Download as pdf or txt
Download as pdf or txt
You are on page 1of 27

C H A P T E R

16
Precipitation
O U T L I N E

16.1 Introduction 621 16.4.2 Global Satellite Mapping of


Precipitation 635
16.2 Surface measurement techniques 622
16.4.3 Global Precipitation Climatology
16.2.1 Rain gauge network 622
Project 636
16.2.2 Ground-based radar 623
16.4.4 Global Precipitation Measurement 638
16.3 Estimation from satellite data 625 16.4.5 Climate Prediction Center
16.3.1 VIS/IR algorithms 625 Morphing 640
16.3.2 Passive microwave algorithms 626
16.5 Global precipitation climatology 641
16.3.3 Active microwave algorithms 628
16.3.4 Multisensor algorithms 629 16.6 Summary 643
16.4 Global and regional datasets 630 References 645
16.4.1 Tropical Rainfall Measuring
Mission 631

climatology of global precipitation and associated large


Abstract climatic events. Section 16.6 summarizes and provides
Precipitation is a fundamental component of the global perspectives for the further development of precipita-
water cycle and a key hydrologic variable of the water tion retrieval in the near future.
cycle for meteorology, climatology, and hydrology.
Accurate observations of precipitation and its regional
and global distributions have long been scientific chal-
lenges. The satellite retrieval of precipitation has 16.1 Introduction
matured over four decades of development. This chap-
ter briefly introduces surface measurement techniques, Precipitation is a fundamental component of
the principles of precipitation retrieval from spaceborne
sensors, major satellite-based precipitation datasets,
the global water cycle and a key hydrologic
and global precipitation climatology. Section 16.2 first variable of the water cycle for meteorology,
introduces surface measurement techniques, including climatology, and hydrology. Accurate observa-
gauge- and radar-based networks. Section 16.3 de- tions of precipitation and its regional and
scribes the principles of precipitation retrieval from global distributions have long been scientific
spaceborne sensors. Section 16.4 introduces major
satellite-based precipitation datasets for regional and
challenges. The satellite retrieval of precipitation
global applications. Section 16.5 discusses the has matured over five decades of development.

Advanced Remote Sensing, Second Edition


https://doi.org/10.1016/B978-0-12-815826-5.00016-7 621 © 2020 Elsevier Inc. All rights reserved.
622 16. Precipitation

This chapter briefly introduces surface measure- ground instrument. Typically, a rain gauge mea-
ment techniques, the principles of precipitation sures the integrated volume of an ensemble of
retrieval from spaceborne sensors, major raindrops, whereas a disdrometer counts and
satellite-based precipitation datasets, and global measures drop size at a specific location.
precipitation climatology. Ground-based meteorological radar was
In a physical sense, precipitation is defined as invented to monitor precipitation with a high
“all liquid or solid phase aqueous particles that spatiotemporal resolution (Michaelides et al.,
originate in the atmosphere and fall to the 2009).
Earth’s surface” (Michaelides et al., 2009). It is
a fundamental component of the global water
cycle. Precipitation is moisture fluxes from the
16.2.1 Rain gauge network
atmosphere to the earth’s surface, and it is a Rain gauges are conventionally used to mea-
key hydrologic variable of the water cycle for sure rainfall or snowfall in situ. More than 40
meteorology, climatology, and hydrology. The rain gauge designs are used throughout the
atmosphere obtains approximately three- world (Linacre, 1992). The size of the canister
quarters of its heat energy from the release of opening and the height of the gauge above the
latent heat by precipitation (Kummerow and ground have yet to be globally standardized
Barnes, 1998). Precipitation plays an important (Shelton, 2009). The World Meteorological
role in the climate system by connecting the Organization (WMO) adopted the British
atmosphere and land surface processes. design as the standard gauge, which has a cylin-
Rain and snow are the principal forms of drical diameter of 127 mm and a rim located
precipitation. Unlike other meteorological 1 m above the ground (Linacre, 1992). In the
parameters, rainfall has a high degree of spatio- United States, the standard rain gauge has a
temporal variability and highly nonnormal 203-mm diameter opening located 800 mm
behavior. Among atmospheric variables, it is above the ground.
the most difficult to measure. The spaceetime Existing rain gauges tend to underestimate
structure of precipitation produces top-down actual precipitation, with a bias of 9% on average
effects on dynamic terrestrial hydrologic pro- (Groisman and Legates, 1994; Duchon and
cesses such as runoff generation and soil mois- Essenberg, 2001; Shelton, 2009). The major cause
ture evolution. Thus, accurate observations of of this inaccuracy is the turbulence over the
precipitation and its regional and global distri- canister mouth. Other relevant problems include
butions are quite challenging. wetting of the internal walls of the gauge, evap-
This chapter first introduces surface measure- oration, splashing into or out of the gauge, and
ment techniques, including gauge- and radar-based blowing snow (Shelton, 2009).
networks. Section 16.2 describes the principles of Sufficiently dense rain gauge networks are
precipitation retrieval from space-borne sensors. often required to characterize spatiotemporal
Section 16.3 introduces major satellite-based pre- variations in precipitation (Liu, 2003). Most
cipitation datasets for regional and global applica- countries have their own gauge networks.
tions. Section 16.4 discusses the climatology of WMO now coordinates the Global Telecommu-
global precipitation and associated large climatic nications Network, which distributes rainfall
events. Section 16.5 presents conclusions and pro- information from participating gauges and
vides perspectives for the further development of forms the largest rain gauge network in the
precipitation retrieval in the near future. world.
Many factors, including accessibility, ease of
16.2 Surface measurement techniques maintenance, and topographical aspects, are
considered when determining the location of a
Rainfall is the most common form of precipi- rain gauge. At the network level, the accuracy
tation. Surface rainfall can be measured with a of an observation depends on the total number
16.2 Surface measurement techniques 623
of gauges and their locations. The required of precipitation above the surface of a specified
minimum density of a rain gauge network is volume (Michaelides et al., 2009). Ground-
determined by the temporal resolution of the based radar systems usually have a 1- to 2-km
desired precipitation measurements, study spatial resolution and temporal revisit times of
objective, geographic configuration of the study 15e30 min (Shelton, 2009). Single observing
region, and economic considerations. The points can provide coverage for large areas,
WMO established a minimum density standard with a high spatiotemporal resolution. This
for constructing precipitation gauge networks feature makes them an attractive alternative to
(Table 16.1). precipitation measurements. Furthermore, com-
Rodda (1969) summarized the problems and posite data from several radar systems extend
solutions of network design and proposed a use- the observation area, offering spatial and tempo-
ful distinction among three levels of networks. ral coverages that are better than those obtained
Level I networks acquire information for with rain gauge networks.
national planning purposes related to gross esti- The principal physical basis of precipitation
mates of water resources, surveillance of major radar (PR) systems is the relationship between
storms, and national databases. Level II net- range-corrected, backscattered return power
works supplement level I networks in particular and the physical properties of targets in the ra-
basins or regions and provide extra information dar beam (Shelton, 2009) (Fig. 16.1). The radar
for local planning efforts. Level III networks equation for spherical drops can be expressed as
gather operational information for local water
management. The networks designed for each jKj2 Z
Pr ¼ cr (16.1)
level can differ, and the acquired knowledge r2r
can be used to classify precipitation networks. where Pr is the average backscattered power
In practice, the entire network of rain gauges returning to the receiving antenna over several
may contain components from all three levels. pulses in Watts; rr is the distance from the target
to the radar in Km; jKj2 is the complex index of
refraction, representing the physical characteris-
16.2.2 Ground-based radar tics of the precipitation particles within the radar
Rain gauges measure point precipitation, resolution volume, with values of 0.93 for water
while meteorological radar systems take and 0.197 for ice; and cr is the radar constant,
snapshots of the drop size distribution (DSD) which groups numerical constants and radar

TABLE 16.1 The minimum density for precipitation gauge networks.

Range of norms for minimum Range of provisional norms in difficult


Region type network (km2/gauge) situations (km2/gauge)

I 600e900 900e3000
IIa 100e250 250e1000
IIb 25
III 1500e10,000

I, Flat regions in temperate, Mediterranean, and tropical zones; IIa, Mountainous regions in temperate, Mediterranean, and tropical zones; IIb,
Small mountain/islands with irregular precipitation patterns, requiring dense hydrographic networks; III, Arid and polar zones.
624 16. Precipitation

and Mueller, 1968). Extensive experimental


results have shown that

Antenna Z ¼ aRb (16.4)


beam
where Z is the radar reflectivity in mm /m3 or
6

Emitted and
Target
log-scale decibel units (dB); R is the rainfall rate
backscattered in mm/h; and a and b are coefficients, with a
energy
ranging from 70 to 500 and b ranging from 1.0
Target to 2.0. The coefficients are often empirically
height
Radar antenna determined or calibrated using rain gauge data.
The ZeR relationship is the principal basis of
Range
many retrieval algorithms, but both Z and R
are subject to uncertainties. The relationship is
scale-dependent, and variations in the relation-
FIGURE 16.1 A diagram of the principles of precipitation
radar. Courtesy of Shelton, M. L., 2008. Hydroclimatology. Cam-
ship depend on the DSD of the precipitation,
bridge: Cambridge University Press. radar clutter produced by ground echoes, the
presence of an enhanced ‘‘bright band’’ related
hardware parameters. This last parameter can be to a melting snow layer, and radar signal attenu-
expressed as ation due to heavy rainfall (Morin et al., 2003;
Shelton, 2009; Villarini and Krajewski, 2010).
Pt G2 l2 qH qv csP p3 PR systems typically use longer wavelengths
cr ¼ (16.2)
1024 ln 2l2 of 4e15 cm (e.g., S-band and C-band), and algo-
where Pt is the pulse power of the radar trans- rithms that have been developed and tested for
mitter in Watts; G is the dimensionless antenna low-frequency, high-power systems. In recent
gain; l is the wavelength of the transmitted years, high-frequency, low-power (e.g., X-band)
wave in m; qH is the horizontal bandwidth in ra- systems have received an increasing amount of
dians; qv is the vertical bandwidth in radians; c is attention. Several methods have been developed
the speed of light (3  108 m/s); and sp is the to correct for rain-path attenuation and estimate
pulse duration in seconds. Z is the radar reflec- DSD parameters for X-band systems (Michae-
tivity factor, which represents the intensity of lides et al., 2009).
the return signal in relation to the density and Radar networks have been established in
size of the targets. It is expressed as several countries. The Next-Generation Radar
Z N system, which includes 159 high-resolution
X X Doppler weather radar systems, was deployed
Z¼ D6i ¼ ni D6i ¼ NðDÞD6 dD
vol 0 in the United States (Brown and Lewis, 2005).
In Japan, the Automated Meteorological Data
(16.3)
Acquisition System (AMeDAS), which includes
where D is the drop diameter in mm, and N(D) 20 ground-based radars, was developed by the
dD is the number of particles per unit volume Japan Meteorological Agency (Makihara et al.,
in the diameter range from D to D þ dD. 1996). The Canadian weather radar network,
Both the reflectivity and the rainfall rates are which consists of 31 weather radars, spans
functions of the raindrop size distribution (Stout Canada’s most populated regions.
16.3 Estimation from satellite data 625
To date, both rain gauge and radar data have precipitation information for approximately 50
been widely applied to a variety of fields, years. Launched in 1997, the Tropical Rainfall
including short-term weather and flood fore- Measuring Mission (TRMM) was the first space-
casting and hydrologic modeling (Yuter, 2003; borne PR system (Kummerow and Barnes, 1998).
Neary et al., 2004). These data are also used for On February 28, 2014, the Global Precipitation
calibration and validation purposes when devel- Measurement (GPM) was launched into space,
oping reliable precipitation retrieval algorithms which is a heritage of the TRMM (Hou et al.,
in satellite remote sensing applications (Krajew- 2014). Over 60 empirical and physical retrieval
ski, 1987). algorithms have been developed for the
launched sensors (Ebert and Manton, 1998;
Kubota et al., 2009). Levizzani et al. (2007)
16.3 Estimation from satellite data provided an overview of the state-of-the-art al-
gorithms, sensors, and techniques used in
Surface-based observations of precipitation precipitation retrievals from space. This section
are mostly performed with rain gauges and, if briefly describes the main types of retrieval algo-
available, ground radar. However, observations rithms, which use VIS/IR, PMW, active micro-
are often impossible over oceans. Furthermore, wave sensors, and various combinations of
for some remote and mountainous areas, obser- these technologies (Table 16.2).
vations are scarce and generally not uniform.
Satellites can be used to compensate for the
limitations of surface-based observations. Since
16.3.1 VIS/IR algorithms
the launch of the first meteorological satellite, The primary idea behind the use of VIS/IR
the Television and Infrared Observation Satellite data to infer precipitation information is that
(TIROS-1), in April 1960, a plethora of satellite cold and bright clouds are related to convection.
data have been used to extract meteorological Convective clouds are likely to produce rain.
information (Kidd, 2001). Passive sensors based Satellite observations of cloud tops can be used
on visible (VIS), infrared (IR), and passive micro- indirectly to derive ground rainfall amounts.
wave (PMW) wavelengths onboard geostation- More specifically, IR cloud-top temperatures
ary (GEO) or low earth-orbiting (LEO) satellites are related to the rainfall probability and inten-
have been primarily used to retrieve sity on the ground. The most widely used

TABLE 16.2 Classification of satellite-based retrieval of precipitation.

Sensors Typical algorithms References

VIS/IR GPI; GMSRA; OPI; GriffitheWoodley Arkin and Meisner (1987); Ba and Gruber (2001); Griffith et al. (1978);
algorithm Xie and Arkin (1997)
PMW The Wilheit algorithm; the SSM/I Wilheit et al. (1977); Ferraro (1997); Kummerow et al. (2001);
algorithm; GPROF
Active The TRMM standard PR algorithm Iguchi et al. (2000)
microwave
Multisensor CMORPH; TMPA; GSMaP Joyce et al. (2004); Huffman et al. (2007); Okamoto et al. (2005)
626 16. Precipitation

algorithm is the Geostationary Operational Envi- Despite their limited accuracy, these algo-
ronmental Satellites (GOES) precipitation index rithms have been applied to a range of scales
(GPI) (Arkin and Meisner, 1987), which can be and fields including operational meteorology.
expressed in millimeters as These algorithms have benefited from the avail-
ability of the GEO VIS/IR observations, and they
GPI ¼ rc Fc t (16.5)
provide an adequate level of temporal precipita-
where rc is a constant (3 mm/h) and t is the tion resolution, with high variability in space,
length of the period in hours for which Fc is the time, and intensity.
mean fractional cloudiness, a dimensionless
number ranging from 0 to 1. Fc is obtained
from the IR pixels associated with brightness 16.3.2 Passive microwave algorithms
temperatures <235K over a spatial domain PMW algorithms are more physically direct
more than 50  50 km. GPI is simple, but its than the VIS/IR algorithms. A simplified form
application is limited to 40 Ne40 S, where the of microwave radiative transfer can be expressed
major rain systems are convective in nature as (Wilheit et al., 1999)
(Kidd, 2001). Validations showed that while
GPI estimates are highly correlated (R2 > 0.7) dRðq; fÞ
¼ AþS (16.6)
with gauge data in the Americas, the eastern ds
and central Pacific, and the western Atlantic, where R(q, f) is the radiance in the direction
biases can be as large as 50%e150% per month specified by the polar angles, q and f; s is
(Arkin and Meisner, 1987). distance in the (q, f) direction; A denotes the ab-
Ba and Gruber (2001) proposed the GOES sorption and concomitant emission; and S repre-
Multispectral Rainfall Algorithm (GMSRA); this sents the loss and gain of radiance due to
algorithm uses all five VIS/IR bands from the scattering. At PMW wavelengths, precipitation
GOES satellite to extract the cloud and precipita- particles are the main source of attenuation of
tion extents and then derives rainfall with correc- the upwelling radiation. The radiation emitted
tions to the cloud and precipitation regimes. from atmospheric particles increases the signal
Validation with gauge-adjusted radar data has received at the sensor. Conversely, scattering
demonstrated that the performance of GMSRA induced by the hydrometeors decreases the
is better than that of GPI. The daily estimates radiation stream.
have positive biases to the radar data on the Over the ocean, because emissivity at PMW
order of several millimeters on the global scale. has a narrow range (0.4e0.5), the background
The other major VIS/IR algorithms include the radiometric signal is low and constant. Thus,
GriffitheWoodley algorithm (Griffith et al., using low-frequency bands (<20 GHz), addi-
1978) and the Outgoing Long-wave Radiation tional emissions from precipitation can be used
(OLR) precipitation index (OPI) (Xie and Arkin, to identify and quantify precipitation. Over
1997). Ebert and Manton (1998) used the Geosta- land, the surface emissivity can be as high as
tionary Meteorological Satellite (GMS) network 0.9, and emissions from hydrometeors cannot
and ground radar data to compare retrieval algo- be measured accurately. Alternatively, scattering
rithms for rainfall in the equatorial western caused by ice particles decreases received radia-
Pacific Ocean. Among the 16 VIS/IR algorithms, tion at high frequencies (>35 GHz). Early
large variations in the magnitude of the retrieved research confirmed that PMW radiometry could
rainfall were observed, while the patterns were be used to estimate surface precipitation
similar. (e.g., Savage and Weinman, 1975). Alishouse
16.3 Estimation from satellite data 627
(1983) found that 18, 19.35, and 37 GHz were SI37 is the scattering portion of the algorithm;
particularly useful to extract atmospheric liquid and RR is the rainfall rate (mm/hr). The rainfall
water content, precipitable water, and rainfall rates are calculated from Eq. (16.7b) when
rate. SI37 > 5K over land. TB22v is the brightness tem-
Based on these principles, many approaches perature at band 22 GHz (K). Q19 and Q37 are the
have been proposed to retrieve precipitation. liquid water estimates from the 19 and 37 GHz
The first retrieval algorithm made simple regres- bands, respectively. The rainfall rates are calcu-
sion steps between surface rainfall rates and the lated from Eq. (16.7e) when Q19 > 0.60 mm or
simulated or measured brightness temperatures Q37 > 0.20. Alternatively, the Algorithm-85 is
(Wilheit et al., 1977). Based on a microwave radi- described as follows over land
ative transfer theory, a model is developed 
between microwave brightness temperatures SIL ¼ 451:9  0:44TB19v  1:775TB22v
 (16.8a)
and rainfall rates. The simulations were þ 0:00575TB222v  TB85v
compared with the brightness temperature at a
wavelength of 1.55 cm for the Electrically Scan- R ¼ 0:00513SIL1:9468 (16.8b)
ning Microwave Radiometer on the Nimbus 5
and over ocean
satellite and the rain rates derived from WSR- 
57 meteorological radar measurements. SIW ¼  174:4 þ 0:72TB19V þ 2:439TB22V
Ground-based validations showed that the esti- 
 0:00504TB222V  TB85V
mates were accurate within a factor of 2 for the
range of 1e25 mm/h over ocean areas. (16.8c)
Ferraro (1997) developed a statisticale 2:0343
R ¼ 0:00188SIW (16.8d)
physical algorithm that is most widely used for
operational uses. The algorithm includes two where SIL is the scattering portion of the algo-
retrieval techniques: one is based primarily on rithm over land from the brightness temperature
85 GHz scattering techniques (Algorithm-85) data at bands TB19v, TB22v, and TB85v. The rain-
with the other on a 37 GHz scattering (Algo- fall rates are estimated from Eq. (16.8b) when
rithm-37) over land and emission only over SIL > 10K for rainfall between 0.45e35 mm/h.
ocean. Specifically, the Algorithm-37 is SIW represents the case over ocean. The rainfall
described as follows over land rates are calculated from Eq. (16.8d) when
SIW > 10K for rainfall between 0.20e35 mm/h.
SI37 ¼ 62:18 þ 0:773TB19v  TB37v (16.7a)
Ground evaluation showed that the algorithm
R ¼ 1:3 þ 1:46SI37 (16.7b) using the Special Sensor Microwave Imager
(SSM/I) data had a retrieval error of 50% over
and over ocean ocean and 75% over tropical and midlatitude
Q19 ¼  2:70½lnð290  TB19v Þ  2:84 land (Ferraro, 1997).
(16.7c) More complex retrieval algorithms are based
 0:40 lnð290  TB22v Þ
on probabilistic theory (Smith et al., 1992;
Q37 ¼  1:15½lnð290  TB37v Þ  2:99 Mugnai et al., 1993). The most widely applied al-
(16.7d) gorithm is the Goddard Profiling Algorithm
 0:32 lnð290  TB22v Þ
(GPROF) (Kummerow et al., 2001). This algo-
rithm requires a database of rainfall profiles
R ¼ 0:001707ð100QÞ1:7359 (16.7e)
and the associated brightness temperatures.
where TB19v and TB37v are brightness tempera- Given a vertical distribution of hydrometeors
ture (K) at band 19 and 37 GHz, respectively; represented by R, the radiative transfer equation
628 16. Precipitation

can be used to determine a brightness tempera- precipitation structures over the ocean and
ture vector, Tb. A Bayesian inversion procedure land to be captured (Iguchi et al., 2000).
is performed to find R for a given Tb. The prob- The standard PR algorithm estimates the
ability of a particular profile R for a given Tb is vertical profiles of attenuation-corrected radar
reflectivity values and precipitation rates. The
Pr ðRjTbÞ ¼ PrðRÞ  Pr ðTbjRÞ (16.9)
measured apparent radar reflectivity factor
where Pr(R) is the probability that a certain R Zm(r) is related to the true effective radar reflec-
will be observed, and Pr(TbjR) is the probability tivity factor Ze(r) by the following equations
of observing Tb given a particular rain profile R. (Iguchi et al., 2000; Iguchi, 2007):
The first term on the right side of the above equa-
tion can be derived using a cloud-resolving Zm ðrÞ ¼ Zmt ðrÞ þ dZm ðrÞ (16.10)
model. The strength of the algorithm relies pri- where r is the distance from the radar to the
marily on the accuracy of the profiles in the data- target and dZm ðrÞ is the measurement error in
set. Validations with rain gauge and ground Zm(r). Zmt(r) can be expressed as
radar data indicated that GPROF was biased
negatively (positively) by 9% (17%), with a corre- Zmt ðrÞ ¼ Ze ðrÞAðrÞ (16.11)
lation of 0.86 (0.8) for monthly averages over the
ocean (land) (Kummerow et al., 2001).
AðrÞ ¼ e0:1 lnð10ÞPIAðrÞ (16.12)
Most existing PMW retrieval algorithms have
been optimized for specific satellite sensors. where A(r) is the attenuation factor. PIA(r) is the
Intercomparisons have revealed that each algo- two-way path-integrated attenuation (PIA) in
rithm has its own strengths and weaknesses dB. Several factors affect the PIA, which can be
and that no algorithm is universally superior. described by
Kummerow et al. (2007) suggested the need for Z r
a transparent, parametric algorithm to ensure PIAðrÞ ¼ 2 ðkp ðsÞ þ kCLW ðsÞ þ kwv ðsÞ þ ko2 ðsÞÞds
uniform rainfall products across sensors. 0
The PMW algorithms are limited to observa- (16.13)
tions currently available from the LEO satellites, The two-way attenuation is accounted for by
typically two per day per satellite. The spatial the factor 2 on the right side of this equation.
resolution of the low-frequency bands used Here, kP, kCLW, kWV, and ko2 represent the attenu-
over the ocean is 50  50 km, and the resolution ations due to precipitation, cloud liquid water
of the high-frequency bands used over land is at (CLW), water vapor (WV), and molecular
most 10  10 km. oxygen, respectively.
The major attenuation comes from the precip-
itation particles and is the key to the algorithm.
16.3.3 Active microwave algorithms The retrieval algorithm of precipitation profiles
The use of active microwave (radar) observa- performs classifications in two steps. First, Ze is
tions from space for precipitation retrieval estimated from the measured vertical profiles
started in November 1997 with the launch of Zm, and this estimation corresponds to the atten-
the TRMM (Kummerow and Barnes, 1998), uation correction. Next, the estimated Ze is
which was the first spaceborne PR. It transmits converted into the precipitation rate (R) using a
energy at 13.8 GHz and measures the power power law.
reflected by the precipitation particles and Comparisons have demonstrated that the PR
earth’s surface. It enables three-dimensional retrieval algorithms are accurate on the order
16.3 Estimation from satellite data 629
of the ground-based rainfall data. Thus, they from the GOES-8 and -10, the Meteosat (meteo-
have been widely used as true values against rological satellite) -5 and -7, and the GMS-5.
which other products are compared and evalu- PMW data include the TRMM Microwave
ated (e.g., Aonashi et al., 2009). However, the Imager (TMI), the SSM/I series, the Advanced
attenuation correction and precipitation esti- Microwave Sounding Unit-B (AMSU-B) series,
mates are affected by uncertainties in the param- and the Advanced Microwave Scanning
eters (Iguchi, 2009). The major parameters RadiometerdEarth Observing System (AMSR-
include the DSD, the phase sate of precipitation E). The shape and intensity of precipitation are
particles, their density and shape, the precipita- morphed during the time between PMW sensor
tion distribution within the PR footprint, attenu- scans using a linear interpolation. The process
ation due to CLW and WV, the freezing height, yields spatially and temporally complete PMW
the surface scattering cross section, and fluctua- retrievals, independent of the IR brightness
tions of the radar echo signal. The major factors temperature. Kubota et al. (2009) showed that
are the phase state of the hydrometeors, the CMORPH achieved better results than other
DSD, and the precipitation temperature, multisensor algorithms in areas around Japan.
followed by the heterogeneity of the precipita- Huffman et al. (2007) proposed the TRMM
tion distribution within the radar resolution cell Multisatellite Precipitation Analysis (TMPA)
(Iguchi, 2009). algorithm. It provides a calibration-based
sequential scheme for combining precipitation
estimates from multiple sensors and gauge ana-
lyses at fine scales (0.25  0.25 and every
16.3.4 Multisensor algorithms 3 h). The involved sensors include the TRMM
Comparative studies have shown that the TMI, AMSR-E, SSM/I, AMSU, Microwave
PMW algorithms are more accurate than the Humidity Sounders (MHS), and GEO-IR. In the
VIS/IR algorithms when retrieving instanta- algorithm, the microwave precipitation esti-
neous precipitation data (Kubota et al., 2009). mates are calibrated and combined. Next, IR
However, the VIS/IR algorithms provided bet- estimates are generated from the calibrated
ter long-term retrieval due to better temporal microwave estimates. The two estimates are
sampling of GEO data (e.g., Bauer and Schanz, combined subsequently. Finally, the rain gauge
1998; Ebert and Manton, 1998). The combined data are incorporated. Overall, the TMPA shows
use of multiple sensors may alleviate the defi- reasonable performance at monthly scales. At
ciencies of single-sensor algorithms. In partic- finer scales, it reproduces the surface
ular, the combination of active and passive observationebased histogram of precipitation.
sensors could substantially improve cloud and Okamoto et al. (2005) developed the Global
precipitation retrieval (Michaelides et al., 2009). Satellite Mapping of Precipitation (GSMaP),
In the last two decades, many algorithms have which uses various attributes of precipitation
been proposed with different combinations of derived from TRMM data to retrieve hydromete-
VIS/IR, PMW, and PR data. Here, we discuss or profiles from PR, statistical rain/no-rain clas-
three common multisensor algorithms. sification, and scattering algorithms. With
Joyce et al. (2004) developed the Climate polarization-corrected temperatures (PCTs) at
Prediction Center Morphing (CMORPH) algo- 37 and 85.5 GHz, the combined scattering algo-
rithm, which uses motion vectors derived from rithms estimate surface rainfalls. For stronger
30-min interval GEO IR satellite imagery to rainfalls, PCT37 is used, and for weaker rainfalls,
propagate the relatively high-quality precipita- PCT85 is used. Monthly surface rainfall amounts
tion retrievals from PMW data. IR data are taken are retrieved from six microwave radiometers
630 16. Precipitation

(MWRs). GSMaP_MWR is combined with TMI, In 2007, the Program to Evaluate High-
AMSR-E, AMSR, and SSM/I (Kubota et al., Resolution Precipitation Products was convened
2007). The combined GSMaP moving vector at the WMO headquarters (Turk et al., 2008).
and Kalman filter approach was proposed to Sapiano and Arkin (2009) intercompared five
combine PWM and IR data for producing global high-resolution products including those
precipitation maps with high spatial and tempo- retrieved from the CMORPH, the TMPA, the
ral resolutions (Aonashi et al., 2009; Ushio et al., Naval Research Laboratory (NRL) blended tech-
2009). nique, the National Environmental Satellite, the
Huffman et al. (2015) proposed the Integrated Data and Information Service Hydro-Estimator,
Multisatellite Retrievals for GPM (IMERG) algo- and the Precipitation Estimation from Remotely
rithm. The algorithm combined the precipitation Sensed Information using Artificial Neural Net-
estimates from the various precipitation-relevant works (PERSIANN) (Hong, 2004; Nguyen
satellite PMW sensors and IR rainfall estimates et al., 2018). The results demonstrate that all
calibrated by MW, then gridded, and intercali- the products effectively represented high-
brated to obtain half-hourly 0.1  0.1 data. resolution precipitation events and adequately
MW sensors include dual-frequency precipita- reproduced the diurnal cycle of precipitation.
tion radar (DPR), PR, the GPM Microwave CMRPH and TMPA performed best among the
Imager (GMI), TMI, SSMI/SSMIS (DMSP), multisensor algorithms, with correlations as
AMSR-E (Aqua), AMSU (NOAA), and MHS. high as 0.7. Kubota et al. (2009) compared
IR sensors mainly include GMS, GOES, and GSMaP, TMPA, CMORPH, PERSIANN, NRL-
Meteosat. GPM-IMERG combined other blended algorithms around Japan with reference
algorithms of Multisensor Precipitation Estima- to the radar-AMeDAS dataset. Validations
tion. As CMORPH algorithm, PMW data are showed that GSMaP and CMORPH performed
morphed using the GEO-IRebased cloud better than the other algorithms. Overall, the
motion vectors. Then GEO-IR estimates are estimates were poor for light rainfall during the
used to fill gaps using a Kalman filter when warm season and for heavy rainfall.
the PMW are too sparse. Following the
PERSIANN-CCS (Hong, 2004), at the 60 NeS
latitude belt, the IR Tb image is segmented into 16.4 Global and regional datasets
separable cloud patches using a watershed algo-
rithm, and then cloud patches are classified into With the development of surface-based and
a number of groups using an unsupervised clus- space-based instruments, many regional and
tering analysis. Precipitation is assigned to each global datasets have been produced. The data-
cloud patch group based on a training set of sets provide invaluable information for studying
PMW precipitation samples. These initial precip- precipitation processes and mechanisms.
itation estimates are then adjusted using coeffi- Table 16.3 summarizes the advantages and
cients between matched PMW precipitation disadvantages of major precipitation measure-
and cloud-patch precipitation. Finally, at high ments. Various approaches have been proposed
latitudes, satellite-sounding-based and experi- to overcome the limitations of given measure-
mental sounding channelebased algorithms ments, such as combinations of satellite PMW
and numerical modelebased estimates have and IR data. Maggioni et al. (2016) and Sun
the potential to add value. Finally, precipitation et al. (2018) provided an overview of the global
gauge analyses are used to adjust the satellite precipitation datasets based on satellite. Here,
estimates. we briefly introduce TRMM, GSMaP, Global
16.4 Global and regional datasets 631
TABLE 16.3 Advantages and limitations of major precipitation measurements.

Instruments Advantages and disadvantages

Surface-based Rain gauge Point measurement, type-dependent accuracy, no ancillary information on rain type, etc.,
available, measures rain accumulation (possibly DSD), operational
Ground radar Limited volume measurement, measures possibly polarization and Doppler-spectrum,
operational, retrieval accuracy depends on rain gauge calibration
Space-based Satellite VIS/IR Available from geostationary satellite with good temporal sampling, high spatial resolution,
indirect estimates
Satellite PMW Wide-swath measurement, good calibration, TB measures sensitive to hydrometeors,
channel selection rather flexible, operational
Satellite radar Profile/narrow swath measurement, rather good calibration, measures Z, experimental

Based on Prigent, C., 2010. Precipitation retrieval from space: an overview. Compt. Rendus Geosci. 342, 380e389. Copyright © 2010 Elsevier Masson SAS.
All rights reserved.

Precipitation Climatology Project (GPCP), GPM, the Visible and Infrared Scanner (VIRS), PR,
and CMORPH datasets, all of which are widely the TMI, the Clouds and Earth’s Radiant Energy
used global datasets dedicated to precipitation System, and the Lighting Imaging Sensor. The
monitoring. first three sensors have contributed greatly to
precipitation retrieval. Table 16.4 lists the major
parameters of each instrument.
PR was the first radar designed specifically for
16.4.1 Tropical Rainfall Measuring rainfall monitoring from space. It provides three-
Mission dimensional rainfall structures to achieve quan-
The TRMM is a joint mission between the titative rainfall measurements over land and
National Aeronautics and Space Administration ocean and to improve the accuracy of TMI
(NASA) of the United States and the National measurements by providing rain structure infor-
Space Development Agency (NASDA) of Japan. mation. TMI is a multichannel, dual-polarized
The objective of the TRMM is to observe rainfall MWR that provides data related to rainfall rates
and energy exchange in tropical and subtropical over the oceans. The TMI data, together with the
regions of the world (Kummerow and Barnes, PR data, were the primary dataset for precipita-
1998). Two-thirds of all precipitation falls in the tion measurements. The TMI data combined
tropics, and tropical rainfall is critical to regu- with the PR and VIRS data were also used to
lating the global hydrological cycle. TRMM derive precipitation profiles. VIRS is a passive,
data are valuable for understanding, diagnosing, cross-track scanning radiometer that measures
and predicting global cycling mechanisms and scene radiance in five spectral bands in the VIS
the El Nino and Southern Oscillation (ENSO). through IR spectral regions. Comparative ana-
The TRMM was designed as an LEO satellite lyses of microwave, VIS, and IR data have
stationed between 35 N and 35 S. It has enabled more precise precipitation estimates
produced a wealth of detailed information on than those possible with only VIS and IR data.
tropical rainfall. It has a circular orbit of The TRMM Science and Data Information
350 km and an inclination angle of 35 (Kum- System (TSDIS) is responsible for satellite rain
merow and Barnes, 1998). The satellite includes data (http://tsdis.gsfc.nasa.gov/). Fig. 16.3
five scientific measuring instruments (Fig. 16.2): shows the TRMM satellite product processing
632 16. Precipitation

flow, which includes the retrieval algorithms. V7


algorithms are currently used for processing.
The TRMM data were available starting on
December 20, 1997. The standard TRMM prod-
ucts can be classified into two groups: satellite
and ground validation (GV) products. There
are three distinct product levels. The TRMM
Level-1 products include VIRS radiance, TMI-
calibrated antenna temperature, PR power, and
reflectivity. Level-2 products contain geophys-
ical parameters and include TMI rain profiles
(2A12), PR surface cross sections (2A21), rain
and bright band heights (2A23), TPR rain
profiles (2A25), and TRMM Combined Instru-
ments (2B31). There are three TRMM level-2
surface rain rate products (2A12, 2A25, and
2B31). Level-3 and level-4 products are space-
and time-averaged products. Table 16.5 details
the products.
GV is an essential component of the TRMM.
The GV products are primarily used to support
the validation of satellite retrieval algorithms.
The selected GV radar sites include Darwin,
Australia, multiple sites in Florida, Houston,
FIGURE 16.2 Configuration of the TRMM. Source from Texas, and the TRMM Large Scale
http://trmm.gsfc.nasa.gov/.

TABLE 16.4 Major parameters of Visible and Infrared Scanner (VIRS), TRMM Microwave Imager (TMI), and
precipitation radar (PR).

Sensor VIRS TMI PR

Frequency/ 0.63, 1.6, 3.75, 1 Dual polarization: 10.65, 19.35, 37, 85.5 GHz; Vertical polarization: 13.8 GHz
wavelength 0.8, and 12 mm vertical: 21 GHz
Scanning Cross track Conical Cross track
mode
Ground 2.2 km (a2.4 km) 4.4 km (a5.1 km) at 85.5 GHz 4.3 km (a5.0 km) at nadir
resolution at nadir
Swath width 720 km 760 km (a878 km) 215 km (a247 km)
(a833 km)
Science Cloud parameters, Surface rain rate, rain type, distribution, and 3D rainfall distribution over both land
applications fire, pollution structure; other atmospheric and oceanic and oceans and latent heat release into
parameters the atmosphere

The TRMM satellite was boosted from an altitude of 350 km to one of 402 km in August 2001.
a
Numbers in parentheses represent postboost values.
Modified from Chiu et al., 2006
16.4 Global and regional datasets 633

TRMM V6 Data Processing Overview


NASA-TSDIS/JAXA-EOC

VIRS TMI PR
Level-1A Level-1A Level-1A

1B01 1B11 1B21


TMI Calibrated
VIRS PR
Radiances

Antenna
Level-1

Radiances Temperatures Poewr

1C21
PR
Reflectivities

2A21 2A23
PR Rain Occurrences,
PR Surface
Instantaneous Geophysical

Storm, & Bright


Cross Section Band Height
Parameters
Level-2

2A12 2A25
TM Rainfall PR Rain and Water
Structure and Content Properties
Latent Heating

2B31
Combined
Rainfall Structure
and Latert Heating

3A25
Products
Gridded
Level-3
TRMM

3A11 3B31 3A26


5ºX 5º

TMI Monthly PR Monthly


Combined Monthly Rainfall Structure PR Monthly
Surface Rainfall Rainfall Structure Surface Rainfall
5ºX 5º & 0.5ºX 0.5º
Gridded 0.25º X0.25º Products

SSM mw 3B42
AMSUB mw Combined Instrument
TRMM & Other

AMSR mw 3-hourly Rainfall


CPC IR Calibration Courtey of TSDIS
Satellites

This diagram was updated with TRMM


Level-3

V6 reprocessing inform action provided


3A45 3B43 by the TSDIS
GPCP or Global Monthly
CAMS Rainfall
Rain Gauge

FIGURE 16.3 The TRMM standard satellite product processing flow. Courtesy of Chiu, L.S., Liu, Z., Rui, H., Teng, W.L., 2006.
Tropical rainfall measuring mission data and access tools. In: Qu, J.J., Gao, W., Kafatos, M., Murphy, R.E., Salomonson, V.V. (Eds.), Earth
Science Satellite Remote Sensing. vol. 2. Berlin, Springer-Verlag.
634 16. Precipitation

TABLE 16.5 Characteristics of the Tropical Rainfall Measuring Mission (TRMM) standard satellite products.

TSDIS
reference Product name Product description

1A01 VIRS Raw Data (VIRS) Reconstructed, unprocessed VIRS (0.63, 1.6, 3.75, 10.8, and 12 mm) data
1A11 TMI Raw Data Reconstructed, unprocessed TMI (10.65, 19.35, 21, 37, and 85.5 GHz)
data
1B01 Visible and Infrared Radiance (VIRS) Calibrated VIRS (0.63, 1.6, 3.75, 10.8, and 12 mm) radiances at 2.2 km
resolution over a 720 km swath
1B11 Microwave Brightness Temperature (TMI) Calibrated TMI (10.65, 19.35, 21, 37, and 85.5 GHz) brightness
temperatures at 5e45 km resolution over a 760 km swath
1B21 Radar Power (PR) Calibrated PR (13.8 GHz) power at 4 km horizontal and 250 m vertical
resolutions over a 220 km swath

1C21 Radar Reflectivity (PR) Calibrated PR (13.8 GHz) reflectivity at 4 km horizontal and 250 m
vertical resolutions over a 220 km swath
2A12 Hydrometeor Profile(TMI) TMI Hydrometeor (cloud liquid water, precipitation water, cloud ice,
precipitation ice) profiles in 14 layers at 5 km horizontal resolution,
along with latent heat and surface rain, over a 760 km swath
2A21 Radar Surface Cross-Section (PR) PR (13.8 GHz) normalized surface cross section at 4 km horizontal
resolution and path attenuation (in case of rain), over a 220 km swath
2A23 Radar Rain Characteristics (PR) Rain type; storm, freezing, and bright band heights; from PR
(13.8 GHz) at 4 km horizontal resolution over a 220 km swath
2A25 Radar Rainfall Rate and Profile (PR) PR (13.8 GHz) rain rate, reflectivity, and attenuation profiles, at 4 km
horizontal, and 250 m vertical, resolutions, over a 220 km swath
2B31 Combined Rainfall Profile Combined PR/TMI rain rate, path-integrated attenuation, and latent
heating at 4 km horizontal and 250 m vertical resolutions, over a
220 km swath
3A11 Monthly 5  5 Oceanic Rainfall Rain rate, conditional rain rate, rain frequency, and freezing height for
a latitude band from 40 N to 40 S, from TMI
3A12 Monthly 5  5 mean 2A12, profile, and 5  5 gridded monthly product comprising the mean 2A12 data,
surface rainfall calculated vertical hydrometeor profiles, and the mean surface rainfall
3A25 Monthly 5  5 and 5  5 Spaceborne Total and conditional rain rate, radar reflectivity, path-integrated
Radar Rainfall attenuation at 2, 4, 6, 10, and 15 km for convective and stratiform rain,
storm, freezing, and bright band heights, and snoweice layer depth
for a latitude band from 40 N to 40 S, from PR

3A25 Monthly 5  5 Surface Rain Total Rain rate probability distribution at surface, 2, and 4 km for a latitude
band from 40 N to 40 S, from PR

3B31 Monthly 5  5 Combined Rainfall Rain rate, cloud liquid water, rain water, cloud ice, graupels at 14
levels for a latitude band from 40 N to 40 S, from PR and TMI
3B46 Monthly 1  1 SSM/I Rain Global rain rate from SSM/I

3B42 3B42: 3-Hour 0.25  0.25 TRMM and Calibrated geosynchronous IR rain rate using TRMM estimates
Other-GPI, Calibration Rainfall
16.4 Global and regional datasets 635
TABLE 16.5 Characteristics of the Tropical Rainfall Measuring Mission (TRMM) standard satellite
products.dcont'd

TSDIS
reference Product name Product description

3B43 3B43: Monthly 0.25


 0.25 TRMM and Merged rain rate from TRMM, geosynchronous IR, SSM/I, rain
Other Sources Rainfall gauges
CSH CSH: Monthly 0.5  0.5 Convective & TRMM Monthly 0.5  0.5 Convective/Stratiform Heating (it is a
Stratiform Heating TRMM experimental product)

Courtesy of Chiu, L.S., Liu, Z., Rui, H., Teng, W.L., 2006. Tropical rainfall measuring mission data and access tools. In: Qu, J.J., Gao, W., Kafatos, M.,
Murphy, R.E., Salomonson, V.V. (Eds.), Earth Science Satellite Remote Sensing. vol. 2. Berlin, Springer-Verlag.

Biosphere-Atmosphere Experiment in Amazonia the Climate Prediction Center (CPC) (Janowiak


(TRMM-LBA), Brazil. Characteristics of the GV et al., 2001). The spatial resolution of the data is
products are listed in Table 16.6. 0.03635 (4 km at the equator). The latitude
ranges from 60 N to 60 S. The temporal resolu-
tion is approximately 30 min. Fig. 16.4 shows
the structure and flow of the GSMaP project.
16.4.2 Global Satellite Mapping of
Currently, the datasets produced from the
Precipitation GSMaP project can be downloaded from their
The GSMaP project, which ran from 2002 until website at http://sharaku.eorc.jaxa.jp/GSMaP_
March 2008, was supported by the Japan Science crest/index.html. The standard version of the
and Technology Agency. Its objectives were to GSMaP datasets includes GSMaP_TMI
create high-precision and high-resolution global (retrieved from TRMM/TMI algorithm),
precipitation maps using data derived from GSMaP_MWR (retrieved from six spaceborne
PMW radiometers, improve the precipitation MWRs), GSMaP_MWRþ (retrieved from six
physical model and rain rate retrieval algo- spaceborne MWRs with AMSU-B products),
rithms, evaluate the accuracy of various GSMaP GSMaP_MVK (retrieved from the MWR-GEO
productions, and prepare for the coming GPM. IR combined algorithm), GSMaP_MVKþ
Since 2007, the activities of the GSMaP project (retrieved from the MWR-GEO IR combined
have been implemented by the Precipitation algorithm with AMSU-B products), and other
Measuring Mission Science Team of the Japan rainfall estimates from a PMW radiometer. The
Aerospace Exploration Agency (JAXA). data from 2006 were produced in a near real-
The GSMaP project used multiple datasets time data processing system developed and
from several sensors as inputs to retrieve rainfall operated at the JAXA Earth Observation
rates. The data sources include the MWR data- Research Center (http://sharaku.eorc.jaxa.jp/
sets observed from the LEO satellites and the GSMaP/index.htm). Table 16.7 details the
infrared (IR) radiometer datasets obtained from spatial and temporal resolution, data period,
the GEO satellites. The PMW sensors include data range, and data source of the GSMaP
imagers and sounders. The available imagers datasets.
are TRMM TMI, AMSR, AMSR-E, and SSM/I. The GSMaP datasets have been compared
The most recent sounder is AMSU-B. In addi- with rain gauge data (Kubota et al., 2007, 2009;
tion, the GEO IR datasets used in the current Aonashi et al., 2009). Overall, the validation re-
version of the data processing system are from sults over the oceans were the best, and the
636 16. Precipitation

TABLE 16.6 Characteristics of the Tropical Rainfall Measuring Mission (TRMM) ground validation (GV) products.

TSDIS
reference Product name Product description

1B51 Radar Reflectivity Volume scan of radar reflectivity, differential reflectivity, and mean velocity
(if available), truncated at 230 km range, at original radar resolution, coordinate,
and sampling
1C51 C51:Half-Hourlya Calibrated Volume scan of calibrated radar reflectivity and differential reflectivity
Radar Reflectivity (if available) and corresponding QC masks, truncated at 200 km range, at original
radar resolution and coordinate
2A52 Half-Hourlya Rain Existence Percent of rain in the radar volume scan
2A53 a
Half-Hourly 2 km Radar Site Instantaneous rain rate over an area of 300 km  300 km
Rain Map
2A54 Half-Hourlya 2 km Radar Site Instantaneous rain type classification (convective, stratiform) over an area of
Rain Type Map 300 km  300 km
2A55 Half-Hourlya Radar Site 3-D Instantaneous radar reflectivity and vertical profile statistics over an area of
Reflectivity 300 km  300 km, at 2 km horizontal and 1.5 km vertical resolution
2A56 1 Minute Average and Peak Time series of rain gauge rain rates over the radar site rain gauge network
Rain Gauge Rain Rate
3A53 5-Day 2 km Ground Radar Site Surface rain total from ground radar
Rain Map
3A54 Monthly 2 km Ground Radar Surface rain total from ground radar
Site Rain Map

3A55 Monthly 2 km Ground Radar Vertical profile of reflectivity and contoured frequency by altitude diagrams for
Site 3-D Rain Map stratiform, convective, and anvil rain over land and water
a
Volume scans at original radar sampling rates within half an hour of TRMM satellite coincidence.
Courtesy of Chiu, L.S., Liu, Z., Rui, H., Teng, W.L., 2006. Tropical rainfall measuring mission data and access tools. In: Qu, J.J., Gao, W., Kafatos, M.,
Murphy, R.E., Salomonson, V.V. (Eds.), Earth Science Satellite Remote Sensing. vol. 2. Berlin, Springer-Verlag.

results over mountainous regions were the 16.4.3 Global Precipitation Climatology
worst. The GSMaP_MVK product achieved a Project
score comparable with the CMORPH products
(Ushio et al., 2009). Heavy precipitation areas Long-term series of satellite-retrieved global
over land and coasts were not easily identified, precipitation are not yet available. GPCP was
and precipitation events with more than established by the World Climate Research
10 mm/h were underestimated (Aonashi et al., Program in 1986. The aim of the GPCP is to
2009). Our results show that GSMaP_MWRþ provide area- and time-averaged global precipi-
and MVKþ underestimated precipitation over tation analysis products. The products play an
the Poyang Lake Basin of China from January important role in understanding seasonal,
2003 to December 2006 (Fig. 16.5). The reasons annual, and long-term global energy and water
for this discrepancy are unknown. cycle variabilities.
16.4 Global and regional datasets 637

FIGURE 16.4 Structure and flow of the GSMaP project. Courtesy of Takahashi, N., Iguchi, T., Aonashi, K., Awaka, J., Eito, H.,
Fujita, M., Hashizume, M., Hirose, M., Inoue, T., Kashiwagi, H., Kozu, T., Satoh, S., Seto, S., Shige, S., Shimizu, S., Takayabu, Y., Oka-
moto, K., 2004. The global satellite mapping of precipitation (GSMaP) project: Part II algorithm and precipitation model development. 2nd
TRMM International Conference. September 6e10, 2004. Tokyo, Japan.

The GPCP product is a mature dataset of The GPCP datasets include three global pre-
merged IR and PMW retrievals and rain gauge cipitation sub-datasets (Huffman et al., 2001;
observations (Gruber and Levizzani, 2008). The Adler et al., 2003; Xie et al., 2003; Adler et al.,
primary IR data sources are GOES, GMS, Meteo- 2018): the GPCP (Version 2 and 2.3) monthly
sat, and NOAA polar-orbiting satellites. The Satellite-Gauge, the Pentad, and the One-
microwave data source is the SSM/I on DMSP Degree Daily (1DD). The primary website for
satellites. The station observations are primarily the GPCP products is the WMO’s World Data
provided by the Global Precipitation Clima- Center (http://lwf.ncdc.noaa.gov/oa/wmo/
tology Centre operated by the German Weather wdcamet-ncdc.html) at NOAA’s National
Service. The satellite precipitation estimates Climatic Data Center. A backup FTP site for
include SSM/I emission estimates, SSM/I the Version 2 and 1DD products is located at
scattering estimates, TOVS (TIROS Operational NASA’s Goddard Space Flight Center (GSFC).
Vertical Sounder)-based estimates, and GPI and The primary product is 2.5 Version 2 monthly
OPI precipitation indices. precipitation analyses for the period from 1979
638 16. Precipitation

TABLE 16.7 Status of each Global Satellite Mapping of Precipitation (GSMaP) product.

Spatial Data
Product resolution Temporal resolution Data period range Data source

GSMaP_TMI 0.25 Hourly, daily, monthly, 1998e2006 40 TRMM TMI


yearly
GSMaP_MWR 0.25 Hourly, daily, monthly, 1998e2006 60 TRMM/TMI, Aqua/AMSR-E, ADEOS-II/
yearly AMSR,
DMSP/SSMI-F10,11,13,14,15
GSMaP_MWRþ0.25 Hourly, daily, monthly 2003e06 60 TRMM/TMI, Aqua/AMSR-E, ADEOS-II/
AMSR,
DMSP/SSMI-F10,11,13,14,15 AMSU-B
GSMaP_MVK 0.1 Hourly, daily, monthly 2003e06 60 TRMM/TMI, Aqua/AMSR-E, ADEOS-II/
AMSR,
DMSP/SSMI-F13,14,15 GOES-8/10,
Meteosat-7/5, GMS
GSMaP_MVKþ 0.1 Hourly, daily, monthly 2003e06 60 TRMM/TMI, Aqua/AMSR-E, ADEOS-II/
AMSR,
DMSP/SSMI-F13,14,15, NOAA/AMSU-B
GOES-8/10, Meteosat-7/5, GMS
GSMaP_NRT 0.25 Hourly, daily, monthly 2007epresent 60 TRMM/TMI, Aqua/AMSR-E,
DMSP/SSMI-F13,14,15, NOAA/AMSU-B
GOES-8/10, Meteosat-7/5, GMS, MTSAT

to the present. Both the Pentad and the 1DD precipitation, and accurately estimating precipi-
products are constrained by monthly analyses. tation in complex terrain areas.
The 1DD dataset was produced on a resolution
of 1 from January 1997 to the present. This data-
set shows flexibility in computing time and 16.4.4 Global Precipitation
spatial averages. The pentad precipitation
analyses are 2.5  2.5 analyses from 1979 to
Measurement
the present and use the pentad CMAP (CPC The GPM is a joint mission between the
Merged Analysis of Precipitation) estimates. NASA of the United States, the NASDA of
The pentad CMAP merges the IR-based GPI, Japan, and other international institutes. The
the SSM/I scattering-based ALG85, the SSM/I objective of the GPM is to improve the global
emission-based algorithm, the MSU-based rainfall estimation and advance scientific under-
method, the OLR-based OPI, and the gauge- standing of the earth’s water and energy cycle.
based analyses. The GPM, an outgrowth of TRMM, have the
Gruber and Levizzani (2008) provided an capability to provide physically based retrievals
extensive assessment of the GPCP datasets, on a global basis with higher resolution and
which have been applied in global change accuracy. It is able to observe weak rainfall
studies. However, the GPCP still faces chal- (0.5 mm/h) and solid precipitation whose obser-
lenges. Major issues include determining precip- vations are significant in regions in mid- and
itation over the open ocean, determining solid high-latitudes and mountainous areas.
(A) (B)

16.4 Global and regional datasets


FIGURE 16.5 Rain gaugeebased monthly measurements for (A) GSMaP MWRþ and (B) MVKþ datasets over the Poyang Lake Basin of China from
January 2003 until December 2006.

639
640 16. Precipitation

TABLE 16.8 Major parameters of GPM Microwave Imager (GMI) and Dual-frequency Precipitation Radar (DPR).

Sensor TMI PR

Frequency/ Dual polarization: 10.65, 18.7, 36.64, 89.0, 166.0 GHz; Vertical polarization: 13.8 GHz
wavelength Vertical: 23.8, 183.31  3, 183.31  7 GHz
Scanning mode Conical Cross track
Ground 3.4 km at 183.31 GHz 5.0 km at nadir
resolution
Swath width 904 km 245 km
Science Surface rain rate, rain type, distribution, and 3D rainfall distribution over both land and
applications structure; other atmospheric and oceanic parameters oceans, and latent heat release into the
atmosphere

GPM Core Observatory (GPMCO) is the heart GSFC is responsible for management of GPM
of the GPM constellation. The GPMCO was Validation Network (GV). GV started on August
designed as an LEO satellite stationed between 8, 2006. The VN database includes TRMM PR
65 N and 65 S. It has a circular orbit of 407 km data and coincident ground radar operational
and an inclination angle of 65 (Smith et al., meteorological networks. Validation areas are
2007). The satellite includes two scientific same as TRMM. VN products are processed by
measuring instruments: DPR and GMI. The VN software (online at http://www.ittvis.
two sensors have contributed greatly to improve com), and the processes include data ingest
precipitation retrieval. Table 16.8 lists the major and preprocessing, the resampling of PR and
parameters of each instrument. GR, and the statistical analysis and display of
DPR was the first dual-frequency radar specif- the matching data volumes (Schwaller and
ically for rainfall monitoring from space. Morris, 2011).
Similarly to the TRMM, it provides three-
dimensional rainfall structures to achieve quan-
titative rainfall measurements. GMI is a
dual-polarized MWR with more multichannels.
16.4.5 Climate Prediction Center
The GMI data combined with the DPR is able Morphing
to derive precipitation profiles in higher latitude. CPC of the NOAA is responsible for
GPM dataset were available starting on CMORPH products. CMORPH dataset were
March 2014 (https://pmm.nasa.gov/data- available starting from December 2012 to
access/downloads/gpm). The standard GPM October 2017 between 60 N and 60 S. CMORPH
products can be classified into two groups: satel- includes 0.07  0.07 , 0.5 h, and 0.25  0.25 ,
lite and GV products. There are three distinct three hourly products (available at https://
product levels. The GPM Level-1 products climatedataguide.ucar.edu/climate-data/
include GMI-calibrated antenna temperature, cmorph-cpc-morphing-technique-high-
DPR power, and reflectivity. Level-2 products resolution-precipitation-60s-60n). CMORPH
contain surface cross sections, DPR rain profiles, first used cloud motion vectors derived from
and rain rate based on Level-1 products. Level-3 GEO IR satellite imagery to propagate the
products are space- and time-averaged global precipitation retrievals from PMW data.
products based on Level-2 products. Table 16.5 Many studies evaluated and compared
details the products. CMORPH with other satellite precipitation
16.5 Global precipitation climatology 641
products, such as TRMM 3B42 and GSMap Convergence Zone area, with expansions into
(Sapiano and Arkin, 2009; Dinku et al., 2010; the Amazon, and over the extreme eastern trop-
Kidd et al., 2012; Li et al., 2013). The results ical Indian Ocean. Two high-precipitation zones
show that CMORPH performed relatively poor were identified along the eastern coastal regions
over wintery, complex land surfaces with light of Asia and North America, coinciding with the
precipitation. Generally, CMORPH underesti- midlatitude storm tracks associated with the
mated a rainfall of 6.5e1.4 mm/day. Kuroshio and Gulf Stream currents. In the south-
ern hemisphere, the South Atlantic Convergence
Zone displayed local precipitation maxima that
16.5 Global precipitation climatology extended southeastward from southern Brazil
and the South Pacific Convergence Zone to
With the aid from satellite remote sensing New Guinea.
systems, spatial and temporal variations of Fig. 16.7 shows the zonal mean precipitation
precipitation can be predicted at various scales. generated from GPCP datasets for land, ocean,
Because the GPCP dataset spans 25 years, it is and the globe over the period 1979e2003. The
commonly used in climatological studies. Here, mean global ocean precipitation levels had
we illustrate the global precipitation climatology off-equatorial maxima and relative minima on
derived from this dataset. the equator, reflecting the mean position of the
Fig. 16.6 shows the mean annual precipitation convergence zones in tropics. In contrast, the
produced with the GPCP dataset for the period zonal mean values of the land-only zonal had a
from 1979 to 2003. The highest annual precipita- single maximum centered on the equator. The
tion levels occurred in the western tropical zonal mean profiles of both hemispheres in
Pacific, the eastern Pacific Intertropical both the ocean- and land-only datasets had

Mean Annual GPCP Precipitation (mm/day)

60N

30N

EQ

30S

60S

60E 120E 180 120W 60W 0 60E

1 2 3 4 5 6 7 8 9

FIGURE 16.6 Mean annual GPCP precipitation for the period 1979e2003. Courtesy of Gruber, A., Levizzani, V., 2008. Assess-
ment of global precipitation products. WCRP Series Report No. 128 and WMO TD-No.1430, 1e55.
642 16. Precipitation

8 maxima in the midlatitudes, reflecting the mean


position of the storm tracks.
7 OCEAN
Precipitation Rate (mm/day)

At a seasonal scale, the mean precipitation


6 LAND
rate appeared to be highest in the near-
TOTAL
5 equatorial tropical belt for all months of the
year (Fig. 16.8). From January to March, the
4
precipitation maximum occurred in the southern
3 hemisphere. From May to early September, the
2 most intense precipitation was in the northern
hemisphere. Low precipitation rates in the
1
tropics and subtropics were centered at approx-
0 imately 20 N from January to April and approx-
90 N 70 50 30 10 10 30 50 70 90 S
imately 15 S from June to September.
FIGURE 16.7 Zonal mean of GPCP precipitation for land Fig. 16.9 shows the variations of monthly
(dashed line), ocean (dotted line), and globe (solid line) over the precipitation for land (green line), the oceans
period 1979e2003. Courtesy of Gruber, A., Levizzani, V., 2008. (blue line), and the globe (black line) over the
Assessment of global precipitation products. WCRP Series Report
No. 128 and WMO TD-No.1430, 1e55.
period 1979e2003. The precipitation rate was

FIGURE 16.8 Annual cycle of zonal precipitation (mm/day) averaged over the period 1979e2003. Courtesy of Gruber, A.,
Levizzani, V., 2008. Assessment of global precipitation products. WCRP Series Report No. 128 and WMO TD-No.1430, 1e55.
16.6 Summary 643
3.5

OCEAN
LAND
TOTAL

3.0
Precipitation Rate (mm/day)

2.5

2.0

1.5
79 81 83 85 87 89 91 93 95 97 99 01 03

FIGURE 16.9 Monthly precipitation for land (green), ocean (blue), and globe (black) over the period 1979e2003. Heavy lines
indicate 12-month running means. Courtesy of Gruber, A., Levizzani, V., 2008. Assessment of global precipitation products. WCRP
Series Report No. 128 and WMO TD-No.1430, 1e55.

higher over the ocean than over the land by day and a low-frequency (2e3 years) variability.
approximately 1 mm/day. The land series had However, no significant relationship between
larger variations than the ocean, even for the ENSO and global precipitation anomalies was
12-month running mean values. The global total observed. Two major volcanic eruptions,
time series were produced by combining the El Chichon and Mt. Pinatubo, did not signifi-
land and ocean precipitation values. cantly influence the anomalies. Conversely, a
The temporal variations of monthly precipita- significant relationship between the precipita-
tion may be relevant to the ENSO, which is a tion anomalies and the Ni~no-3.4 SST index over
quasiperiodic climate pattern characterized by the oceans or land for both dry and wet condi-
roughly 5-year variations in the surface temper- tions was observed. The relationship was strong
ature of the tropical eastern Pacific Ocean. especially in 1986/87 and 1997/98, two largest
ENSO is often described with the Ni~ no-3.4 sea ENSO events of the 20th century.
surface temperature (SST) index. Fig. 16.10
shows the total (top), ocean (middle), and land
(bottom) tropical (30 Ne30 S) averages of
16.6 Summary
monthly precipitation anomalies. The mean
tropical precipitation showed month-to-month
Precipitation includes rainfall and snowfall,
variabilities generally within 0.2 mm/day.
which are fundamental process in the global wa-
The 12-month running mean of global precipita-
ter cycle. It has a high degree of spatiotemporal
tion had a low amplitude of less than 0.1 mm/
variability, highly nonnormal in behavior. The
644 16. Precipitation

0.6 3
0.4 TOTAL 2

mm day-1
0.2 1

ºC
-0.0 0
-0.2 –1
-0.4 El Chichon Pinatubo –2
St. Helens
-0.6 –3
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03
0.6 3
0.4 2
OCEAN
mm day-1

0.2 1

ºC
-0.0 0
-0.2 –1
-0.4 St. Helens El Chichon Pinatubo –2
-0.6 –3
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03
0.6 3
Pinatubo
0.4 2
mm day-1

0.2 1

ºC
-0.0 0
-0.2 –1
-0.4 St. HelensEl Chichon LAND –2
-0.6 –3
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03

FIGURE 16.10 Tropical (30 N-30 S) averages of monthly precipitation anomalies (mm/day) for total (top), ocean (mid-
dle), and land (bottom). Vertical dashed lines indicate months with significant volcanic eruptions. The thin black curves are
the Ni~no-3.4 SST index ( C). Heavy lines are 12-month running means. Courtesy of Gruber, A., Levizzani, V., 2008. Assessment
of global precipitation products. WCRP Series Report No. 128 and WMO TD-No.1430, 1e55.

spaceetime structure of precipitation produces Fast development of space-based instruments


top-down effects on numerous hydrologic and retrieval approaches fosters to produce
processes. Many rain gauge networks and many regional and global precipitation datasets.
ground-based and space-borne radar techniques The TRMM, GSMaP, and GPCP datasets are
have been developed to capture spatial and well-known examples of global long-term series
temporal features of precipitation. precipitation datasets. The global datasets
Since 1970s, tens of satellite retrieval algo- provide invaluable information for studying
rithms have been proposed on a basis of satellite precipitation processes and mechanisms at
observation from meteorological sensors and PR. regional and global scales.
The algorithms use VIS/IR, PMW, active micro- Global warming and climate change are likely
wave data, or various combinations of these to continue, necessitating a comprehensive
technologies. The GPI algorithm is the most understanding of global precipitation for the
commonly used VIS/IR algorithm. The Ferraro sustainable development of our changing earth.
and the GPROF algorithms are the most In addition to surface-based instruments,
commonly used PWV algorithms. Combination space-based observations, especially those that
of active and passive sensors substantially use information from the TRMM, have advanced
improved cloud and precipitation retrieval and our spatial and temporal understandings of
was particularly welcomed in the last two precipitation intensity. As an international effort,
decades. the GPM launch in 2014 extends the TRMM
References 645
measurements to latitudes up to 65 ; increases Brown, R.A., Lewis, J.M., 2005. Path to NEXRAD: Doppler ra-
the sampling frequency to 3 h; improves sensi- dar development at the national severe storms laboratory.
Bull. Am. Meteorol. Soc. 86, 1459e1470.
tivity to light and heavy rainfalls, with a detect- Chiu, L.S., Liu, Z., Rui, H., Teng, W.L., 2006. Tropical rainfall
able range from 1 mm/h to more than 200 mm/ measuring mission data and access tools. In: Qu, J.J.,
h; and discriminates between rain and snow. Gao, W., Kafatos, M., Murphy, R.E., Salomonson, V.V.
Long-term global precipitation monitoring is (Eds.), Earth Science Satellite Remote Sensing, vol. 2.
a challenge. Combining observations at different Springer-Verlag, Berlin.
Dinku, T., Ruiz, F., Connor, S.J., Ceccato, P., 2010. Validation
VIS, IR, and microwave wavelengths from pas- and intercomparison of satellite rainfall estimates over
sive and active modes and from both low and Colombia. J. Appl. Meteorol. Climatol. 49 (5), 1004e1014.
geostationary orbits can produce hourly samples Duchon, C., Essenberg, G., 2001. Comparative rainfall obser-
and long-term precipitation records for the entire vations from pit and aboveground rain gauges with and
globe. The International Precipitation Working without wind shields. Water Resour. Res. 37, 3253e3263.
Ebert, E.E., Manton, M.J., 1998. Performance of satellite rain-
Group is focusing on satellite-based quantitative fall estimation algorithms during TOGA COARE.
precipitation measurements. With the interna- J. Atmos. Sci. 55, 1537e1557.
tional efforts, we are approaching an important Ferraro, R.R., 1997. Special sensor microwave imager derived
juncture for the retrieval of global precipitation global rainfall estimates for climatological applications.
information from multiple sensors. J. Geophys. Res. 102, 16715e16735.
Griffith, C.G., Woodley, W.L., Grube, P.G., Martin, D.W.,
Stout, J., Sikdar, D.N., 1978. Rain estimates from geosyn-
chronous satellite imagery: visible and infrared studies.
References Mon. Weather Rev. 106, 1153e1171.
Groisman, P., Legates, D., 1994. The accuracy of United States
Adler, R.F., Huffman, G.J., Chang, A., Ferraro, R., Xie, P., precipitation data. Bull. Am. Meteorol. Soc. 75, 215e228.
Janowiak, J., Rudolf, B., Schneider, B., Curtis, S., Gruber, A., Levizzani, V., 2008. Assessment of global precip-
Bolvin, D., Gruber, A., Susskind, J., Arkin, P., 2003. The itation products. In: WCRP Series Report No. 128 and
version 2 Global Precipitation Climatology Project WMO TD-No.1430, pp. 1e55.
(GPCP) monthly precipitation analysis (1979-present). Hong, Y., 2004. Precipitation estimation from remotely
J. Hydrometeorol. 4, 1147e1167. sensed information using artificial neural network-cloud
Adler, R.F., Sapiano, M.R.P., Huffman, G.J., Wang, J.J., classification system. J. Hydrometeorol. 43, 1834e1852.
Gu, G.J., Bolvin, D., Chiu, L., Schneider, U., Becker, A., Hou, A.Y., Kakar, R.K., Neeck, S., Azarbarzin, A.A.,
Nelkin, E., Xie, P.P., Ferraro, R., Shin, D.B., 2018. The Kummerow, C.D., Kojima, M., Oki, R., Nakamura, K.,
global precipitation climatology project (GPCP) monthly Iguchi, T., 2014. The global precipitation measurement
analysis (new version 2.3) and a review of 2017 global mission. Bull. Am. Meteorol. Soc. 95, 701e722.
precipitation. Atmosphere 9, 14. Huffman, G.J., Adler, R.F., Morrissey, M., Bolvin, D.T.,
Alishouse, J.C., 1983. Total precipitable water and rainfall de- Curtis, S., Joyce, R., McGavock, B., Susskind, J., 2001.
terminations from the SeaSat scanning multichannel mi- Global precipitation at one-degree daily resolution from
crowave radiometer. J. Geophys. Res. 88, 1929e1935. multi-satellite observations. J. Hydrometeorol. 2, 36e50.
Aonashi, K., Awaka, J., Hirose, M., Kozu, T., Kubota, T., Huffman, G.J., Alder, R., Bolvin, D.T., Gu, G., Nelkin, E.J.,
Liu, G., Shige, S., Kida, S., Seto, S., Takahashi, N., Bowman, K.P., Hong, Y., Stocker, E.F., Wolff, D.B., 2007.
Takayabu, Y.N., 2009. GSMaP passive microwave precip- The TRMM multisatellite precipitation analysis (TMPA):
itation retrieval algorithm: algorithm description and quasi-global, multiyear, combined-sensor precipitation
validation. J. Meteorol. Soc. Jpn. 87A, 119e136. estimates at fine scales. J. Hydrometeorol. 8, 38e55.
Arkin, P., Meisner, B.N., 1987. The relationship between Huffman, G.J., Bolvin, D.T., Braithwaite, D., Hsu, K.,
large-scale convective rainfall and cold cloud over the Joyce, R., Kidd, C., Nelkin, E.J., Xie, P., 2015. NASA
Western Hemisphere during 1982-84. Mon. Weather Global Precipitation Measurement Integrated Multi-
Rev. 115, 51e74. Satellite Retrievals for GPM (IMERG). Algorithm Theoret-
Ba, M.B., Gruber, A., 2001. GOES multispectral rainfall algo- ical Basis Doc., Version 4.5, 30 pp. Available online at:
rithm (GMSRA). J. Appl. Meteorol. 40, 1500e1514. http://pmm.nasa.gov/sites/default/files/document_
Bauer, P., Schanz, L., 1998. Outlook for combined TMI-VIRS files/IMERG_ATBD_V4.5.pdf.
algorithms for TRMM: lessons from the PIP and AIP
projects. J. Atmos. Sci. 55, 1714e1729.
646 16. Precipitation

Iguchi, T., Kozu, T., Meneghini, R., Awaka, J., Okamoto, K., Levizzani, V., Bauer, P., Turk, F.J., 2007. Measuring Precipita-
2000. Rain-profiling algorithm for the TRMM precipita- tion from Space: EURAINSAT and the Future. Springer,
tion radar. J. Appl. Meteorol. 39, 2038e2052. Dordrecht.
Iguchi, T., 2007. Space-borne radar algorithms. In: Li, Z., Yang, D., Hong, Y., 2013. Multi-scale evaluation of
Levizzani, V., Bauer, P., Turk, F.J. (Eds.), Measuring Pre- high-resolution multi-sensor blended global precipitation
cipitation from Space: EURAINSAT and the Future. products over the Yangtze river. J. Hydrol. 500 (14),
Springer-Verlag, Berlin. 157e169.
Iguchi, T., 2009. Uncertainties in the rain profiling algorithm Linacre, E., 1992. Climate Data and Resources: A Reference
for the TRMM precipitation radar. J. Meteorol. Soc. Jpn. and Guide. Routledge.
87A, 1e30. Liu, G., 2003. Satellite remote sensing: precipitation. In:
Janowiak, J., Joyce, R.J., Yahosh, Y., 2001. A real-time global Holton, J.R., Curry, J.A., Pyle, J.A. (Eds.), Encyclopedia
half-hourly pixel-resolution IR dataset and its of Atmospheric Sciences. Academic Press, London,
applications. Bull. Am. Meteorol. Soc. 82, 205e217. pp. 1972e1979.
Joyce, R.J., Janowiak, J.E., Arkin, P.A., Xie, P., 2004. Maggioni, V., Meyers, P.C., Robinson, M.D., 2016. A review
CMORPH: a method that produces global precipitation of merged high-resolution satellite precipitation product
estimates from passive microwave and infrared data at accuracy during the tropical rainfall measuring mission
high spatial and temporal resolution. J. Hydrometeorol. (TRMM) Era. J. Hydrometeorol. 17, 1101e1117.
5, 487e503. Makihara, Y., Uekiyo, N., Tabata, A., Abe, Y., 1996. Accuracy
Kidd, C., 2001. Satellite rainfall climatology: a review. Int. J. of radar-AMeDAS precipitation. IEICE Trans. Commun.
Climatol. 21, 1041e1066. E79-B, 751e762.
Kidd, C., Bauer, P., Turk, J., Huffman, G.J., Joyce, R., Michaelides, S., Levizzani, V., Anagnostou, E., Bauer, P.,
Hsu, K.L., Braithwaite, D., 2012. Intercomparison of Kasparis, T., Lane, J.E., 2009. Precipitation: measurement,
high-resolution precipitation products over northwest remote sensing, climatology and modeling. Atmos. Res.
europe. J. Hydrometeorol. 13 (1), 67e83. 94, 512e533.
Krajewski, W., 1987. Cokriging radar-rainfall and rain gage Morin, E., Krajewski, W.F., Goodrich, D.C., Gao, X.,
data. J. Geophys. Res. 92, 9571e9580. Sorooshian, S., 2003. Estimating rainfall intensities from
Kubota, T., Shige, S., Hashizume, H., Aonashi, K., weather radar data: the scale-dependency problem.
Takahashi, N., Seto, S., Hirose, M., Takayabu, Y.N., J. Hydrometeorol. 4, 782e797.
Ushio, T., Nakagawa, K., Iwanami, K., Kachi, M., Mugnai, A., Smith, E.A., Triopli, G.J., 1993. Foundations for
Okamoto, K., 2007. Global precipitation map using statistical-physical precipitation retrieval from passive
satellite-borne microwave radiometers by the GSMaP microwave satellite measurements. Part II: emission-
project: production and validation. IEEE Trans. Geosci. source and generalized weighting-function properties of
Remote Sens. 45, 2259e2275. a time-dependent cloud-radiation model. J. Appl. Mete-
Kubota, T., Ushio, T., Shige, S., Kida, S., Kachi, M., orol. 32, 17e39.
Okamoto, K., 2009. Verification of high-resolution Neary, V., Habib, E., Fleming, M., 2004. Hydrologic modeling
satellite-based rainfall estimates around Japan using a with NEXRAD precipitation in middle Tennessee.
gauge-calibrated ground-radar dataset. J. Meteorol. Soc. J. Hydrol. Eng. 9, 339e349.
Jpn. 87A, 203e222. Nguyen, P., Ombadi, M., Sorooshian, S., Hsu, K.,
Kummerow, C., Barnes, W., 1998. The tropical rainfall AghaKouchak, A., Braithwaite, D., Ashouri, H.,
measuring mission (TRMM) sensor package. J. Atmos. Thorstensen, A.R., 2018. The PERSIANN family of global
Ocean. Technol. 15, 809e817. satellite precipitation data: a review and evaluation of
Kummerow, C.D., Hong, Y., Olson, W.S., Yang, S., products. Hydrol. Earth Syst. Sci. 22, 5801e5816.
Adler, R.F., McCollum, J., Ferraro, R., Petty, G., Okamoto, K., Iguchi, T., Takahashi, N., Iwanami, K.,
Shi, D.B., Wilheit, T.T., 2001. The evolution of the God- Ushio, T., 2005. The global satellite mapping of precipita-
dard Profiling Algorithm (GPROF) for rainfall estimation tion (GSMaP) project. In: 25th IGARSS Proceedings,
from passive microwave sensors. J. Appl. Meteorol. 40, pp. 3414e3416.
1801e1820. Prigent, C., 2010. Precipitation retrieval from space: an
Kummerow, C.D., Masunaga, & H., Bauer, P., 2007. A next- overview. Compt. Rendus Geosci. 342, 380e389. Copy-
generation microwave rainfall retrieval algorithm for right © 2010 Elsevier Masson SAS. All rights reserved.
use by TRMM and GPM. In: Levizzani, V., Bauer, P., Rodda, J.C., 1969. Hydrological network design e needs,
Turk, F.J. (Eds.), Measuring Precipitation from Space: problems and approaches. WMO Rep. 12, 1e58.
EURAINSAT and the Future. Springer-Verlag, Berlin.
References 647
Sapiano, M.R.P., Arkin, P.A., 2009. An intercomparison and Takahashi, N., Iguchi, T., Aonashi, K., Awaka, J., Eito, H.,
validation of high-resolution satellite precipitation esti- Fujita, M., Hashizume, M., Hirose, M., Inoue, T.,
mates with 3-hourly gauge data. J. Hydrometeorol. 10, Kashiwagi, H., Kozu, T., Satoh, S., Seto, S., Shige, S.,
149e166. Shimizu, S., Takayabu, Y., Okamoto, K., 2004. The global
Savage, R.C., Weinman, J.A., 1975. Preliminary calculations satellite mapping of precipitation (GSMaP) project: Part II
of the upwelling radiance from rain clouds at 37.0 and algorithm and precipitation model development. In: 2nd
19.35 GHz. Bull. Am. Meteorol. Soc. 56, 1272e1274. TRMM International Conference. September 6-10, 2004.
Schwaller, M.R., Morris, K.R., 2011. A ground validation Tokyo, Japan.
network for the global precipitation measurement Turk, F.J., Arkin, P., Ebert, E.E., Sapiano, M.R.P., 2008. Eval-
mission. J. Atmos. Ocean. Technol. 28 (3), 301e319. uating high-resolution precipitation products. Bull. Am.
Shelton, M.L., 2009. Hydroclimatology. Cambridge Univer- Meteorol. Soc. 89, 1911e1916.
sity Press, Cambridge. Ushio, T., Sasashige, K., Kubata, T., Shige, S., Okamoto, K.,
Smith, E., Asrar, G., Furuhama, Y., Ginati, A., Aonashi, K., Inoue, T., Takahashi, N., Iguchi, T.,
Kummerow, C., Levizzani, V., Mugnai, A., Kachi, M., Oki, R., Morimoto, T., Kawasaki, Z., 2009.
Nakamura, K., Adler, R., Casse, V., Cleave, M., A Kalman filter approach to the Global Satellite Mapping
Debois, M., Durning, J., Entin, J., Houser, P., Iguchi, T., of Precipitation (GSMaP) from combined passive micro-
Kakar, R., Kaye, J., Kojima, M., Lettenmaier, D.P., wave and infrared radiometric data. J. Meteorol. Soc.
Luther, M., Mehta, A., Morel, P., Nakazawa, T., Jpn. 87A, 137e151.
Neeck, S., Okamoto, K., Oki, R., Raju, G., Shepherd, M., Villarini, G., Krajewski, W.F., 2010. Review of the different
Stocker, E., Testud, J., Wood, E.F., 2007. International sources of uncertainty in single polarization radar-based
global precipitation measurement (gpm) program and estimates of rainfall. Surv. Geophys. 31, 107e129.
mission: an overview. In: Levizzani, V., Bauer, P., Wilheit, T.T., Change, A.T.C., Rao, M.S.V., Rodgers, E.B.,
Turk, F.J. (Eds.), Measuring Precipitation From Space, Ad- Theon, J.S., 1977. A satellite technique for quantitatively
vances In Global Change Research, vol. 28. Springer, mapping rainfall rates over oceans. J. Appl. Meteorol.
Dordrecht. 16, 551e560.
Smith, E.A., Mugnai, A., Cooper, H.J., Triopli, G.J., Xiang, X., Wilheit, T., Kummerow, C., Ferraro, R., 1999. EOS/AMSR
1992. Foundations for statistical-physical precipitation rainfall. Algorithm Theor. Basis Doc. 1e59.
retrieval from passive microwave satellite measurements. Xie, P., Arkin, P.A., 1997. Global precipitation: a 17 year
Part I: brightness-temperature properties of a time- monthly analysis based on gauge observations, satellite
dependent cloud-radiation model. J. Appl. Meteorol. 31, estimates, and predictions. J. Clim. 9, 840e858.
506e531. Xie, P., Janowiak, J.E., Arkin, P.A., Adler, R.F., Gruber, A.,
Stout, G., Mueller, E.A., 1968. Survey of relationships be- Ferraro, R.R., Huffman, G.J., Curtis, S., 2003. GPCP
tween rainfall rate and radar reflectivity in the measure- pentad precipitation analyses: an experimental dataset
ment of precipitation. J. Appl. Meteorol. 7, 465e474. based on gauge observations and satellite estimates.
Sun, Q.H., Miao, C.Y., Duan, Q.Y., Ashouri, H., J. Clim. 16, 2197e2214.
Sorooshian, S., Hsu, K.L., 2018. A review of global precip- Yuter, S.E., 2003. Radar: precipitation radar. In: Holton, J.R.,
itation data sets: data sources, estimation, and Curry, J.A., Pyle, J.A. (Eds.), Encyclopedia of Atmospheric
intercomparisons. Rev. Geophys. 56, 79e107. Sciences. Academic Press, London.

You might also like