfall process and precipitation duration with ground observations than the latest FY-2H QPE. Therefore, the precipitation retrieval algorithm of FY-2H QPE still had room to improve. It is necessary to introduce error correction algorithms, especially in complex topography for rainstorm events.

Study on the Vertical Structure and the Evolution of Precipitation Particle Spectrum Parameters of Stratocumulus Clouds over North China Based on Aircraft Observation

The understanding of the macro- and micro-structure, particle spectrum parameters, and their evolutions in different parts of stratocumulus clouds based on aircraft observation data, is important basic data for the development of cloud microphysical parameterization schemes and the quantitative retrieval of cloud-precipitation by radar and satellite detections. In this study, a total of ten vertical measurements during three aircraft observations were selected to analyze the vertical distribution of cloud microphysical properties in different parts of stratocumulus clouds in Hebei, North China. It was found that the downdraft in the cumulus cloud area was stronger than that in the stratiform cloud area, with the temperature at the same height higher than that in the stratiform cloud area, and the height of the 0 °C layers was correspondingly higher. In terms of particle spectrum parameters, the intercept and slope parameters of particle spectrum below melting levels in the cumulus part were higher than those in stratiform clouds area in the same weather process. In different vertical detection, it was found that the ice particles have begun to melt in the negative temperature layer near 0 °C level, and there might be sublimation, fragmentation, and aggregation in the melting process of ice phase particles. In addition, the melting process changed the spectral parameters greatly and also changed the correlation between the intercept and slope of the particle spectrum. The slope below the 0 °C level increased with the increase of intercept, which was greater than that above the 0 °C level. The relationship obtained between the intercept parameter of the particle’s spectrum and temperature, and the correlation between the maximum diameter and slope parameter of the particle spectrum, have certain reference significance for cloud physical parameterization and the quantitative retrieval of cloud precipitation by radar and satellite in North China and similar climate background areas.

A framework for merging precipitation retrievals and gauge-based observations based on a novel concept namely virtual gauges

The merging of multi-source precipitation retrievals (PRs) and gauge-based observations (GO) provides a new opportunity for precipitation field estimation. However, the uncertainties associated with PRs may become relatively more evident when the gauge network better captures the spatial distribution of the rainfall fields. To dynamically balance the utilisation of PRs, we propose a merging framework based on a novel concept; namely, the virtual gauge, where the grid cells utilising PRs are regarded as virtual gauges, and the framework is henceforth referred to as VG. The main steps were as follows: i) determine the locations of virtual gauges from multi-source PRs, ii) estimate rainfall at virtual gauges using a basic merging method, and iii) spatially interpolate using both actual and virtual gauges. Accordingly, the case study employed random forest and inverse distance weight as the basic merging and interpolation methods of VG. Evaluation using real world data over a region where nearly each 0.1° grid cell contains a ground gauge indicates that VG improves around 7–11% over its basic methods and improves around 32–240% over its inputting PRs. VG performed better and was more stable than the basic methods under various gauge densities, rainfall intensities, and rainfall distributions. The results showed that VG could supplement the spatial information of rainfall fields missed by the gauge network and reduce interference caused by the uncertainties of PRs. Overall, the framework integrates the advantages of existing merging and spatial interpolation methods by adjusting the number, locations, and rainfall estimations of virtual gauges.

Multi‐Task Learning for Simultaneous Retrievals of Passive Microwave Precipitation Estimates and Rain/No‐Rain Classification

AbstractSatellite‐based precipitation estimations provide frequent, large‐scale measurements. Deep learning has recently shown significant potential for improving estimation accuracy. Most studies have employed a two‐stage framework, which is a sequential architecture of a rain/no‐rain binary classification task followed by a rain rate regression task. This study proposes a novel precipitation retrieval framework in which these two tasks are simultaneously trained using multi‐task learning approach (MTL). Furthermore, a novel network architecture and loss function were designed to reap the benefits of MTL. The proposed two‐task model successfully achieved a better performance than the conventional single‐task model possibly due to efficient knowledge transfer between tasks. Furthermore, the product intercomparison showed that our product outperformed existing products in rain rate retrieval and also yielded better skills in the rain/no‐rain retrieval task.

Precipitation Estimation Based on Infrared Data with a Spherical Convolutional Neural Network

Abstract Precipitation is a vital process in the water cycle. Accurate estimation of the precipitation rate underpins the success of hydrological simulations, flood predictions, and water resource management. Satellite infrared (IR) data, with high temporal resolution and wide coverages, have been commonly used in precipitation inversion. However, existing IR-based precipitation retrieval algorithms suffer from various problems such as overestimation in dry regions, poor performance in extreme rainfall events, and reliance on an empirical cloud-top brightness–rain rate relationship. To resolve these problems, we construct a deep learning model using a spherical convolutional neural network to properly represent Earth’s spherical surface. With data input directly from IR bands 3, 4, and 6 of the operational Geostationary Operational Environmental Satellite (GOES), our new model of Precipitation Estimation based on IR data with Spherical Convolutional Neural Network (PEISCNN) was first trained and tested with a 3-month-long dataset, and then validated in a 2-yr period. Compared to the commonly used IR-based precipitation product PERSIANN-CCS (Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks–Cloud Classification System), PEISCNN showed significant improvement in the metrics of POD, CSI, RMSE, and CC, especially in the dry region and for extreme rainfall events. Decomposed with the four-component error decomposition (4CED) method, the overestimation of PEISCNN was averaged 47.66% lower than the CCS at the hourly scale. The PEISCNN model may provide a promising way to produce an improved IR-based precipitation product to benefit a wide range of hydrological applications. Significance Statement An IR-based precipitation algorithm is irreplaceable in satellite precipitation inversion, since an IR sensor can provide observations of high frequency, fine temporal resolution, and wide coverage. Considering the spherical nature of Earth’s surface which has been overlooked in previous IR-based precipitation retrieval algorithms, we proposed a new deep learning model PEISCNN, which can address the problems that exist in IR-based precipitation estimations such as overestimation in dry regions, deficiency in extreme rainfall events, and reliance on the empirical cloud-top brightness–rain rate relationship. PEISCNN provides a new insight to improve the accuracy of the satellite IR-based or multisensor-based precipitation estimation, and it has great potential to benefit a range of related hydrological research, applications in water resource management, and flood predictions.

Evaluation of SWER(Ze) Relationships by Precipitation Imaging Package (PIP) during ICE-POP 2018

Abstract Improving estimation of snow water equivalent rate (SWER) from radar reflectivity (Ze), known as a SWER(Ze) relationship, is a priority for NASA’s Global Precipitation Measurement (GPM) mission ground validation program as it is needed to comprehensively validate spaceborne precipitation retrievals. This study investigates the performance of eight operational and four research-based SWER(Ze) relationships utilizing Precipitation Imaging Probe (PIP) observations from the International Collaborative Experiment for Pyeongchang 2018 Olympic and Paralympic Winter Games (ICE-POP 2018) field campaign. During ICE-POP 2018, there were 10 snow events that are classified by synoptic conditions as either cold low or warm low, and a SWER(Ze) relationship is derived for each event. Additionally, a SWER(Ze) relationship is derived for each synoptic classification by merging all events within each class. Two new types of SWER(Ze) relationships are derived from PIP measurements of bulk density and habit classification. These two physically based SWER(Ze) relationships provided superior estimates of SWER when compared to the operational, event-specific, and synoptic SWER(Ze) relationships. For estimates of the event snow water equivalent total, the event-specific, synoptic, and best-performing operational SWER(Ze) relationships outperformed the physically based SWER(Ze) relationship, although the physically based relationships still performed well. This study recommends using the density or habit-based SWER(Ze) relationships for microphysical studies, whereas the other SWER(Ze) relationships are better suited toward hydrologic application.

How accurate are infrared-only and rain gauge-adjusted multi-satellite precipitation products in the southwest monsoon precipitation estimation across India?

A dense network of rain gauges and considerably large variability of the southwest monsoon precipitation across the country make India a suitable test-bed to evaluate any satellite-based precipitation product. In this paper, three real-time infrared-only precipitation products derived from the INSAT-3D satellite namely, INSAT Multispectral Rainfall (IMR), Corrected IMR (IMC) and Hydro-Estimator (HEM) and three rain gauge-adjusted Global Precipitation Measurement (GPM)-based multi-satellite precipitation products namely, Integrated Multi-satellitE Retrievals for GPM (IMERG), Global Satellite Mapping of Precipitation (GSMaP) and an Indian merged satellite-gauge product (INMSG) have been evaluated over India at a daily timescale for the southwest monsoon seasons of 2020 and 2021. An evaluation against rain gauge-based gridded reference dataset shows noticeable reduction of bias in IMC product over IMR, primarily over the orographic regions. However, INSAT-3D infrared-only precipitation retrieval algorithms have limitations in shallow and convective precipitation estimation. Among rain gauge-adjusted multi-satellite products, INMSG is shown to be the best product in the monsoon precipitation estimation over India due to use of rather larger number of rain gauges than IMERG and GSMaP products. All satellite-derived precipitation products, i.e. infrared-only and gauge-adjusted multi-satellite products underestimate heavy monsoon precipitation by 50-70%. The bias decomposition analysis indicates that a simple statistical bias correction would considerably improve the performance of the INSAT-3D precipitation products over the central India, but the same might not work over the west coast due to rather larger contributions of both positive and negative hit bias components. Although rain gauge-adjusted multi-satellite precipitation products show very small or negligible total biases in the monsoon precipitation estimation, positive and negative hit bias components are considerable over the west coast and central India. Furthermore, rain gauge-adjusted multi-satellite precipitation products underestimate very heavy to extremely heavy precipitation with larger magnitudes than the INSAT-3D derived precipitation products over the central India. Among the rain gauge-adjusted multi-satellite precipitation products, INMSG has smaller bias and error than IMERG and GSMaP products for very heavy to extremely heavy monsoon precipitation over the west coast and central India. Preliminary results of this study would be useful for end users in choosing a better precipitation product for real-time and research applications as well as for algorithm developers in further improving these products.

Atmospheric humidity and temperature sounding from the CubeSat TROPICS mission: Early performance evaluation with MiRS

Atmospheric humidity and temperature are the two most important geophysical variables for numerical weather forecasting and relevant Earth studies, contributing to the reduction of over 60% forecast errors. Conventional large satellite microwave radiometers provide the primary sounding data of moisture and temperature, with the V-band (near 50 GHz) and G-band (near 183 GHz) as the main spectra. In recent years, CubeSats are rapidly developing with compact sizes, novel designs, and low cost. The NASA TROPICS is an ongoing CubeSat constellation mission for studying tropical meteorology and storm systems, with the first Pathfinder satellite launched on 30 June 2021, which will provide a median revisit rate better than 60 min. TROPICS has innovative channels at F-band (near 118 GHz) and 205 GHz, providing a new perspective for atmospheric sounding. In this study, we have extended the NOAA MiRS system and applied it to TROPICS Pathfinder early-phase, provisionally calibrated data, focusing on moisture and temperature. We have examined the retrieval quality and compared with the ECMWF analysis and MiRS retrievals from NOAA-20 ATMS. An experiment of subsetting ATMS to match up with TROPICS channels is also conducted. Here we focus on atmospheric humidity and temperature retrievals, since they are of primary importance for microwave sounders and is tied to the core of the TROPICS mission along with precipitation retrieval. Compared to traditional channels, TROPICS F-band and 205 GHz provide new information content with distinct sensitivity to moisture and hydrometeors, showing for example over 20 K larger dynamic range of brightness temperature at 205 GHz compared to 190 GHz. The retrieved total precipitable water compared to ECMWF has a correlation coefficient of 0.955, 0.985, and 0.977 for TROPICS, ATMS, and ATMS subset, respectively. For water vapor profile, the standard deviation of retrieval to ECMWF is 0.93, 0.76, and 0.80 g/kg for the three experiments, respectively, and regarding temperature, it is 2.5, 1.5, and 1.6 K, respectively. The retrieval shows dependence on surface type and clouds, where land and cloudy conditions degrade retrieval compared to ocean and clear condition. The early performance is promising, and the successful MiRS extension paves the way to explore improved data when the calibration validation is complete and from the forthcoming constellation.

How Accurately Can Warm Rain Realistically Be Retrieved with Satellite Sensors? Part II: Horizontal and Vertical Heterogeneities

Abstract There are many sources of uncertainty in satellite precipitation retrievals of warm rain. In this paper, the second of a two-part study, we focus on uncertainties related to spatial heterogeneity and surface clutter. A cloud-resolving model simulation of warm, shallow clouds is used to simulate satellite observations from three theoretical satellite architectures—one similar to the Global Precipitation Measurement Core Observatory, one similar to CloudSat, and one similar to the planned Atmosphere Observing System (AOS). Rain rates are then retrieved using a common optimal estimation framework. For this case, retrieval biases due to nonuniform beamfilling are very large, with retrieved rain rates negatively (low) biased by as much as 40%–50% (depending on satellite architecture) at 5 km horizontal resolution. Surface clutter also acts to negatively bias retrieved rain rates. Combining all sources of uncertainty, the theoretical AOS satellite is found to outperform CloudSat in terms of retrieved surface rain rate, with a bias of −19% as compared with −28%, a reduced spread of retrieval errors, and an additional 17.5% of cases falling within desired uncertainty limits. The results speak to the need for additional high-resolution modeling simulations of warm rain so as to better characterize the uncertainties in satellite precipitation retrievals.

A Microphysics-Scheme-Consistent Snow Optical Parameterization for the Community Radiative Transfer Model

Abstract The satellite observational data assimilation community requires consistent hydrometer descriptions—including mass–size relation and particle size distribution—to be used in both the forecast model and observation operator. We develop a microphysics-scheme-consistent snow and graupel single-scattering property database to meet this requirement. In this database, snowflakes are modeled as a mixture of small column and large aggregated ice particles, the mixing ratios of which may be adjusted to satisfy a given mass–size relation. Snow single-scattering properties are computed for four different mass–size relations. Subsequently, the snow description in the Thompson microphysics scheme is used as an example to demonstrate how microphysics-scheme-consistent snow bulk optical properties are derived. The Thompson-scheme-consistent snow bulk optical properties are added to the Community Radiative Transfer Model (CRTM), version 2.4.0. With CloudSat Cloud Profiling Radar (CPR) snow and liquid precipitation retrievals as the inputs, CRTM simulations are performed over global oceans and compared with four collocated Global Precipitation Measurement (GPM) Microwave Imager (GMI) high-frequency channel observations. The CRTM simulated brightness temperatures show agreement with the GMI observed brightness temperatures in cases of light-to-moderate precipitation over extratropical and polar ice-free oceans, with root-mean-square errors of 4.3, 13.0, 1.8, and 3.3 K in the 166-GHz (vertical polarization), 166-GHz (horizontal polarization), 183 ± 3-GHz (vertical polarization), and 183 ± 7-GHz (vertical polarization) channels, respectively. The result demonstrates the potential of using the newly developed microphysics-scheme-consistent snow optical parameterization in data assimilation applications.

Multi‐Year Analysis of Rain‐Snow Levels at Marquette, Michigan

AbstractThis study uses observations from a ground‐based instrument suite to investigate the rain‐snow level (RSL) in stratiform rainfall from January 2014 to April 2020 in the Upper Great Lakes Region. The height above the surface where ice melts to rain, the rain‐snow level, influences microphysical assumptions in remote sensing precipitation retrievals and the ability of space‐based radar to discriminate surface precipitation phase because of ground clutter. The instrument suite is installed at the Marquette, MI (MQT) National Weather Service station adjacent to Lake Superior. Rain events and the RSLs are studied through a ground‐based vertically profiling radar (Micro Rain Radar), a custom NASA‐developed video disdrometer (Precipitation Imaging Package), and reanalysis products from ECMWF ERA5 and NASA MERRA‐2. Distinct macro and microphysical characteristics are observed in precipitation events with shallow RSLs (<1.8 km above ground level [AGL]) and intermediate RSLs (>1.8 km AGL). Intermediate RSLs correspond to rain events with relatively higher rain rates and a higher concentration of small drops in the drop size distributions (DSDs). Shallow RSL DSDs contain relatively higher concentrations of large drops with lower fall speeds suggesting that partially melted snowflakes may be reaching the surface. Reflectivity‐rain‐rate relationships are also impacted by microphysical differences associated with RSL regimes. Radar‐detected RSLs agree with reanalysis‐derived melt levels‐especially at wet‐bulb isotherms of 0.5°C and 1°C. Seasonal differences such as shallow RSLs in winter, fall, and spring have subsequent implications for satellite detectability.

A Cloud-Dependent 1DVAR Precipitation Retrieval Algorithm for FengYun-3D Microwave Soundings: A Case Study in Tropical Cyclone Mekkhala

A cloud-dependent 1-D variational (1DVAR) precipitation retrieval algorithm applied to FengYun-3D microwave soundings (CD1DVAR-FY3DMS) is developed in this study. Compared with the current 1DVAR precipitation retrieval framework, cloud scene identification (CSI) is first proposed to delineate the different weather conditions including clear sky, stratiform clouds, and convective clouds. Results of this study in a typical tropical cyclone event demonstrate that: 1) the precipitation retrievals considering CSI (correlation coefficient ~0.72 and root mean square error ~1.63 mm/h) outperform those without distinguishing cloud scenes (correlation coefficient ~0.66 and root mean square error ~1.82 mm/h); 2) identifying cloud scenes in a variational scheme could significantly improve the accuracy of retrieving heavy precipitation volumes, with the capturing abilities improved by ~100%; and 3) the FengYun-3 constellation has great potential to complement global microwave retrievals. In addition, these findings could provide valuable references and pathfinders for further improving the retrieval accuracy of microwave-based precipitation estimates, especially for strong convective zones.

Deep Learning for Precipitation Retrievals Using ABI and GLM Measurements on the GOES-R Series

Deep Learning for Precipitation Retrievals Using ABI and GLM Measurements on the GOES-R Series

Development of precipitation retrieval based on multimode radar remote sensing

Development of precipitation retrieval based on multimode radar remote sensing

PrecipGradeNet: A New Paradigm and Model for Precipitation Retrieval with Grading of Precipitation Intensity

Near-real-time precipitation retrieval plays an important role in the study of the evolutionary process of precipitation and the prevention of disasters caused by heavy precipitation. Compared with ground-based precipitation observations, the infrared precipitation estimations from geostationary satellites have great advantages in terms of geographical coverage and temporal resolution. However, precipitation retrieved from multispectral infrared data still faces challenges in terms of accuracy, especially in extreme cases. In this paper, we propose a new paradigm for satellite multispectral infrared data retrieval of precipitation and construct a new model called PrecipGradeNet. This model uses FY-4A L1 FDI data as the input, IMERG precipitation data as the training target, and improves the precipitation retrieval accuracy by grading the precipitation intensity through Res-UNet, a semantic segmentation network. To evaluate the precipitation retrieval of the model, we compare the retrieval results with the FY-4A L2 QPE operational product to the IMERG precipitation. IMERG is considered as the ground truth. We evaluate the precipitation retrieval from the precipitation fall area identification, the precipitation intensity interval discrimination, and the precipitation quantification. Experimental results show that PrecipGradeNet has better overall performance compared with the FY-4A QPE product in precipitation fall area identification with POD increased by 48% and CSI and HSS improved by 21% and 14%. PrecipGradeNet also has better performance in light precipitation with POD increased by 114% and CSI and HSS improved by 64% and 52%, and better overall precipitation quantification, with RMSE and CC improved by 16% and 15%. In addition, PrecipGradeNet avoids the overall bias in the low and extreme high precipitation cases. Therefore, the new paradigm proposed in this paper has the potential to improve the retrieval accuracy of satellite precipitation estimation products. This study suggests that the application of semantic segmentation methods may provide a new path to correct the intensity bias of the satellite-based precipitation products.

PRSOT: Precipitation Retrieval from Satellite Observations Based on Transformer

Precipitation with high spatial and temporal resolution can improve the defense capability of meteorological disasters and provide indispensable instruction and early warning for social public services, such as agriculture, forestry, and transportation. Therefore, a deep learning-based algorithm entitled precipitation retrieval from satellite observations based on Transformer (PRSOT) is proposed to fill the observation gap of ground rain gauges and weather radars in deserts, oceans, and other regions. In this algorithm, the multispectral infrared brightness temperatures from Himawari-8, the new-generation geostationary satellite, have been used as predictor variables and the Global Precipitation Measurement (GPM) precipitation product has been employed to train the retrieval model. We utilized two data normalization schemes, area-based and pixel-based normalization, and conducted comparative experiments. Comparing the estimated results with the GPM product on the test set, PRSOT_Pixel_based model achieved a Probability Of Detection (POD) of 0.74, a False Alarm Ratio (FAR) of 0.44 and a Critical Success Index (CSI) of 0.47 for two-class metrics, and an Accuracy (ACC) of 0.75 for multi-class metrics. Pixel-based normalization is more suitable for meteorological data, highlighting the precipitation characteristics and obtaining better comprehensive retrieval performance in visualization and evaluation metrics. In conclusion, the proposed PRSOT model has made a remarkable and essential contribution to precipitation retrieval and outperforms the benchmark machine learning model Random Forests.

An improved near-real-time precipitation retrieval for Brazil

Abstract. Observations from geostationary satellites can provide spatially continuous coverage at continental scales with high spatial and temporal resolution. Because of this, they are commonly used to complement ground-based precipitation measurements, whose coverage is often more limited. We present Hydronn, a neural-network-based, near-real-time precipitation retrieval for Brazil based on visible and infrared (Vis–IR) observations from the Advanced Baseline Imager (ABI) on the Geostationary Operational Environmental Satellite 16 (GOES-16). The retrieval, which employs a convolutional neural network to perform Bayesian precipitation retrievals, was developed with the aims of (1) leveraging the full potential of latest-generation geostationary observations and (2) providing probabilistic precipitation estimates with well-calibrated uncertainties. The retrieval is trained using more than 3 years of collocations with combined radar and radiometer retrievals from the Global Precipitation Measurement (GPM) core observatory over South America. The accuracy of instantaneous precipitation estimates is assessed using a separate year of GPM combined retrievals and compared to retrievals from passive microwave (PMW) sensors and HYDRO, the Vis–IR retrieval that is currently in operational use at the Brazilian Institute for Space Research. Using all available channels of the ABI, Hydronn achieves accuracy close to that of state-of-the-art PMW precipitation retrievals in both precipitation estimation and detection despite the lower information content of the Vis–IR observations. Hourly, daily, and monthly precipitation accumulations are evaluated against gauge measurements for June and December 2020 and compared to HYDRO, the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN) Cloud Classification System (CCS), and the Integrated Multi-satellitE Retrievals for GPM (IMERG). Compared to HYDRO, Hydronn reduces the mean absolute error for hourly accumulations by 21 % (22 %) compared to HYDRO by 44 % (41 %) for the mean squared error (MSE) and increases the correlation by 138 % (312 %) for June (December) 2020. Compared to IMERG, the improvements correspond to 16 % (14 %), 12 % (12 %), and 20 % (56 %), respectively. Furthermore, we show that the probabilistic retrieval is well calibrated against gauge measurements when differences in the distributions of the training data and the gauge measurements are accounted for. Hydronn has the potential to significantly improve near-real-time precipitation retrievals over Brazil. Furthermore, our results show that precipitation retrievals based on convolutional neural networks (CNNs) that leverage the full range of available observations from latest-generation geostationary satellites can provide instantaneous precipitation estimates with accuracy close to that of state-of-the-art PMW retrievals. The high temporal resolution of the geostationary observation allows Hydronn to provide more accurate precipitation accumulations than any of the tested conventional precipitation retrievals. Hydronn thus clearly shows the potential of deep-learning-based precipitation retrievals to improve precipitation estimates from currently available satellite imagery.