Browse
Abstract
Few statistical downscaling applications have provided gridded products that can provide downscaled values for a no-gauge area as is done by dynamical downscaling. In this study, a gridded statistical downscaling scheme is presented to downscale summer precipitation to a dense grid that covers North China. The main innovation of this scheme is interpolating the parameters of single-station models to this dense grid and assigning optimal predictor values according to an interpolated predictand–predictor distance function. This method can produce spatial dependence (spatial autocorrelation) and transmit the spatial heterogeneity of predictor values from the large-scale predictors to the downscaled outputs. Such gridded output at no-gauge stations shows performances comparable to that at the gauged stations. The area mean precipitation of the downscaled results is comparable to other products. The main value of the downscaling scheme is that it can obtain reasonable outputs for no-gauge stations.
Abstract
Few statistical downscaling applications have provided gridded products that can provide downscaled values for a no-gauge area as is done by dynamical downscaling. In this study, a gridded statistical downscaling scheme is presented to downscale summer precipitation to a dense grid that covers North China. The main innovation of this scheme is interpolating the parameters of single-station models to this dense grid and assigning optimal predictor values according to an interpolated predictand–predictor distance function. This method can produce spatial dependence (spatial autocorrelation) and transmit the spatial heterogeneity of predictor values from the large-scale predictors to the downscaled outputs. Such gridded output at no-gauge stations shows performances comparable to that at the gauged stations. The area mean precipitation of the downscaled results is comparable to other products. The main value of the downscaling scheme is that it can obtain reasonable outputs for no-gauge stations.
Abstract
Large hail is a primary contributor to damages and loss around the world, in both agriculture and infrastructure. The sensitivity of passive microwave radiometer measurements to scattering by hail led to the development of proxies for severe hail, most of which use brightness temperature thresholds from 37-GHz and higher-frequency microwave channels on board weather satellites in low-Earth orbit. Using 16+ years of data from the Tropical Rainfall Measuring Mission (TRMM; 36°S–36°N), we pair TRMM brightness temperature–derived precipitation features with surface hail reports in the United States to train a hail retrieval on passive microwave data from the 10-, 19-, 37-, and 85-GHz channels based on probability curves fit to the microwave data. We then apply this hail retrieval to features in the Global Precipitation Measurement (GPM) domain (from 69°S to 69°N) to develop a nearly global passive microwave–based climatology of hail. The extended domain of the GPM satellite into higher latitudes requires filtering out features that we believe are over icy and snowy surface regimes. We also normalize brightness temperature depression by tropopause height in an effort to account for differences in storm depth between the tropics and higher latitudes. Our results show the highest hail frequencies in the region of northern Argentina through Paraguay, Uruguay, and southern Brazil; the central United States; and a swath of Africa just south of the Sahel. Smaller hot spots include Pakistan, eastern India, and Bangladesh. A notable difference between these results and many prior satellite-based studies is that central Africa, while still active in our climatology, does not rival the aforementioned regions in retrieved hailstorm frequency.
Abstract
Large hail is a primary contributor to damages and loss around the world, in both agriculture and infrastructure. The sensitivity of passive microwave radiometer measurements to scattering by hail led to the development of proxies for severe hail, most of which use brightness temperature thresholds from 37-GHz and higher-frequency microwave channels on board weather satellites in low-Earth orbit. Using 16+ years of data from the Tropical Rainfall Measuring Mission (TRMM; 36°S–36°N), we pair TRMM brightness temperature–derived precipitation features with surface hail reports in the United States to train a hail retrieval on passive microwave data from the 10-, 19-, 37-, and 85-GHz channels based on probability curves fit to the microwave data. We then apply this hail retrieval to features in the Global Precipitation Measurement (GPM) domain (from 69°S to 69°N) to develop a nearly global passive microwave–based climatology of hail. The extended domain of the GPM satellite into higher latitudes requires filtering out features that we believe are over icy and snowy surface regimes. We also normalize brightness temperature depression by tropopause height in an effort to account for differences in storm depth between the tropics and higher latitudes. Our results show the highest hail frequencies in the region of northern Argentina through Paraguay, Uruguay, and southern Brazil; the central United States; and a swath of Africa just south of the Sahel. Smaller hot spots include Pakistan, eastern India, and Bangladesh. A notable difference between these results and many prior satellite-based studies is that central Africa, while still active in our climatology, does not rival the aforementioned regions in retrieved hailstorm frequency.
Abstract
Using three years of observations from the Dual-Frequency Precipitation Radar (DPR) aboard the Global Precipitation Measurement (GPM) Core Observatory, properties of the cores of deep convection are examined. First, deep convective systems are selected, defined as GPM precipitation features with maximum 20-dBZ echo-top heights above 10 km. The cores of deep convection are described by the profiles of Ku- and Ka-band radar reflectivity at the location of the highest echo top in each deep convective system. Then the dual-frequency ratio (DFR) profile is derived by subtracting Ka-band from Ku-band radar reflectivity. It is found that values of DFR are larger over land than over ocean in general near the top of the convection, which is consistent with larger ice particles in stronger updrafts in continental convection. The magnitude of DFR at 12 km is positively correlated with the convection intensity indicated by 20- and 30-dBZ echo tops. The microphysical properties including volume-weighted mean diameter, ice water content, and total ice particle number concentration are derived using a simple lookup table approach. Under the same particle size distribution assumption, the cores of deep convection over land have larger ice particle size, higher ice water content, and lower particle concentration than those over ocean at levels above 10 km, but with some distinct regional variations.
Abstract
Using three years of observations from the Dual-Frequency Precipitation Radar (DPR) aboard the Global Precipitation Measurement (GPM) Core Observatory, properties of the cores of deep convection are examined. First, deep convective systems are selected, defined as GPM precipitation features with maximum 20-dBZ echo-top heights above 10 km. The cores of deep convection are described by the profiles of Ku- and Ka-band radar reflectivity at the location of the highest echo top in each deep convective system. Then the dual-frequency ratio (DFR) profile is derived by subtracting Ka-band from Ku-band radar reflectivity. It is found that values of DFR are larger over land than over ocean in general near the top of the convection, which is consistent with larger ice particles in stronger updrafts in continental convection. The magnitude of DFR at 12 km is positively correlated with the convection intensity indicated by 20- and 30-dBZ echo tops. The microphysical properties including volume-weighted mean diameter, ice water content, and total ice particle number concentration are derived using a simple lookup table approach. Under the same particle size distribution assumption, the cores of deep convection over land have larger ice particle size, higher ice water content, and lower particle concentration than those over ocean at levels above 10 km, but with some distinct regional variations.
Abstract
Global Precipitation Measurement (GPM) Microwave Imager (GMI) brightness temperatures (BTs) were simulated over a case of severe convection in Texas using ground-based S-band radar and the Atmospheric Radiative Transfer Simulator. The median particle diameter D o of a normalized gamma distribution was varied for different hydrometeor types under the constraint of fixed radar reflectivity to better understand how simulated GMI BTs respond to changing particle size distribution parameters. In addition, simulations were conducted to assess how low BTs may be expected to reach from realistic (although extreme) particle sizes or concentrations. Results indicate that increasing D o for cloud ice, graupel, and/or hail leads to warmer BTs (i.e., weaker scattering signature) at various frequencies. Channels at 166.0 and 183.31 ± 7 GHz are most sensitive to changing D o of cloud ice, channels at ≥89.0 GHz are most sensitive to changing D o of graupel, and at 18.7 and 36.5 GHz they show the greatest sensitivity to hail D o . Simulations contrasting BTs above high concentrations of small (0.5-cm diameter) and low concentrations of large (20-cm diameter) hailstones distributed evenly across a satellite pixel showed much greater scattering using the higher concentration of smaller hailstones with BTs as low as ~110, ~33, ~22, ~46, ~100, and ~106 K at 10.65, 18.7, 36.5, 89.0, 166.0, and 183.31 ± 7 GHz, respectively. These results suggest that number concentration is more important for scattering than particle size given a constant S-band radar reflectivity.
Abstract
Global Precipitation Measurement (GPM) Microwave Imager (GMI) brightness temperatures (BTs) were simulated over a case of severe convection in Texas using ground-based S-band radar and the Atmospheric Radiative Transfer Simulator. The median particle diameter D o of a normalized gamma distribution was varied for different hydrometeor types under the constraint of fixed radar reflectivity to better understand how simulated GMI BTs respond to changing particle size distribution parameters. In addition, simulations were conducted to assess how low BTs may be expected to reach from realistic (although extreme) particle sizes or concentrations. Results indicate that increasing D o for cloud ice, graupel, and/or hail leads to warmer BTs (i.e., weaker scattering signature) at various frequencies. Channels at 166.0 and 183.31 ± 7 GHz are most sensitive to changing D o of cloud ice, channels at ≥89.0 GHz are most sensitive to changing D o of graupel, and at 18.7 and 36.5 GHz they show the greatest sensitivity to hail D o . Simulations contrasting BTs above high concentrations of small (0.5-cm diameter) and low concentrations of large (20-cm diameter) hailstones distributed evenly across a satellite pixel showed much greater scattering using the higher concentration of smaller hailstones with BTs as low as ~110, ~33, ~22, ~46, ~100, and ~106 K at 10.65, 18.7, 36.5, 89.0, 166.0, and 183.31 ± 7 GHz, respectively. These results suggest that number concentration is more important for scattering than particle size given a constant S-band radar reflectivity.
Abstract
Satellite retrieval algorithms and model microphysical parameterizations require guidance from observations to improve the representation of ice-phase microphysical quantities and processes. Here, a parameterization for ice-phase particle size distributions (PSDs) is developed using in situ measurements of cloud microphysical properties collected during the Global Precipitation Measurement (GPM) Cold-Season Precipitation Experiment (GCPEx). This parameterization takes advantage of the relation between the gamma-shape parameter μ and the mass-weighted mean diameter D m of the ice-phase PSD sampled during GCPEx. The retrieval of effective reflectivity Z e and ice water content (IWC) from the reconstructed PSD using the μ–D m relationship was tested with independent measurements of Z e and IWC and overall leads to a mean error of 8% in both variables. This represents an improvement when compared with errors using the Field et al. parameterization of 10% in IWC and 37% in Z e . Current radar precipitation retrieval algorithms from GPM assume that the PSD follows a gamma distribution with μ = 3. This assumption leads to a mean overestimation of 5% in the retrieved Z e , whereas applying the μ–D m relationship found here reduces this bias to an overestimation of less than 1%. Proper selection of the a and b coefficients in the mass–dimension relationship is also of crucial importance for retrievals. An inappropriate selection of a and b, even from values observed in previous studies in similar environments and cloud types, can lead to more than 100% bias in IWC and Z e for the ice-phase particles analyzed here.
Abstract
Satellite retrieval algorithms and model microphysical parameterizations require guidance from observations to improve the representation of ice-phase microphysical quantities and processes. Here, a parameterization for ice-phase particle size distributions (PSDs) is developed using in situ measurements of cloud microphysical properties collected during the Global Precipitation Measurement (GPM) Cold-Season Precipitation Experiment (GCPEx). This parameterization takes advantage of the relation between the gamma-shape parameter μ and the mass-weighted mean diameter D m of the ice-phase PSD sampled during GCPEx. The retrieval of effective reflectivity Z e and ice water content (IWC) from the reconstructed PSD using the μ–D m relationship was tested with independent measurements of Z e and IWC and overall leads to a mean error of 8% in both variables. This represents an improvement when compared with errors using the Field et al. parameterization of 10% in IWC and 37% in Z e . Current radar precipitation retrieval algorithms from GPM assume that the PSD follows a gamma distribution with μ = 3. This assumption leads to a mean overestimation of 5% in the retrieved Z e , whereas applying the μ–D m relationship found here reduces this bias to an overestimation of less than 1%. Proper selection of the a and b coefficients in the mass–dimension relationship is also of crucial importance for retrievals. An inappropriate selection of a and b, even from values observed in previous studies in similar environments and cloud types, can lead to more than 100% bias in IWC and Z e for the ice-phase particles analyzed here.
Abstract
Retrievals of falling snow from space-based observations represent key inputs for understanding and linking Earth’s atmospheric, hydrological, and energy cycles. This work quantifies and investigates causes of differences among the first stable falling snow retrieval products from the Global Precipitation Measurement (GPM) Core Observatory satellite and CloudSat’s Cloud Profiling Radar (CPR) falling snow product. An important part of this analysis details the challenges associated with comparing the various GPM and CloudSat snow estimates arising from different snow–rain classification methods, orbits, resolutions, sampling, instrument specifications, and algorithm assumptions. After equalizing snow–rain classification methodologies and limiting latitudinal extent, CPR observes nearly 10 (3) times the occurrence (accumulation) of falling snow as GPM’s Dual-Frequency Precipitation Radar (DPR). The occurrence disparity is substantially reduced if CloudSat pixels are averaged to simulate DPR radar pixels and CPR observations are truncated below the 8-dBZ reflectivity threshold. However, even though the truncated CPR- and DPR-based data have similar falling snow occurrences, average snowfall rate from the truncated CPR record remains significantly higher (43%) than the DPR, indicating that retrieval assumptions (microphysics and snow scattering properties) are quite different. Diagnostic reflectivity (Z)–snow rate (S) relationships were therefore developed at Ku and W band using the same snow scattering properties and particle size distributions in a final effort to minimize algorithm differences. CPR–DPR snowfall amount differences were reduced to ~16% after adopting this diagnostic Z–S approach.
Abstract
Retrievals of falling snow from space-based observations represent key inputs for understanding and linking Earth’s atmospheric, hydrological, and energy cycles. This work quantifies and investigates causes of differences among the first stable falling snow retrieval products from the Global Precipitation Measurement (GPM) Core Observatory satellite and CloudSat’s Cloud Profiling Radar (CPR) falling snow product. An important part of this analysis details the challenges associated with comparing the various GPM and CloudSat snow estimates arising from different snow–rain classification methods, orbits, resolutions, sampling, instrument specifications, and algorithm assumptions. After equalizing snow–rain classification methodologies and limiting latitudinal extent, CPR observes nearly 10 (3) times the occurrence (accumulation) of falling snow as GPM’s Dual-Frequency Precipitation Radar (DPR). The occurrence disparity is substantially reduced if CloudSat pixels are averaged to simulate DPR radar pixels and CPR observations are truncated below the 8-dBZ reflectivity threshold. However, even though the truncated CPR- and DPR-based data have similar falling snow occurrences, average snowfall rate from the truncated CPR record remains significantly higher (43%) than the DPR, indicating that retrieval assumptions (microphysics and snow scattering properties) are quite different. Diagnostic reflectivity (Z)–snow rate (S) relationships were therefore developed at Ku and W band using the same snow scattering properties and particle size distributions in a final effort to minimize algorithm differences. CPR–DPR snowfall amount differences were reduced to ~16% after adopting this diagnostic Z–S approach.
Abstract
The Goddard convective–stratiform heating (CSH) algorithm has been used to retrieve latent heating (LH) associated with clouds and cloud systems in support of the Tropical Rainfall Measuring Mission and Global Precipitation Measurement (GPM) mission. The CSH algorithm requires the use of a cloud-resolving model to simulate LH profiles to build lookup tables (LUTs). However, the current LUTs in the CSH algorithm are not suitable for retrieving LH profiles at high latitudes or winter conditions that are needed for GPM. The NASA Unified-Weather Research and Forecasting (NU-WRF) Model is used to simulate three eastern continental U.S. (CONUS) synoptic winter and three western coastal/offshore events. The relationship between LH structures (or profiles) and other precipitation properties (radar reflectivity, freezing-level height, echo-top height, maximum dBZ height, vertical dBZ gradient, and surface precipitation rate) is examined, and a new classification system is adopted with varying ranges for each of these precipitation properties to create LUTs representing high latitude/winter conditions. The performance of the new LUTs is examined using a self-consistency check for one CONUS and one West Coast offshore event by comparing LH profiles retrieved from the LUTs using model-simulated precipitation properties with those originally simulated by the model. The results of the self-consistency check validate the new classification and LUTs. The new LUTs provide the foundation for high-latitude retrievals that can then be merged with those from the tropical CSH algorithm to retrieve LH profiles over the entire GPM domain using precipitation properties retrieved from the GPM combined algorithm.
Abstract
The Goddard convective–stratiform heating (CSH) algorithm has been used to retrieve latent heating (LH) associated with clouds and cloud systems in support of the Tropical Rainfall Measuring Mission and Global Precipitation Measurement (GPM) mission. The CSH algorithm requires the use of a cloud-resolving model to simulate LH profiles to build lookup tables (LUTs). However, the current LUTs in the CSH algorithm are not suitable for retrieving LH profiles at high latitudes or winter conditions that are needed for GPM. The NASA Unified-Weather Research and Forecasting (NU-WRF) Model is used to simulate three eastern continental U.S. (CONUS) synoptic winter and three western coastal/offshore events. The relationship between LH structures (or profiles) and other precipitation properties (radar reflectivity, freezing-level height, echo-top height, maximum dBZ height, vertical dBZ gradient, and surface precipitation rate) is examined, and a new classification system is adopted with varying ranges for each of these precipitation properties to create LUTs representing high latitude/winter conditions. The performance of the new LUTs is examined using a self-consistency check for one CONUS and one West Coast offshore event by comparing LH profiles retrieved from the LUTs using model-simulated precipitation properties with those originally simulated by the model. The results of the self-consistency check validate the new classification and LUTs. The new LUTs provide the foundation for high-latitude retrievals that can then be merged with those from the tropical CSH algorithm to retrieve LH profiles over the entire GPM domain using precipitation properties retrieved from the GPM combined algorithm.
Physical Evaluation of GPM DPR Single- and Dual-Wavelength AlgorithmsAbstract
A physical evaluation of the rain profiling retrieval algorithms for the Dual-Frequency Precipitation Radar (DPR) on board the Global Precipitation Measurement (GPM) Core Observatory satellite is carried out by applying them to the hydrometeor profiles generated from measured raindrop size distributions (DSD). The DSD-simulated radar profiles are used as input to the algorithms, and their estimates of hydrometeors’ parameters are compared with the same quantities derived directly from the DSD data (or truth). The retrieval accuracy is assessed by the degree to which the estimates agree with the truth. To check the validity and robustness of the retrievals, the profiles are constructed for cases ranging from fully correlated (or uniform) to totally uncorrelated DSDs along the columns. Investigation into the sensitivity of the retrieval results to the model assumptions is made to characterize retrieval uncertainties and identify error sources. Comparisons between the single- and dual-wavelength algorithm performance are carried out with either a single- or dual-wavelength constraint of the path integral or differential path integral attenuation. The results suggest that the DPR dual-wavelength algorithm generally provides accurate range-profiled estimates of rainfall rate and mass-weighted diameter with the dual-wavelength estimates superior in accuracy to those from the single-wavelength retrievals.
Abstract
A physical evaluation of the rain profiling retrieval algorithms for the Dual-Frequency Precipitation Radar (DPR) on board the Global Precipitation Measurement (GPM) Core Observatory satellite is carried out by applying them to the hydrometeor profiles generated from measured raindrop size distributions (DSD). The DSD-simulated radar profiles are used as input to the algorithms, and their estimates of hydrometeors’ parameters are compared with the same quantities derived directly from the DSD data (or truth). The retrieval accuracy is assessed by the degree to which the estimates agree with the truth. To check the validity and robustness of the retrievals, the profiles are constructed for cases ranging from fully correlated (or uniform) to totally uncorrelated DSDs along the columns. Investigation into the sensitivity of the retrieval results to the model assumptions is made to characterize retrieval uncertainties and identify error sources. Comparisons between the single- and dual-wavelength algorithm performance are carried out with either a single- or dual-wavelength constraint of the path integral or differential path integral attenuation. The results suggest that the DPR dual-wavelength algorithm generally provides accurate range-profiled estimates of rainfall rate and mass-weighted diameter with the dual-wavelength estimates superior in accuracy to those from the single-wavelength retrievals.
Ground Validation of Surface Snowfall Algorithm in GPM Dual-Frequency Precipitation RadarAbstract
Extensive evaluations have been performed on the dual-frequency classification module in the Global Precipitation Mission (GPM) Dual-Frequency Precipitation Radar (DPR) level-2 algorithm. Both rain type classification and melting-layer detection continue to show promising results in the validations. Surface snowfall identification is a feature newly added in the classification module to the recently released version to provide a surface snowfall flag for each qualified vertical profile. This algorithm is developed upon vertical features of Ku- and Ka-band reflectivity and dual-frequency ratio from DPR. In this paper, we validate this surface snowfall identification algorithm with ground radars including NEXRAD, NASA Polarimetric Radar (NPOL), and CSU–CHILL radar during concurrent precipitation events and GPM validation campaign Olympic Mountain Experiment (OLYMPEX). Other ground truth such as Precipitation Imaging Package (PIP) and ground report is also included in the validation. Based on 16 validation cases in the years 2014–18, the average match ratio between surface snowfall flag from space radar and ground radar is around 87.8%. Promising agreements are achieved with different validation sources. Algorithm limitation and potential improvement are discussed.
Abstract
Extensive evaluations have been performed on the dual-frequency classification module in the Global Precipitation Mission (GPM) Dual-Frequency Precipitation Radar (DPR) level-2 algorithm. Both rain type classification and melting-layer detection continue to show promising results in the validations. Surface snowfall identification is a feature newly added in the classification module to the recently released version to provide a surface snowfall flag for each qualified vertical profile. This algorithm is developed upon vertical features of Ku- and Ka-band reflectivity and dual-frequency ratio from DPR. In this paper, we validate this surface snowfall identification algorithm with ground radars including NEXRAD, NASA Polarimetric Radar (NPOL), and CSU–CHILL radar during concurrent precipitation events and GPM validation campaign Olympic Mountain Experiment (OLYMPEX). Other ground truth such as Precipitation Imaging Package (PIP) and ground report is also included in the validation. Based on 16 validation cases in the years 2014–18, the average match ratio between surface snowfall flag from space radar and ground radar is around 87.8%. Promising agreements are achieved with different validation sources. Algorithm limitation and potential improvement are discussed.
Abstract
The study evaluated a numerical weather model (WRF)-based satellite precipitation adjustment technique with 81 heavy precipitation events that occurred in three tropical mountainous regions (Colombia, Peru, and Taiwan). The technique was applied on two widely used near-real-time global satellite precipitation products—the National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Center morphing technique (CMORPH) and the Global Satellite Mapping of Precipitation project (GSMaP)—for each precipitation event. The WRF-adjusted satellite products along with the near-real-time and gauge-adjusted satellite products as well as the WRF simulation were evaluated by independent gauge networks at daily scale and event total scale. Results show that the near-real-time precipitation products exhibited severe underestimation relative to the gauge observations over the three tropical mountainous regions. The underestimation tended to be larger for higher rainfall accumulations. The WRF-based satellite adjustment provided considerable improvements to the near-real-time CMORPH and GSMaP products. Moreover, error metrics show that WRF-adjusted satellite products outperformed the gauge-adjusted counterparts for most of the events. The effectiveness of WRF-based satellite adjustment varied with events of different physical processes. Thus, the technique applied on satellite precipitation estimates of these events may exhibit inconsistencies in the bias correction.
Abstract
The study evaluated a numerical weather model (WRF)-based satellite precipitation adjustment technique with 81 heavy precipitation events that occurred in three tropical mountainous regions (Colombia, Peru, and Taiwan). The technique was applied on two widely used near-real-time global satellite precipitation products—the National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Center morphing technique (CMORPH) and the Global Satellite Mapping of Precipitation project (GSMaP)—for each precipitation event. The WRF-adjusted satellite products along with the near-real-time and gauge-adjusted satellite products as well as the WRF simulation were evaluated by independent gauge networks at daily scale and event total scale. Results show that the near-real-time precipitation products exhibited severe underestimation relative to the gauge observations over the three tropical mountainous regions. The underestimation tended to be larger for higher rainfall accumulations. The WRF-based satellite adjustment provided considerable improvements to the near-real-time CMORPH and GSMaP products. Moreover, error metrics show that WRF-adjusted satellite products outperformed the gauge-adjusted counterparts for most of the events. The effectiveness of WRF-based satellite adjustment varied with events of different physical processes. Thus, the technique applied on satellite precipitation estimates of these events may exhibit inconsistencies in the bias correction.
