Browse
Enhancing PMW Satellite Precipitation Estimation: Detecting Convective ClassAbstract
A decades-long effort in observing precipitation from space has led to continuous improvements of satellite-derived passive microwave (PMW) large-scale precipitation products. However, due to a limited ability to relate observed radiometric signatures to precipitation type (convective and stratiform) and associated precipitation rate variability, PMW retrievals are prone to large systematic errors at instantaneous scales. The present study explores the use of deep learning approach in extracting the information content from PMW observation vectors to help identify precipitation types. A deep learning neural network model (DNN) is developed to retrieve the convective type in precipitating systems from PMW observations. A 12-month period of Global Precipitation Measurement mission Microwave Imager (GMI) observations is used as a dataset for model development and verification. The proposed DNN model is shown to accurately predict precipitation types for 85% of total precipitation volume. The model reduces precipitation rate bias associated with convective and stratiform precipitation in the GPM operational algorithm by a factor of 2 while preserving the correlation with reference precipitation rates, and is insensitive to surface type variability. Based on comparisons against currently used convective schemes, it is concluded that the neural network approach has the potential to address regime-specific PMW satellite precipitation biases affecting GPM operations.
Abstract
A decades-long effort in observing precipitation from space has led to continuous improvements of satellite-derived passive microwave (PMW) large-scale precipitation products. However, due to a limited ability to relate observed radiometric signatures to precipitation type (convective and stratiform) and associated precipitation rate variability, PMW retrievals are prone to large systematic errors at instantaneous scales. The present study explores the use of deep learning approach in extracting the information content from PMW observation vectors to help identify precipitation types. A deep learning neural network model (DNN) is developed to retrieve the convective type in precipitating systems from PMW observations. A 12-month period of Global Precipitation Measurement mission Microwave Imager (GMI) observations is used as a dataset for model development and verification. The proposed DNN model is shown to accurately predict precipitation types for 85% of total precipitation volume. The model reduces precipitation rate bias associated with convective and stratiform precipitation in the GPM operational algorithm by a factor of 2 while preserving the correlation with reference precipitation rates, and is insensitive to surface type variability. Based on comparisons against currently used convective schemes, it is concluded that the neural network approach has the potential to address regime-specific PMW satellite precipitation biases affecting GPM operations.
Abstract
The Goddard convective–stratiform heating (CSH) algorithm, used to estimate cloud heating in support of the Tropical Rainfall Measuring Mission (TRMM), is upgraded in support of the Global Precipitation Measurement (GPM) mission. The algorithm’s lookup tables (LUTs) are revised using new and additional cloud-resolving model (CRM) simulations from the Goddard Cumulus Ensemble (GCE) model, producing smoother heating patterns that span a wider range of intensities because of the increased sampling and finer GPM product grid. Low-level stratiform cooling rates are reduced in the land LUTs for a given rain intensity because of the rain evaporation correction in the new four-class ice (4ICE) scheme. Additional criteria, namely, echo-top heights and low-level reflectivity gradients, are tested for the selection of heating profiles. Those resulting LUTs show greater and more precise variation in their depth of heating as well as a tendency for stronger cooling and heating rates when low-level dBZ values decrease toward the surface. Comparisons versus TRMM for a 3-month period show much more low-level heating in the GPM retrievals because of increased detection of shallow convection, while upper-level heating patterns remain similar. The use of echo tops and low-level reflectivity gradients greatly reduces midlevel heating from ~2 to 5 km in the mean GPM heating profile, resulting in a more top-heavy profile like TRMM versus a more bottom-heavy profile with much more midlevel heating. Integrated latent heating rates are much better balanced versus surface rainfall for the GPM retrievals using the additional selection criteria with an overall bias of +4.3%.
Abstract
The Goddard convective–stratiform heating (CSH) algorithm, used to estimate cloud heating in support of the Tropical Rainfall Measuring Mission (TRMM), is upgraded in support of the Global Precipitation Measurement (GPM) mission. The algorithm’s lookup tables (LUTs) are revised using new and additional cloud-resolving model (CRM) simulations from the Goddard Cumulus Ensemble (GCE) model, producing smoother heating patterns that span a wider range of intensities because of the increased sampling and finer GPM product grid. Low-level stratiform cooling rates are reduced in the land LUTs for a given rain intensity because of the rain evaporation correction in the new four-class ice (4ICE) scheme. Additional criteria, namely, echo-top heights and low-level reflectivity gradients, are tested for the selection of heating profiles. Those resulting LUTs show greater and more precise variation in their depth of heating as well as a tendency for stronger cooling and heating rates when low-level dBZ values decrease toward the surface. Comparisons versus TRMM for a 3-month period show much more low-level heating in the GPM retrievals because of increased detection of shallow convection, while upper-level heating patterns remain similar. The use of echo tops and low-level reflectivity gradients greatly reduces midlevel heating from ~2 to 5 km in the mean GPM heating profile, resulting in a more top-heavy profile like TRMM versus a more bottom-heavy profile with much more midlevel heating. Integrated latent heating rates are much better balanced versus surface rainfall for the GPM retrievals using the additional selection criteria with an overall bias of +4.3%.
