This paper describes the basic structure and flow of the rain profiling algorithm for the TRMM Precipitation Radar, and discusses the major assumptions and sources of error in the algorithm. In particular, it describes how the uncertainties in individual parameters affect the attenuation correction and rain estimates. Major parameters involved are the drop size distribution, the phase state of precipitating particles, their density and shape, inhomogeneity of precipitation distribution within the footprint, attenuation due to cloud liquid water and water vapor, freezing height, uncertainty of the surface scattering cross section, and fluctuation of the radar echo signal. Among these parameters that affect the rain estimates, the effect of inhomogeneity of rain distribution is summarized in detail. The paper also describes how these parameters are taken into account in different versions of the standard algorithm 2A25.
This paper presents the statistical properties of bright band (BB) measured by the Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR). Since BB is detected by PR standard version 6 (V6) algorithm 2A23 V6, this paper first outlines 2A23 V6, which classifies rain into three main categories: stratiform, convective, and other. This paper studies BB from the viewpoint of algorithm development. It demonstrates that the detected BB count strongly depends on the antenna scan angle. The BB height (HBB) detected by the PR is compared with the 0°C height computed using the National Centers for Environmental Prediction-Department of Energy Atmospheric Model Intercomparison Project-II reanalysis (NCEP2) data. On average, the BB height is about 500 m below the 0°C height, which is computed from the NCEP2 data. However, a large difference between the above two heights is sometimes observed, mainly due to the false detection of BB. The reliability of BB detection by the current PR algorithm 2A23 V6 could be increased by introducing a simple filter that rejects BBs by regarding them as false when the number of BBs detected in one antenna scan (consisting of 49 antenna beam directions) is less than or equal to 5. An improved version of the filter is planned to be installed in the coming 2A23 V7. Statistics of 10 years of PR data indicate a small but discernible effect of the August 2001 boost of the altitude of the TRMM satellite on the rain type statistics and on the BB statistics. The zonal mean of the maximum value of reflectivity factor in the BB peak (ZmaxBB) over water is almost constant in latitude, but the zonal mean of ZmaxBB over land is less at latitudes above 20 degrees in the northern hemisphere. The width of BB increases as ZmaxBB increases. The zonal mean of the BB width in mid latitude exhibits a small but smooth decrease as the latitude increases.
This paper studies the feasibility of estimating raindrop size distribution (DSD) parameters from Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) measurements. A methodology is described for DSD estimation with PR, in which parameter “ε” or “a” in the Z-R relation Z=aRb is used as a DSD parameter. The ε parameter is an adjustment factor for α in the relation between the attenuation coefficient κ and the effective radar reflectivity factor Ze (κ = αZeβ) that makes the attenuation correction stable by using the path-integrated attenuation estimated from the surface echo as a reference. ε is also recognized as a path-averaged DSD parameter. Large (small) ε corresponds to small (large) a, i.e., to small (large) median volume diameters (D0s) with the assumption of the gamma DSD model. ε exhibits a clear diurnal variation over land suggesting that afternoon convection causes DSDs with large D0s. In contrast, there is no significant diurnal variation over the ocean. ε also exhibits a clear negative correlations with the storm-top height deduced from the PR and with the lightning flash rate, both of which again suggest that deep convections over land produce large D0s. There are several error sources that may produce bias errors in the DSD estimates: non-uniform beam filling (NUBF) within the PR antenna beam, non-liquid hydrometeors aloft (such as hail), and variation in the Normalized Radar Cross Section (NRCS) under rain as compared with no-rain conditions. Preliminary evaluations are performed on these error sources, which generally cause negative errors in ε (i.e., overestimation of raindrop size). Nevertheless, comparisons of PR- and disdrometer-estimated a (in Z=aRb) generally are in agreement at various locations over both land and oceanic sites. This result suggests the feasibility of PR estimation of DSD. It is concluded that, at the present stage, PR estimates of global DSD distribution should be considered to be “qualitative.” Nevertheless, it would be useful to study the tuning of spaceborne radar algorithms and climatological studies of cloud microphysics.
Observations of shallow, isolated convection from the Precipitation Radar (PR) aboard the Tropical Rainfall Measuring Mission (TRMM) observatory were compared with simulations from a rain cell model to examine effects of sub-resolution scale convection on space-borne radar observations. A large sample of shallow isolated convection was obtained from PR data over the central Pacific Intertropical Convergence Zone. Storm top heights, ranged from 1000 to 3750 m, with a mode at 1500 m. A secondary classification was made to identify 3 classes with progressively smaller spatial scales: groups of rainy fields-of-view (FOVs), rainy FOVs with one neighboring rainy FOV, and solitary rainy FOVs. Smaller spatial scales were associated with shallower storms and lower radar reflectivity (dBZ) values. The solitary class was the focus of the modeling comparison. The first objective of the modeling exercise was to explore the possibility that information on the horizontal scale and rain rate intensity of sub-resolution shallow convection could be extracted from the PR observations by finding an optimal fit between simulated and observed dBZ distributions. The height and diameter of simulated rain cells were linked by assuming an average aspect ratio of one. Simulated and observed probability distributions of dBZ were compared for two modes of rain cell variability: a fixed aspect ratio with variable cell dBZ, and a fixed cell dBZ with variable aspect ratio. Both modes resulted in excellent comparisons with observed distributions. The second modeling objective was to compare rain rates from the simulated “ground truth” and simulated “retrieved” rain rates, and to determine the effective sampling area of the modeled FOV. In general, simulated retrieved rain rates had a positive bias, as expected from the non-linear Z-R relation. However, more complex patterns of potential negative biases were evident, depending on the size and intensity of modeled rain cells.
The Precipitation Radar (PR) on the Tropical Rainfall Measuring Mission (TRMM) satellite, which was launched on 28 November 1997, has been collecting global rainfall data for more than 10 years. The monthly estmated-surface rainfall (e_SurfRain) amounts estimated by PR have decreased on average since the satellite altitude was changed from 350 km to 402.5 km in August 2001 to extend its lifetime. There is no significant decrease in e_SurfRain in the five angle bins around the nadir (near-nadir) or in rainfall amounts at a 2-km height. The major causes of the changes in rain estimates due to the orbit boost are (1) sensitivity degradation, owing to the increase of satellite altitude, (2) an increase in the range of surface clutter owing to the increase in the footprint size, and (3) a mismatch between the transmission and reception angles of adjacent radar beams for one in every 32 pulses. The decrease in the echo from a distributed target near the surface attributable to the change of the satellite altitude from 350 km to 402.5 km is 1.21 dB (= 20log(402.5/350)). From a simulation result, we estimate a decrease of 0.5% in surface-rainfall estimates due to the sensitivity degradation. Because the height of the clutter-free bottom increases at off-nadir angles due to the increase in the footprint size from 4.3 km to 5.0 km, the sampling range bins for near-surface rain have been raised, causing the PR to miss low-rain systems more often after the boost than before the boost. Although a beam-mismatch correction is executed in the current 1B21 algorithm, the correction algorithm still results in a negative bias in rain-rate estimates near the surface in the second half of the PR scan. The changes in the surface-rainfall estimates after the boost due to the second and third causes can be estimated as −2.5% and −2.9% by comparing the estimates at off-nadir bins with those at near-nadir bins with the assumption that the true rain rates measured at all angle bins should be the same, on average. It is concluded that the total effect of the orbit boost is a decrease of 5.9% (= 0.5 + 2.5 + 2.9%) in the e_SurfRain amounts in PR version 6 products.
The performance of the version-5 and version-6 Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) products before and after the satellite orbit boost is assessed through a series of comparisons with Weather Surveillance Radar (WSR)-88D ground-based radar in Melbourne, Florida. Analysis of the comparisons of radar reflectivity near the storm top from the ground radar and both versions of the PR indicates that the PR bias relative to the WSR radar at Melbourne is on the order of 1 dB for both pre- and post-boost periods, indicating that the PR products maintain accurate calibration after the orbit boost. Comparisons with the WSR-88D near-surface reflectivity factors indicate that both versions of the PR products accurately correct for attenuation in stratiform rain. However, in convective rain, both versions exhibit negative biases in the near-surface radar reflectivity with version-6 products having larger negative biases than version-5. Rain rate comparisons between the ground and space radars show similar characteristics.
Rainfall rate fields based on TRMM spaceborne radar observations are compared to those based on the new NOAA Next-Generation Quantitative Precipitation Estimation (QPE) high-resolution national mosaic product (Q2). These rainfall fields can be considered as radar products with the largest coverage currently available from space and ground-based radar observations. They probably can also be considered as the most advanced radar rainfall rate products covering a large area. How well do these two products agree? While the accumulated rain rates from all overpasses combined differ by less than 10%, a comparison between the satellite and ground radar probability distribution functions (pdfs) of the instantaneous rain rate shows very large discrepancies. In general, systematic anomalies over the continental U.S. in TRMM radar pdfs compared to the ground-radar pdfs can be recognized. The pdfs of the TRMM radar are generally shifted towards lower rain rates. Moreover, double peaks occur more frequently in the Q2 than in the TRMM radar pdf. Initial results from the comparisons between these two advanced products are presented.
This paper describes a precipitation-retrieval algorithm for the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) that was developed under the Global Satellite Mapping of Precipitation project (GSMaP) by improving the authors' previous algorithm. The basic idea of the GSMaP algorithm is to find the optimal precipitation for which the brightness temperatures (TBs) calculated by the radiative-transfer model (RTM) fit best with the observed TBs. The main improvements of the GSMaP algorithm over the authors' previous work are as follows: (1) use of precipitation-related variable models (precipitation profiles, drop-size distribution, etc.) and precipitation detection and inhomogeneity estimation methods based on TRMM observation studies; (2) use of scattering signals of the TMI Polarization-Corrected Temperature (PCT) at 37 and 85 GHz (PCT37, PCT85) and scattering-signal correction for tall precipitation (thickness between precipitation top level and freezing level (Dtop) larger than 6 km) over land and coastal areas. In order to validate the GSMaP algorithm, we compared its retrievals from TMI TBs in 1998 with the TRMM Precipitation Radar (PR) and Goddard Profiling Algorithm (GPROF) retrievals (2A12 version 6). The results show that (1) over land and coastal areas, the GSMaP retrievals agreed better with PR than GPROF for tall precipitation (Dtop>4 km) weaker than 10 mm h-1, while both GSMaP and GPROF underestimated PR precipitation rates for precipitation heavier than 10 mm h-1; (2) over ocean, the GSMaP retrievals agreed better with PR than GPROF for precipitation heavier than 10 mm h-1, while GSMaP slightly overestimated precipitation weaker than 10 mm h-1 compared to PR and GPROF; (3) The GSMaP algorithm failed to detect some precipitation areas weaker than 2 mm h-1 over sub-tropical oceans. Experimental algorithms with different precipitation-related variable models and retrieval methods using scattering signals were applied to TMI TBs in July 1998 to examine the effect of the above improvements to the GSMaP algorithm. The results show that the improvement of the precipitation profile alleviated the underestimation of precipitation heavier than 10 mm h-1 over land and coastal areas, that the combined use of new physical-related variable models alleviated the underestimation of precipitation heavier than 10 mm h-1 over ocean, and that the use of PCT37 and scattering-signal correction reduced the overestimation of tall precipitation (Dtop>4 km) weaker than 10 mm h-1 over land and coastal areas.
A system has been developed and implemented that integrates passive microwave radiometer data with infrared radiometer data in order to have high temporal (1 hour) and spatial (0.1 degree) resolution global precipitation estimates. The product (GSMaP_MVK) is produced based on a Kalman filter model that refines the precipitation rate propagated based on the atmospheric moving vector derived from two successive IR images. The proposed method was evaluated and compared with other high-resolution precipitation products and the ground-based data collected by the Automated Meteorological Data Acquisition System (AMeDAS) near Japan. It was clearly shown that the approach described in this paper performed better than without the Kalman filter, and the time series of the hourly global precipitation pattern demonstrated the potential capabilities for weather monitoring and typhoon tracking. The GSMaP_MVK product achieved a score comparable to the CMORPH and the 3B42RT products.
Nonuniform beamfilling (NUBF) is a major error source in physical retrieval algorithms for estimating rain rates using satellite-borne passive microwave radiometers. The NUBF effects over the ocean within the beam effective field of view in the 10 GHz channel (FOV10) for the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) are investigated from simultaneous measurements made by the TMI and the Precipitation Radar (PR) aboard the TRMM satellite. They are investigated with respect to variability of a non-uniform parameter defined by the lognormal assumption from January to December 2000. The parameter computed using surface rain rates estimated from PR data tends to be small (large) for stratiform (convective) rainfall in the FOV10. The parameter computed using surface rain rates estimated from brightness temperature (Tb) at 85 GHz is systematically lower in stratiform rainfall than the reference derived by the PR. Systematic differences are not found for the convective cases. To evaluate the effects on rain retrievals due to the systematic differences of the non-uniform parameters, a simple adjustment is applied, and rain retrievals from the observed TMI Tb at 10 GHz vertical polarization are performed over the ocean. Relatively large increases in rain rates retrieved using corrected parameters are found in the tropics, while the increases of the corrected parameters are similar for all latitudes. Effects of the non-uniform parameter differences on the rain retrievals are nonlinear and could be closely related to high background values of the non-uniform parameter in the tropics and more frequent high rainfall intensities.
The rain/no-rain classification (RNC) for the Tropical Rainfall Measuring Mission (TRMM) Imager (TMI) fails in detecting shallow rain observed by the TRMM Precipitation Radar (PR). In this study, the RNC method is revised to use the 37-GHz emission more efficiently to identify shallow rain and is applied to the TMI observation. The results are then evaluated against the RNC made by the PR observation, considered as the “truth.” The revised RNC method (GSMaP2) is compared with the original RNC method (GSMaP1) and the Goddard profiling algorithm (GPROF). GSMaP2 performs well for shallow rain behind cold fronts in the extratropics, where GSMaP1 and GPROF fail, using the 37-GHz emission signature. Through a whole year, a global comparison shows that GSMaP2 performs better than GSMaP1 and GPROF over mid-latitudes. However, GSMaP2 fails in detecting shallow isolated rain over sub-tropical oceans owing to a globally constant value for the vertically integrated cloud liquid water path (LWP) assumed in the forward calculation. Therefore, we parameterize the LWP as a function of storm height from the PR observation over the region where shallow isolated rain is predominant. GSMaP3, in which the parameterization of the LWP is applied to GSMaP2, improves detection of shallow isolated rain over sub-tropical oceans.
This paper presents an evaluation of over-land rain-rate estimates by both the Global Satellite Mapping of Precipitation (GSMaP) algorithm and the standard (GPROF) algorithm for the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI), by comparing them with estimates by the standard algorithm for TRMM Precipitation Radar (PR). This study has the following advantages over previous studies: (1) the errors in rain-rate estimates are decomposed into those caused by rain/no-rain classification and those caused by rain-rate retrieval, (2) the quantitative effects of bright band height (BBH) and land surface physical temperature on retrieval are evaluated; and (3) the role of lower frequency channels (37.0 GHz and lower) for retrieval is investigated. GSMaP yields monthly average and zonal mean rain-rate estimates close to those estimated by the standard algorithm for PR, as it refers to a database produced with PR data. However, GSMaP and GPROF overestimate (underestimate) rain rates for tall (shallow), stratiform (convective), and evening (morning) rainfall. Dependence on storm height (SH) is unavoidable as long as the algorithm relies on the scattering signal caused by solid precipitation for higher frequency channels (85.5 GHz). Lower frequency channels are secondarily used in some algorithms to mitigate the above bias characteristics to some degree. In GSMaP Version 4.7, the severe overestimation seen in GSMaP Version 4.5 when SH is higher than 10 km is mitigated by using 37.0 GHz observations as a scattering signal. The following are indicated for stratiform rainfall with a bright band (SRBB). While the rain-rate estimates are negatively dependent on BBH in GSMaP, the use of 21.3 GHz and 10.7 GHz observations resulted in the cancellation of the dependence on the BBH in GPROF. Although lower-frequency observations are subject to variation in land surface physical temperature, no significant effects of land surface physical temperature on the rain-rate estimates were observed in this study. To improve over-land rain-rate estimates, it is important to make the most effective use of lower-frequency observations.
Global rainfall products of high spatial and temporal resolutions have been provided using combined data from passive microwave (PMW) sensors in low Earth orbit and infrared (IR) radiometers in geostationary Earth orbit (GEO). This study compared six satellite rainfall estimates around Japan with reference to a ground-radar dataset calibrated by rain gauges provided by the Japan Meteorological Agency (JMA) from January through December 2004. Validation results tended to be better for the products with temporal interpolation based upon the morphed technique using GEO IR information. Satellite estimates were poor for light rainfall during the warm season and for very heavy rainfall. Further analyses of satellite estimates were conducted in terms of data sources and surface types. Effective performance by the merger of PMW sounders over the ocean was verified by radar validation, in addition to the best results of the PMW imagers. Overall, validation results over the ocean were best, and results over mountainous regions were worst. Performance was poor over coasts and small islands, due to the problem of PMW retrievals. This study focused on hydrometeor profiles of orographic heavy rainfall over the Japanese Archipelago, which could be related to the poor performance of satellite estimates in very heavy rainfall.
This paper presents recent progress in inter-satellite microwave radiometric cross-calibration to eliminate brightness temperature measurement biases between a pair of radiometer channels operating at slightly different frequencies and incidence angles. The motivation of this research is to develop robust analytical cross-calibration techniques for inter-calibration of various satellite radiometer instruments, with the first projected application being the multi-satellite Global Precipitation Measurement (GPM) constellation to be launched in 2013. The significance of this work is that it will allow the formation of consistent multi-decadal time series of geophysical measurements for multiple satellite microwave radiometers that are free of instrumental biases and other long-term changes in radiometric calibration, which will allow researchers to study global climate change. Descriptions are given for two independent calibration techniques: a Taylor series expansion of the oceanic brightness temperature (Tb) spectrum between dissimilar radiometer channels and a non-linear regression among multi-channel Tb measurements. In the first approach, predictions were made of Tb's at a destination frequency from Tb's of a close by source frequency by expansion of the oceanic brightness temperature spectrum in a Taylor series centered at the source frequency. The relationships between Tb's and frequencies were derived from simulations using a radiative transfer model (RTM), which accounts for the total collected emissions from the ocean surface and the atmosphere. Further, earth incidence angle differences between radiometer channels were transformed in a similar manner using the partial derivatives of Tb with incidence angle derived from RTM simulations. In the second approach, we used a prediction algorithm that relies on the correlation between radiometer Tb's at various frequencies and polarizations and which uses a regression on the Tb's and their non-linear transformations developed using an independent radiative transfer model. As a demonstration, near-simultaneous pair-wise ocean Tb comparisons are presented between the TRMM Microwave Imager (TMI), which is not sun synchronous, and the sun-synchronous polar orbiting WindSat, using oceanic Tb observations from 2003-04. The corresponding results between these two inter-satellite calibration techniques are highly correlated, and results demonstrate that fixed channel-by-channel differences, of order 1-2 K exist between TMI and WindSat. These are significant radiometric calibration differences, which can be removed prior to forming joint data sets of geophysical parameter retrievals.
The Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) 2A12 product consists of unique components configured for land and oceanic precipitation retrievals. This design was based on the vastly different physical characteristics of the retrieval, involving primarily emission over ocean and entirely scattering over land. This paper describes the current status of the TRMM Version 6 (V6) 2A12 product over land and envisioned improvements for TRMM TMI V7 and GPM GMI V1. On a global scale, the 2A12 land algorithm exhibits biases when compared with the TRMM 2A25 (Precipitation Radar (PR) based) and rain gauges. These range from 6 percent for GPCC to 20 percent for 2A25. Closer comparison also reveals regional and seasonal biases, with the largest positive biases found in warm-season convective zones and over semi-arid regions. Some negative biases are found in warm-rain precipitation regimes where scattering at 85 GHz is unable to detect a precipitation signal. On an instantaneous time scale, 2A12 land also produces a positive bias when compared with high-quality radar data from Melbourne, Florida, a TRMM ground validation site. The largest discrepancies occur for rain rates of less than 2 mm h-1. A number of known “anomalies” are highlighted, including overestimation of rainfall in deep convective systems, underestimation in warm-rain regimes, and a number of features associated with the screening component of the algorithm (e.g., snow cover, deserts, etc.). Future improvements for TRMM TMI V7 are described and include the use of ancillary data to determine the underlying surface characteristics and the development of improved brightness-temperature to rain-rate relationships with a more robust data set of TMI and PR matchups, stratified by atmospheric parameters (i.e., surface temperature, atmospheric moisture, etc.) obtained from Numerical Weather Prediction (NWP) model fields. Finally, the promise of an improved land algorithm through the use of high-frequency microwave measurements is described. This will form the basis for the Global Precipitation Measurement (GPM) Global Microwave Imager (GMI) V1 algorithm.
The unresolved rainfall structure within the Field of View (FOV) of a microwave radiometer, such as TRMM Microwave Imager (TMI), causes errors in the rainfall estimation due to the nonlinear relationship between the rain rate and brightness temperature. We have used rainfall structure data from the TRMM Precipitation Radar (PR) to estimate the bias caused by this structure, the random (FOV-to-FOV) error, and a rough estimate of the regional/seasonal scale uncertainty in the bias. We have expressed this bias in terms of a rainfall variance, V, over the FOV in the form: V=a<r>b where <r> is the FOV average rain rate in mm h-1. The data were best fit by a=0.144 mm2h-2 and b=2.46 at the pre-boost 19.35 GHz resolution. The beam filling error is very small relative to the rain rate at small rain rates but increases with increasing rain rate.
The Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) and Precipitation Radar (PR) have enhanced the accuracy of rainfall estimation from satellites over ocean and land. An algorithm to merge TRMM Multi-satellite Precipitation Analysis (TMPA) satellite estimates with the India Meteorological Department (IMD) rain-gauge values is tested for the Indian monsoon region. A daily merged gauge and satellite data product (NMSG) at 1° latitude-longitude resolution for the Indian monsoon region is prepared to depict the large-scale aspects of monsoon rainfall. The satellite product used as a first guess is the TRMM TMPA for daily estimates. Incorporation of IMD gauge data corrects the mean biases of the TMPA values. TMPA alone is able to depict the space-time distribution of monsoon rainfall patterns. The merging of gauge data enhances the value of the satellite information; therefore, the NMSG is more representative than TMPA. Daily, monthly, and seasonal fields are prepared and compared with the land-only gridded data of the India Meteorological Department National Climate Centre (IMDNCC) at the same resolution. This inter-comparison with another independent dataset confirms the utility of the NMSG, produced by this objective analysis algorithm. The comparison of the merged data with the TMPA data reveals the regions where the satellite estimates have mean biases. Objective statistical scores also confirm the goodness of NMSG. The NMSG data are meant for use in verification of large-scale rainfall features from numerical models for the monsoon region.
A new climatology of tropical surface rain is described based on a composite of ten years of precipitation retrievals and analyses from the Tropical Rainfall Measuring Mission (TRMM). This TRMM Composite Climatology (TCC) consists of a combination of selected TRMM rainfall products over both land and ocean. This new climatology will be useful as a summary of surface rain estimates from TRMM (not replacing the individual products) and should be useful as a ready comparison with other non-TRMM estimates and for comparison with calculated precipitation from general circulation models. The TCC mean precipitation for each calendar month and for the annual total is determined by a simple mean of the three chosen products (slightly different combination of products over land and ocean). Over ocean areas, the three TRMM products are those based on the passive microwave (2A12), radar (2A25) and combined retrievals (2B31). Over land, the multi-satellite product (3B43) is substituted for the passive microwave product. The standard deviation (σ) at each point among the three estimates gives a measure of dispersion, which can be used as an indicator of confidence and as an estimate of error. The mean annual precipitation over the TRMM domain of 35°N to 35°S in the new climatology is 2.68 mm d-1 (ocean and land combined) with a σ of .05 mm d-1, or 2.0%. The ocean (land) value is 2.74 mm d-1 (2.54) with a σ/mean of 2.1% (5.4%). The larger dispersion (and assumed error) over land is due to the greater difficulty of satellite rain retrieval over land, especially with passive microwave techniques and especially in mountains and along coasts. The maps of σ and σ/mean indicate these regions of less confidence, including areas over the ocean such as the eastern Pacific Ocean. Examples of values for different latitude bands, seasonal variations, and relations of the individual inputs to the composite mean are given. Comparison with analyses from the Global Precipitation Climatology Project (GPCP) indicates lower values than GPCP for the TRMM composite in middle latitudes over the ocean and over northern Australia and India during their respective summer monsoons.
A long-term observation from the Tropical Rainfall Measuring Mission (TRMM) is analyzed to investigate the Madden-Julian Oscillation (MJO), Kelvin wave, and equatorial Rossby (ER) wave in austral summer seasons. TRMM Precipitation Radar (PR) and Visible/Infrared Scanner (VIRS) measurements are jointly used to clarify convective progression associated with individual modes of the tropical oscillations. Variability in the dynamic and thermodynamic environment involving sea surface temperature (SST), column relative humidity (CRH), and moisture convergence is also examined. A sea surface warming is found to precede the peak MJO convection by ∼ 10 days, while a prior SST increase is not as evident for the Kelvin and ER waves. Moisture convergence and CRH exhibit a horseshoe-like pattern in the composite MJO map, constituted of a pair of off-equatorial maxima and a weak equatorial peak. The Kelvin wave has a moist anomaly on the equator leading the convective peak as theoretically expected, while the moist anomaly also extends poleward without being accompanied by moisture convergence. Moisture convergence leads CRH by a day or two for the Kelvin and ER waves. Moisture convergence is, in contrast, virtually concurrent with CRH for the MJO. The correlation between CRH and deep stratiform coverage is diverse among the three modes of the tropical oscillations. Shallow cumulus and cumulus congestus lead the MJO convective peak by ∼ 1 day, followed by lingering non-precipitating high clouds. ER wave convection is led by moisture convergence but lagged by CRH. The convective progression appears not to proceed in a monotonic way for the ER wave. A possible mechanism to explain MJO propagation is discussed as suggested by a synthesis of the present findings. The Kelvin wave guides the eastward migration of MJO convection onset over the Indian Ocean. The role of the Kelvin wave in MJO propagation diminishes as the MJO enters the west Pacific with the convective area shifting away from the equator. Instead, the Kelvin wave convective heating induces poleward moisture transport and moistens the off-equatorial mid troposphere. The resultant moist anomaly is hypothesized to help trigger the MJO convective burst upon the arrival of ER wave disturbances. Such cooperative processes involving the Kelvin and ER waves could act as a driving engine of some, if not all, MJO episodes.
Global and regional interannual variations of rainfall characteristics over the tropics were examined by applying empirical orthogonal function (EOF) analysis to TRMM PR 3A25 data from December 1997 to December 2007. The TRMM PR 3A25 data and other TRMM datasets detect the interannual variation of rainfall over the tropics, in concert with the SST change, which is closely related to the El Niño/La Niña cycle, including the pseudo-El Niño periods in 2002 and 2004. In addition, we examined the impact of the altitude boost of the TRMM satellite from 350 km to 400 km in August 2001 and found that the boost affects the annual cycle in light rain rates (total, convective, and stratiform). A baseline shift occurs in the annual cycle of light convective rain rate (convRainH) with more (less) frequent occurrence before (after) the boost. In addition, amplitude changes were observed in the annual cycle of light stratiform and total rain rates with more (less) frequent occurrence before (after) the boost. However, for the interannual variation, the coherent spatio-temporal structure of the El Niño/La Niña cycle is evident, and the interannual variation does not indicate any boost impact. In contrast, for heavy convRainH, the number events decreases after the boost, especially over land; however, it is difficult to conclude that this reduction is related to the boost.
Temporal variations of rainfall characteristics over the ocean to the south of the Japan Archipelago during the Baiu season are quantitatively analyzed, and their relationships to the variations of environmental conditions are discussed. The Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) data are utilized for the precipitation, upper-air observation data at six oceanic stations of the Japan Meteorological Agency (JMA), and the reanalysis data produced by JMA are utilized for the environmental data, to study the nine Baiu seasons from 1998 to 2006. Calendar-day averaged time series of nine-year TRMM PR data show a significant increase of tall rain in the latter period of the Baiu season, which accompanies the destabilization of the environmental thermodynamic conditions. Utilizing the values of Convective Available Potential Energy (CAPE), Baiu periods are classified into the earlier-period type (EPT) days and the latter-period type (LPT) days, such that EPT/LPT days are those with negative/positive CAPE anomalies from the entire Baiu-period average. Between the two periods, the convective rain ratio and the stratiform rain characteristics change significantly. In the EPT days, weak stratiform rain associated with the Baiu front is dominant and characterized by an intensity of ∼2.5 mm h-1 and a rain-top height (RTH) of ∼5 km. In the LPT days, rain is dominated by cloud clusters along the Baiu front with larger contribution from the convective rain, associated with stronger and deeper stratiform rain characterized by ∼5.0 mm h-1 intensity and 7.0-7.5 km RTH. Four rain types are classified with the rainfall characteristics obtained from TRMM PR data at 1 degree x 1 degree grids, with the thresholds of convective rain ratio at 35% and rainfall intensity at 2.5 mm h-1. Type 1 represents the weak stratiform rain along the Baiu front, type 2 rain has well organized cloud clusters, and type 3-rain corresponds to relatively shallow convective rain often found under the influence of the subtropical high. The type-4-rain region can be interpreted as a temporally-varying mixture of rain types 1 and 2. Statistics of rainfall types and environmental conditions suggest that dominant rainfall characteristics may be diagnosed utilizing environmental variables such as the equivalent potential temperature at 1000 hPa and the mean sea-level pressure.
The regional characteristics of scale-based precipitation systems on the basis of the size of the systems were investigated using a 10-year dataset from the Precipitation Radar aboard the Tropical Rainfall Measuring Mission satellite. The size of a precipitation system is defined as the number of contiguous rain pixels. The first result is on the diurnal features of precipitation systems stratified by size as small (<102 km2) and large (>104 km2) rain area and stratiform/convective types, respectively. It was found that large overland precipitation systems developed mostly in the evening, following the early-afternoon rainfall maxima of small systems. While large systems showed clear migration properties, small systems had rain peaks at nearly the same local time, that is, without propagation. The diurnal features of small systems were almost uniform over the land or oceans. However, the vertical profiles of rainfall rates obtained for small systems exhibited regional variations. The second result is on the spatial and interannual variations in extreme events. The extreme events are defined as 1000 systems with the highest volumetric rainfall. These events were unevenly distributed and appeared particularly around the mouth of the La Plata basin and the northwestern and western Pacific. The year-to-year variations in these extreme events over the global tropics were consistent with the occurrence of tropical cyclones over the western Pacific.
Rain rates from four algorithms are examined in tropical cyclones (TCs). Old and new versions of the Remote Sensing Systems (RSS) rain estimates (RSS V03 and RSS V04) are compared with the standard version 6 TMI 2A12 and PR 2A25 algorithms, after averaging those down to the 0.25° scale used by RSS. RSS V03 produces more rain by a factor of two than the others, frequently assigning rain rates up to 25 mm h-1 (which is an internal limit for that product). Among the three current algorithms, PR 2A25 produces the most rain when averaged over a 0 to 100 km radius in hurricanes. This results from PR 2A25 assigning much higher grid-scale rain rates (up to 100 mm h-1) in the small fraction of grid boxes having heaviest rain. TMI 2A12 has the least rain, assigning moderate rain rates (5 mm h-1) to more grid boxes than the other products. The differences between algorithms are greatest for the inner regions of Category 3 to 5 hurricanes. In weaker TCs, or further away from the TC center, the three current algorithms tend to agree on mean rain rate. However, they arrive at these areal means from completely different distributions of grid-scale rain rates. PR 2A25 gets a greater fraction of its rain from grid boxes having high rain rates, with little contribution from the light and moderate rain rates. RSS V04 gets much of its rain from grid boxes with 10 mm h-1. TMI 2A12 gets less rain around 10 mm h-1, but balances that with greater contributions both from the occasional higher (15 mm h-1) and more common lower (5 mm h-1) rain rates. At the 0.25° scale, the TMI-based products are better correlated with each other than with PR 2A25. The RSS products are better correlated with PR 2A25 than TMI 2A12 is. All the correlations increase when more zero-rain or light-rain grid boxes are included (i.e., the weaker TCs or greater distances from the center).
The life cycle of deep convective systems over the eastern tropical Pacific (30°N to 30°S, 180 to 90°W) was studied in terms of cloud types, as classified by a split window (11 μm and 12 μm). Hourly split window image data of Geostationary Operational Environmental Satellite (GOES-W) from January 2001 to December 2002 was used in this study. Deep convective systems consist mostly of optically thick cumulus type clouds in the earlier stage and a cirrus type cloud area that increases with time in the later stage. During this analysis period and over the analysis area, the life stage of deep convective system, to a large extent, can be identified by computing the percentage of cirrus type clouds within the deep convective system from a single snap shot of the split window image. Coincident Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) observations were used to study the relationship between the percentage of cirrus type clouds within a deep convective system (i.e., its life stage) and the rainfall rate. It was found that the rainfall rate tends to be larger in the earlier stage of the life cycle when a smaller percentage of cirrus type cloud is present within the deep convection.
This study makes use of a network of nearly 2100 rain gauges over India in order to statistically modify the Tropical Rainfall Measuring Mission (TRMM) algorithm 3B42. The validation and usefulness of the modified product is determined against rain gauge datasets and from the training and forecast phase of the Florida State University (FSU) multimodel superensemble. We use downscaled member model forecasts to construct superensemble forecasts. The member model forecasts are scaled down using the modified high-resolution TRMM rains during the training phase of the multimodel superensemble. We demonstrate that a mesoscale superensemble thus constructed has forecast efficiencies superior to those of all of the member models and to the ensemble mean (ENSM) and bias-corrected ensemble mean (BcorENSM). The forecast procedure includes a high-resolution downscaling of the model forecast rain and the construction of a multimodel superensemble. This procedure clearly provides a much-improved forecast of rain using metrics such as equitable threat scores (ETSs), anomaly correlations, root mean square (RMS) errors, and their bias scores. We also demonstrate that the performance of this modified TRMM algorithm provides results very clearly comparable to those obtained from the direct use of rain gauges throughout India and the TRMM 3B42 covering a domain of 50°S to 50°N.
This study investigates precipitation over the Maritime Continent, comparing the precipitation simulated by a 20 km-grid Meteorological Research Institute General Circulation Model (MRI-GCM) and the near-surface rain data of TRMM 2A25. The focus is particularly the diurnal cycle and its phase distribution of precipitation. The features of the simulated precipitation mostly agree well with observations made over islands and straits having horizontal scales smaller than 200 km. However, these are quite different around larger islands, such as Sumatra and Borneo, particularly in the phase of the diurnal cycle. The MRI-GCM indicates maximum precipitation in the early afternoon on these islands, while the observed precipitation has its maximum at night. In particular, over the inland areas of the larger islands, the simulated diurnal cycle has almost a reversed phase. The simulated precipitation is remarkably weaker than the observation around the western coast of Sumatra Island, where a large discrepancy is also found in the phase distribution along a line perpendicular to the coast. A higher-resolution simulation using a non-hydrostatic model without convective parameterizations substantially improves the phase distribution over Borneo Island. The non-hydrostatic model simulates well the migration of the precipitation zone and the daily maximum at night in the inland areas. In contrast, the GCM fails to simulate the diurnal cycle over islands whose horizontal scale is larger than 200 km, although the 20 km grid spacing is small enough to reproduce the major aspects of the local circulations. The cause for this seems to be the cumulus convective parameterization, which may not adequately represent the coupling of convection and local circulations.
This paper compares satellite microwave radiometer and ground-based radar observations with the Japan Meteorological Agency nonhydrostatic model (JMA-NHM) simulations, using a bulk microphysical parameterization, for a typical rainfall system associated with the Baiu front around the Okinawa Islands, Japan, on 8 June 2004. The JMA-NHM correctly replicated the shape, location and intensity of the precipitation associated with the observed rainfall system. Radar reflectivities and microwave brightness temperatures (TBs) were simulated using output from the JMA-NHM simulations. They were then compared with concurrent corresponding observations by the National Institute of Information and Communications Technology (NICT), CRL Okinawa Bistatic Polarimetric Radar (COBRA), and Advanced Microwave Scanning Radiometer for EOS (AMSR-E). Fairly good agreement was obtained between the simulated and observed reflectivities under the melting layer and TBs at low frequency (18.7 GHz), indicating that the JMA-NHM adequately simulated the amount of liquid hydrometeors. However, the intensity of scattering in the simulations was stronger than that in the COBRA observations above the melting layer and the AMSR-E observations at high frequencies (36.5 and 89.0 GHz). This was due to the fact that the JMA-NHM overestimated the amount and size of snow particles as a result of large depositional growth. The excessive snow contents were reduced by adjusting some of the microphysical processes in the JMANHM: the snowfall speeds were increased and a riming threshold for snow to graupel conversion was changed. These adjustments helped to reduce the amount and size of snow, resulting in further agreement with the COBRA observations. These adjustments also further improved the simulated TBs at high frequencies, especially at 36.5 GHz. However, differences still exist between the simulated and the observed TBs at high frequencies, suggesting that additional adjustment to and improvement of the snow microphysical processes are needed for the application of the model to microwave remote sensing of precipitation.