JGR Atmospheres - 2017 - Cannon

PUBLICATIONS
Journal of Geophysical Research: Atmospheres

RESEARCH ARTICLE GPM Satellite Radar Measurements of Precipitation
10.1002/2017JD027355
and Freezing Level in Atmospheric Rivers:
Key Points:
• Precipitation in atmospheric rivers
Comparison With Ground-Based
accounts for nearly 10% more oceanic
precipitation in satellite radar
Radars and Reanalyses
estimates than reanalyses Forest Cannon1 , F. Martin Ralph1, Anna M. Wilson1 , and Dennis P. Lettenmaier2
• Over the eastern North Pacific Ocean,
25% of satellite radar-derived 1
Center for Western Weather and Water Extremes, Scripps Institution of Oceanography, University of California, San Diego,
precipitation rates exceed the 99th
percentile of reanalyses La Jolla, CA, USA, 2Department of Geography, University of California, Los Angeles, CA, USA
• Satellite radar freezing level height
estimates agree exceptionally well
with ground-based radar observations Abstract Atmospheric rivers (ARs) account for more than 90% of the total meridional water vapor flux in
and reanalyses midlatitudes, and 25–50% of the annual precipitation in the coastal western United States. In this study,
reflectivity profiles from the Global Precipitation Measurement Dual-Frequency Precipitation Radar
(GPM-DPR) are used to evaluate precipitation and temperature characteristics of ARs over the western coast
Correspondence to: of North America and the eastern North Pacific Ocean. Evaluation of GPM-DPR bright-band height using a
F. Cannon, network of ground-based vertically pointing radars along the West Coast demonstrated exceptional
fcannon@ucsd.edu
agreement, and comparison with freezing level height from reanalyses over the eastern North Pacific Ocean
also consistently agreed, indicating that GPM-DPR can be used to independently validate freezing level in
Citation: models. However, precipitation comparison with gridded observations across the western United States
Cannon, F., Ralph, F. M., Wilson, A. M., &
Lettenmaier, D. P. (2017). GPM satellite indicated deficiencies in GPM-DPR’s ability to reproduce the spatial distribution of winter precipitation, likely
radar measurements of precipitation related to sampling frequency. Over the geographically homogeneous oceanic portion of the domain,
and freezing level in atmospheric rivers: sampling frequency was not problematic, and significant differences in the frequency and intensity of
Comparison with ground-based radars
and reanalyses. Journal of Geophysical precipitation between GPM-DPR and reanalyses highlighted biases in both satellite-observed and modeled
Research: Atmospheres, 122, AR precipitation. Reanalyses precipitation rates below the minimum sensitivity of GPM-DPR accounted for a
12,747–12,764. https://doi.org/10.1002/ 20% increase in total precipitation, and 25% of radar-derived precipitation rates were greater than the 99th
2017JD027355
percentile precipitation rate in reanalyses. Due to differences in the proportions of precipitation in
Received 5 JUL 2017 convective, stratiform bright-band, and non-bright-band conditions, AR conditions contributed nearly 10%
Accepted 17 NOV 2017 more to total precipitation in GPM-DPR than reanalyses.
Accepted article online 27 NOV 2017
Published online 14 DEC 2017
1. Introduction
Atmospheric rivers (ARs) are a key component of the global water budget that account for a major fraction of
precipitation over midlatitude oceans and coastlines, including the U.S. West Coast (e.g., Dettinger, 2013;
Gimeno et al., 2014; Kingston et al., 2016; Lavers & Villarini, 2013; Ralph et al., 2017, 2004; Viale & Nuñez,
2011; Zhu & Newell, 1998). Despite the importance of ARs to global moisture fluxes and regional water
resources, understanding and forecasting these features remains a challenge due in part to their formation
and propagation over the ocean, where in situ and ground-based observations are extremely limited. One
of the more promising recent technological developments that could help fill this data gap is the launch
of the Global Precipitation Measurement Mission (GPM) satellite (Hou et al., 2014; Skofronick-Jackson et al.,
2016), which carries a Dual-Frequency Precipitation Radar (DPR) that helps monitor global precipitation
between 65°N and 65°S. This monitoring includes precipitation rate estimation, snow versus rain discrimina-
tion, freezing level height estimation, and discernment of convective versus stratiform precipitation types
(Iguchi et al., 2010).
However, the ability of GPM to fill gaps in AR measurements offshore, and over the western United States, has
yet to be evaluated. This study uses a unique set of observations and methodologies to evaluate the
Dual-Frequency Precipitation Radar (DPR) on board GPM, including a network of ground-based precipitation
profiling radars that regularly monitor snow-level aloft in precipitation (White et al., 2013) and a state-of-the-
art AR detection method (Rutz et al., 2014), completely independent of GPM. The paper also evaluates the
ability of GPM-DPR measurements to accurately represent the regional climatology of precipitation across
©2017. American Geophysical Union.
major portions of the U.S. West Coastal states, and over the Pacific Ocean to its west. And, finally, it compares
All Rights Reserved. the GPM-derived climatology of precipitation in the region in terms of rain rates to that provided by a
CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,747

21698996, 2017, 23, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JD027355 by Cochrane Thailand, Wiley Online Library on [31/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Atmospheres 10.1002/2017JD027355
modern reanalysis data set (Modern-Era Retrospective Analysis for Research Application, version 2, MERRA-2)
to search for potential differences. This study is motivated by the following issues:
1. Precipitation is a key term in the water vapor budget within ARs and yet is poorly observed over the
oceans (L’Ecuyer et al., 2015).
2. The altitude of the freezing level is a key factor in determining precipitation impacts on landfall, including
mountain snowpack formation/depletion, streamflow, and flooding (Neiman et al., 2014; White et al.,
2010), yet is poorly observed ahead of AR landfall.
3. Model predictions of AR landfall, precipitation amounts, and freezing level contain serious errors (Ralph
et al., 2010, 2016) that might be reduced by better initialization.
4. GPM offers an opportunity to help fill these gaps in AR monitoring, and yet quantitative evaluations of
GPM-DPR performance in measuring precipitation and freezing level in ARs are not available.
5. Although it is known that GPM-DPR is limited in its ability to measure light rain rates (Iguchi et al., 2010),
and that global models produce excessive light rain and drizzle (Stephens et al., 2010), comparisons of
these two key sources of precipitation information have not been conducted for AR environments.
In this study, GPM-DPR along-track reflectivity profiles were evaluated over the West Coast and eastern North
Pacific Ocean. GPM was launched in February 2014, following the success of the Tropical Rainfall
Measurement Mission (TRMM; Huffman et al., 2007). A key sensor on board TRMM was the Ku band precipita-
tion radar (PR), which measured three-dimensional reflectivity of precipitation droplets in the atmosphere.
GPM-DPR exceeds TRMM PR’s capabilities by utilizing differential attenuation between a similar Ku band
(13.6 GHz) sensor and an additional Ka band (35.5 GHz) radar to provide information on drop size distribu-
tions and precipitation intensity that improve precipitation rate and freezing level estimates. This informa-
tion additionally enables the discernment of precipitation type and microphysics (Hamada & Takaybu,
2016; Iguchi et al., 2010), which describe various cloud particle formation, growth, and precipitation pro-
cesses, depending on altitude and environmental conditions (e.g., Martner et al., 2008; Matrosov, 2007,
2013). Previously, Matrosov (2013) employed a W band radar (~94 GHz) on board the CloudSat satellite
(CloudSat CPR; Stephens et al. 2002) to investigate precipitation processes within ARs over the eastern
North Pacific Ocean over a 3 year period (2007–2009). While the methodology for analyzing ARs over the
ocean in the present study is similar to that of Matrosov (2013), it is important to note that CloudSat
CPR-based findings are likely to differ from GPM-DPR results based on differences in each radar’s
minimum sensitivity.
The present study provides novel analyses of GPM-DPR measurements in atmospheric rivers and discusses
the benefits and limitations of applying GPM-DPR to observing precipitation processes over the western
United States and eastern North Pacific Ocean. Despite notable caveats, it is shown that GPM-DPR improves
observation of AR precipitation and freezing level over the data sparse oceans and that these measurements
are useful for evaluating the simulation of ARs by global models. Differences in precipitation frequency and
intensity between GPM-DPR and reanalyses are highlighted by discrepancies of nearly 10% in the contribu-
tion of ARs to total precipitation over the eastern North Pacific Ocean, while comparisons of freezing level
demonstrate remarkable agreement in height and variability.
2. Data and Methodology

2.1. Satellite Precipitation Radar
This research utilized the GPM-DPR standard level 2 (L2) product, version 04 (2ADPR; Iguchi et al., 2010), from
the matched beam scan swath. Radar reflectivity in three dimensions, precipitation rate, bright-band height,
and precipitation type variables were evaluated (listed in the GPM 2ADPR file structure documentation as,
“zFactorCorrected,” “precipRateESurface,” “heightBB,” and “typePrecip,” respectively). The reflectivity data
used have a vertical bin sampling of 125 m (176 vertical bins) and a horizontal resolution of ~5.2 km. The
along-track horizontal footprint of the Ka and Ku band matched scan swath is ~125 km (25 horizontal bins
in the along-track direction). Only data from the five cross-track bins closest to the nadir-pointing ray at
the center of the swath were used on account of increasing surface clutter with increasing distance from
nadir (Kubota et al., 2016), and in order to avoid sampling precipitation features at an angle, which bias both
precipitation and bright-band retrievals (Hamada & Takaybu, 2016; Hirose et al., 2012). GPM completes
roughly 16 orbits per day between 65°N and 65°S, and the 2ADPR product has a latency of less than 24 h.

Figure 1. (a) Example GPM-DPR swaths during an AR on 8 February 2015. (b) A subset of the domain (red box) over Central California is shown in the center panel
with NEXRAD reflectivity over topography. The full swath is shown in black, with colored dots indicating measured precipitation rates at individual GPM-DPR
bins. White dots indicate vertically profiling radars. (c) The vertical profile of GPM-DPR reflectivity from the Ka band and topography along the center of the swath
over California. Black circles indicate the identified bright-band height and colored bars above the plot indicate stratiform (blue) and convective (red) precipitation
identification by GPM-DPR. (d) The FMCW S band radar measured radial velocity provided by NOAA ESRL Physical Sciences Division.
The study period spanned three winter seasons (October through March; 2014–2017), focusing on the period
of the year during which ARs routinely make landfall in California. The study domain covers the eastern North
Pacific Ocean and western North America (160°W–110°W; 20°N–60°N) (Figure 1a).
GPM satellite tracks recorded during an AR event impacting California on 8 February 2015 demonstrate the
challenges presented by the satellite’s sampling frequency (Figure 1a). Integrated water vapor (IWV) from
reanalysis (described in section 2.4) indicates the location of the AR feature of interest. In this example,
GPM made an overpass of landfalling AR conditions (elongated region of IWV > 20 mm) in Central
California at approximately 21:05 UTC. Only 5 of GPM’s 16 daily orbits passed through the study domain
on the day that the AR made landfall, 2 of which sampled some portion of the AR, and only 1 of which
sampled the event over the west coast. The limited track frequency and narrow swath width illustrate that
large areas of the domain are unsampled during any given event and that it is relatively rare for GPM-DPR
to sample an AR during landfall. However, a single AR propagating across the Pacific over several days (this
particular AR propagated for approximately 7 days) is likely to be sampled multiple times by the near-polar
orbiting satellite.
Along-track GPM-DPR precipitation estimates from the five angle bins that were used in this study are plotted
over composite reflectivity at 21:05 UTC from the National Weather Service’s Next-Generation Weather
Radars (NEXRAD) to illustrate the data sampling methodology (Figure 1b). Local precipitation maxima were
observed by both GPM-DPR and NEXRAD over the coastal ranges and foothills of the Sierra Nevada
Mountains, indicating strong spatial agreement between observations. An example reflectivity profile from
the center bin of the Ka band from Figure 1b is also plotted over topography (Figure 1c). Additional
GPM-DPR variables (e.g., precipitation rate and type, and bright-band height) that were evaluated in this
study were derived from reflectivity by the algorithms described in the GPM-DPR theoretical basis document
(Iguchi et al., 2010) and Awaka et al. (2016). Comparison of GPM-DPR precipitation characteristics with
ground-based observations are discussed in section 3. Figure 1d is included to demonstrate the use of a

Figure 2. All GPM-DPR overpasses, with individual bins color coded by measured precipitation rate, are shown for the (a) 2014–2015, (b) 2015–2016, and (c) 2016–
2017 winter seasons, and (d) the percentage of all winter overpasses that measured precipitation, within 0.25° grid boxes.
unique network of vertically profiling radars to validate GPM-DPR bright-band measurements (discussed in
section 2.3).
The location of all GPM-DPR precipitation retrievals within the study domain for each winter season is shown
in Figures 2a–2c. Swaths are vertically stacked and color coded according to precipitation rate. It is notable
that there were relatively few precipitation retrievals along the West Coast in all years, with the minimum
occurring during the exceptionally dry 2014–2015 winter season (California Department of Water
Resources, 2016). The percentage of overpasses that recorded precipitation over the entire study period
(Figure 2d) indicates that precipitation retrievals were most frequent along the coastal ranges of western
Canada, Washington, and Oregon (>25% of overpasses), and moderately frequent in the North Pacific above
30°N (10–25% of overpasses). GPM-DPR had the lowest precipitation sampling frequency over a large region
of semipermanent high pressure in the southeast portion of the domain, including Southern California. These
figures reflect the regional precipitation climatology (Adler et al., 2003) and are useful for understanding the
spatial variability of GPM-DPR sampling-related issues, which complicate comparisons of the satellite product
with in situ measurements and reanalyses.
2.2. In Situ Precipitation

GPM-DPR precipitation rates were compared with colocated precipitation data from the Parameter
Regression on Independent Slopes Method (PRISM; Daly et al., 1994), which is an interpolated precipitation
product with daily temporal and 4 km horizontal resolution. PRISM daily resolution data are not optimal
for comparison with GPM-DPR, given the instantaneous sampling of the satellite; however, the improved
spatial coverage afforded by interpolation made considerably more comparisons possible than with indivi-
dual stations. Rather than compare absolute precipitation amounts, bulk statistics of the spatial patterns of

normalized precipitation during the winter season were evaluated. Comparison with GPM-DPR along-track
precipitation retrievals was achieved by sampling the single PRISM grid point whose centroid was nearest
the center bin of the GPM swath on the date of overpass.
2.3. Ground-Based Snow-Level Radars
GPM-DPR bright-band height, based on a reflectivity peak in the Ku band, and the slope of differential reflec-
tivity between the Ka band and Ku band, within a 3 km vertical range centered 1 km below model-estimated
freezing level elevation (Iguchi et al., 2010; Le & Chandrasekar, 2013b), was validated using a unique network
of 27 ground-based vertically pointing radars across California, Oregon, and Washington (Neiman et al., 2014;
White et al., 2013). This network includes S band radars (3 GHz), Frequency-Modulated Continuous Wave
radars (FMCW), and wind profiling radars (449 MHz and 915 MHz) that were developed and are maintained
by the National Oceanic and Atmospheric Administration’s Physical Science Division (NOAA ESRL-PSD) (ftp://
ftp1.esrl.noaa.gov/psd2/data/realtime//). This unique observing network is described in White et al. (2013).
White et al. (2002) describe the methodology for automated bright-band retrievals from radars operating
in the 400–4000 MHz frequency range, which gives the scientific basis for the publicly available bright-band
height data in the NOAA PSD archive. Estimation is based on identification of a peak in signal-to-noise ratio
(equivalent to reflectivity) and increasing Doppler vertical velocity, indicating the increased fall speed of rain
compared to snow (White et al., 2002). Each ground-based radar considered has hourly to subhourly
temporal resolution overlapping some portion of the study period. An example time series of the vertical
structure of radial velocity and snow level measurements from the New Exechequer Dam FMCW radar is
shown for the 8 February 2015 AR event (Figure 1d). The FMCW radar measured a bright-band at
~2,775 m at the time of GPM’s overpass, while the DPR measured a bright-band at ~2,650 m in the bin
corresponding to the FMCW site. Although the melting layer typically has a width of several hundred meters,
due to the noninstantaneous melting of ice hydrometeors, the ground-based profiling radars identify bright-
band heights as the reflectivity maxima within this layer, typically near its center (White et al., 2002).
GPM-DPR’s bright-band height similarly refers to the peak of Ku band reflectivity and/or the peak of the
differential reflectivity slope within the melting layer (Le & Chandrasekar, 2013b).
2.4. Reanalysis
This research used NASA Modern-Era Retrospective Analysis for Research Application, version 2 (MERRA-2)
(Molod et al., 2015; Rienecker et al., 2011). MERRA-2, integrated at a resolution of 0.67° longitude by 0.5°
latitude with 72 Lagrangian vertical levels, was chosen on account of its refined representation of the atmo-
sphere’s hydrological cycle (Rienecker et al., 2011) and its high spatiotemporal resolution, which minimized
inconsistencies in comparisons with instantaneous GPM-DPR retrievals. Bias-corrected time-averaged
precipitation at 1 h temporal resolution, and instantaneous temperature and geopotential height, interpo-
lated to 42 pressure levels, at 3 h temporal resolution were evaluated. Freezing-level elevations were calcu-
lated for each 3 h snapshot by reverse-interpolating the temperature and geopotential height profiles at
each horizontal grid point and subsequently identifying the geopotential height at the 0°C isotherm within
that grid point (Harris et al., 2000). In the case of multiple 0°C levels in a single vertical profile, due to inver-
sions, only the lowest level was retained. Comparison with GPM along-track retrievals was achieved by
sampling the single MERRA-2 grid point whose centroid was nearest the center of the GPM swath, at the near-
est time slice in the 1-hourly precipitation and 3-hourly temperature data sets. MERRA’s improved simulation
of the hydrological cycle relies on assimilating precipitation estimates, including those from the GPM micro-
wave imager. However, the GPM-DPR measurements evaluated here are not assimilated and are independent.
3. GPM-DPR In Situ Validation

3.1. Precipitation
Before applying GPM-DPR to study AR precipitation characteristics over the ocean, the sensor was evaluated
over the western United States, where ARs contribute significantly to annual precipitation and ample in situ
data are available for validation. Although precipitation processes and instrument challenges (e.g., ground
clutter) are different between ocean and land portions of the study domain, this comparison highlights
important considerations regarding GPM-DPR’s performance in representing regional precipitation patterns
with a relatively short period of accumulated data. There have been several field campaigns and comprehen-
sive validation experiments, such as the Olympic Mountains Experiment (OLYMPEX; Houze et al., 2017), that

Figure 3. (a) Accumulated PRISM precipitation during the 2014–2017 winter seasons study period. (b) The spatially normalized sum of GPM-DPR instantaneous pre-
cipitation rates for all overpasses during the study period, within 0.25° grid boxes. Data were normalized by the domain maximum and minimum values to emphasize
spatial gradients.
have specifically focused on quantifying the performance of GPM-DPR instrumentation and algorithms in
measuring reflectivity and deriving related precipitation products. The comparisons presented here are
based on spatiotemporal sampling over multiple seasons and are intended to illustrate how low-sampling
frequency affects subsequent analyses, rather than as an evaluation of instrumentation or
algorithm performance.
Total accumulated PRISM precipitation during the study period demonstrates the characteristic pattern of
orographic enhancement over the western United States (Dettinger et al., 2004; Roe, 2005), indicated by
steep precipitation gradients across the Coast Ranges, Cascade Mountains, and Sierra Nevada Mountains
(Figure 3a). Additionally, there is a climatological north-south precipitation gradient (Daly et al., 1994) that
was particularly pronounced over the study period due to below average precipitation in California during
the 2014/2015 and 2015/2016 winter seasons (He et al., 2017). Figure 3b demonstrates the ability of
GPM-DPR to reproduce PRISM’s salient precipitation patterns. The plotted data show the normalized sum
of GPM-DPR precipitation measurements from all overpasses during the study period, binned to 0.25° resolu-
tion on account of sparsity (outside of the Pacific Northwest, there are no individual 0.25° grid points with
more than 10 GPM overpasses that recorded precipitation during the study period). Here normalization
focuses the comparison on spatial precipitation patterns rather than absolute values.
The pattern of orographic enhancement observed in PRISM (Figure 3a) is degraded in the GPM-DPR plot
(Figure 3b), primarily due to insufficient sampling (Figure 2). The spatial correlation between Figure 2b and
PRISM, resampled to 0.25° resolution (not shown), is 0.65. GPM-DPR’s results were best in western Oregon
and Washington, where frequent precipitation was observed. Given the small sample size for locations in
Southern and Central California, where precipitation events are less frequent, and the variability of instanta-
neous precipitation rates during events is large, GPM-DPR did not produce a statistically robust precipitation
climatology for this region. Increasing the number of cross-track bins used in the analyses from 5 to 11
improves the spatial correlation with PRISM to 0.7 and slightly improved the qualitative representation of
orographic precipitation gradients (not shown). Despite the moderate improvement gained through
increased cross-track sampling, this research employed only five cross-track bins to minimize the impact of
known issues in precipitation and bright-band height estimation in off-nadir bins (Hirose et al., 2012;
Kubota et al., 2016), which would unnecessarily introduce uncertainty into further analyses.
The hypothesis that differences were attributable to sampling deficiency and not measurement or algorithm
error is further supported by previous analyses from the OLYMPEX experiment, which demonstrated that
airborne instrumentation and algorithms similar to those developed for GPM-DPR are ideally suited to
measure orographic precipitation processes (Houze et al., 2017). Additionally, GPM-DPR’s predecessor,

Table 1
The Principal Characteristics of the Vertically Pointing Radars That Provide Bright-Band Height Measurements—Name, Coordinates, Altitude, Profiling Frequency,
Installation Date, and Project Name
ID Name Latitude Longitude Elevation (m) Instrument Install Project
1 Forks 47.97 124.40 95 449 MHz wind profiler 7/21/2015 DOE

2 Astoria 46.15 123.88 3 449 MHz wind profiler 3/9/2015 DOE
3 Troutdale 45.55 122.39 12 915 MHz wind profiler 8/26/2015 DOE
a a
3 Troutdale 45.55 122.39 12 3 GHz precipitation profiler 12/2/2015 DOE
4 Wasco 45.59 120.67 462 915 MHz wind profiler 9/2/2015 DOE
5 Boardman 45.82 119.81 107 915 MHz wind profiler 10/5/2015 DOE
6 Condon 45.24 120.17 891 915 MHz wind profiler 10/8/2015 DOE
7 Prineville 44.29 120.90 959 915 MHz wind profiler 8/5/2015 DOE
8 North Bend 43.42 124.24 5 449 MHz wind profiler 10/15/2015 DOE
9 Happy Camp 41.79 123.38 366 FMCW precipitation profiler 2/2/2012 HTW
10 Shasta Dam 40.72 122.43 202 FMCW precipitation profiler 12/9/2009 HTW
11 Chico 39.70 121.91 42 915 MHz wind profiler 6/2/2000 HTW
12 Oroville 39.53 121.49 114 FMCW precipitation profiler 12/8/2011 HTW
13 Colfax 39.08 120.94 644 FMCW precipitation profiler 12/10/2008 HTW
14 New Exchequer 37.60 120.27 259 FMCW precipitation profiler 12/3/2010 HTW
15 Pine Flat Dam 36.83 119.33 184 FMCW precipitation profiler 12/2/2010 HTW
16 Visalia 36.31 119.39 81 915 MHz wind profiler 3/17/2010 HTW
17 Kernville 35.75 118.42 816 FMCW precipitation profiler 1/2/2012 HTW
18 San Bernardino 34.20 117.34 602 FMCW precipitation profiler 3/12/2013 HTW
19 Santa Barbara 34.43 119.85 2 449 MHz wind profiler 1/20/2016 HTW
20 San Luis Res. 37.06 121.07 81 FMCW precipitation profiler 4/2/2013 HTW
21 Saint Helena 38.55 122.49 135 FMCW precipitation profiler 7/9/2014 HTW
22 Middletown 38.75 122.71 972 3 GHz precipitation profiler 12/10/2014 HTW
23 Santa Rosa 38.51 122.80 32 3 GHz precipitation profiler 10/7/2014 SCWA
24 Bodega Bay 38.32 123.07 15 449 MHz wind profiler 3/22/2013 HTW
25 Cazadero 38.61 123.22 478 3 GHz precipitation profiler 11/11/2004 HTW
26 Hopland 39.00 123.09 253 3 GHz precipitation profiler 10/22/2015 HTW
27 McKinleyville 40.97 124.11 56 449 MHz wind profiler 11/18/2015 HTW
Note. The ID numbers correspond to the map in Figure 4a.
a
DOE = Department of Energy; HTW = NOAA Hydrometeorological test bed—West; CWA = Sonoma County water agency.
Figure 4. (a) Locations of vertically pointing in situ radars (described in Table 1 with corresponding index numbers). Comparison of in situ radar bright-band mea-
surements and (b) concomitant colocated GPM-DPR bright-band height measurements within 100 km and 1 h. Measurements when the daily bright-band standard
deviation at the in situ radar was <200 m are colored according to distance and time lag. Bright-band heights units are meters above sea level.

Table 2
Comparisons of Measured Bright-Band Height at Each In Situ Radar With Concomitant, Colocated GPM-DPR Bright-Band Height Measurements
Radius < 100 km Radius < 50 km
Radius < 100 km time < 1 h time < 0.5 h
time < 1 h standard deviation < 200 m standard deviation < 200 m
ID Name ε (m) n σ (m) ε (m) n σ (m) ε (m) n σ (m)
1 Forks 299.2 15 436.7 171.2 6 157.5 177.5 5 162.3

2 Astoria 304.0 19 498.0 96.6 8 97.4 31.8 3 31.7
3 Troutdale 511.9 14 715.9 144.3 5 91.7 40.8 2 24.3
3* Troutdale* 465.2 16 674.5 139.4 7 109.3 77.7 3 55.9
4 Wasco 350.5 6 718.1 69.1 3 72.8 63.1 2 87.0
5 Boardman 88.4 5 68.7 88.4 5 68.7 124.9 2 73.2
6 Condon 245.5 3 203.8 245.5 3 203.8 202.3 3 268.8
7 Prineville 328.1 2 389.0 53.0 1 NA NA 0 NA
8 North Bend 383.4 18 423.8 267.5 9 290.5 171.9 4 105.1
9 Happy Camp 393.0 21 593.0 299.3 6 151.4 103.6 1 NA
10 Shasta Dam 91.0 6 82.7 243.5 2 160.4 160.4 2 117.4
11 Chico 222.2 4 139.5 147.4 1 NA NA 0 NA
12 Oroville 192.8 8 231.1 190.1 3 214.9 31.9 1 NA
13 Colfax 166.2 5 123.2 54.6 2 5.3 2.0 1 NA
14 New Exchequer 436.4 11 743.6 136.1 6 61.9 112.0 5 44.2
15 Pine Flat Dam 61.4 6 44.1 68.9 5 44.9 76.8 3 64.0
16 Visalia 189.9 4 117.9 189.9 4 117.9 149.8 2 71.2
17 Kernville 248.7 3 90.2 254.1 1 NA NA 0 NA
18 San Bernardino 117.7 3 80.1 163.3 2 19.8 NA 0 NA
19 Santa Barbara NA 0 NA NA 0 NA NA 0 NA
20 San Luis Res. 652.0 11 652.0 408.5 4 554.8 NA 0 NA
21 Saint Helena 270.7 1 NA NA 0 NA NA 0 NA
22 Middletown 396.0 8 715.3 157.0 4 77.8 190.0 2 106.5
23 Santa Rosa 216.7 10 317.5 72.3 4 23.6 73.8 2 24.1
24 Bodega Bay 240.9 6 472.1 73.0 2 82.3 43.5 2 0.9
25 Cazadero 304.1 10 511.7 44.3 4 45.5 82.4 2 40.6
26 Hopland 332.4 8 717.9 65.3 2 53.8 NA 0 NA
27 McKinleyville 166.6 12 135.5 100.8 6 71.3 76.2 2 54.6
284.2 235 380.6 151.7 105 118.9 99.6 49 78.3
Note. Mean absolute error (ε), the number of records (n), and the standard deviation (σ) are given. The data correspond to Figure 4b. NA = not applicable. Bold
values indicate column-averaged error, column total records, and column-averaged standard deviation. The asterisk indicates a the 915 MHz profiler at
Troutdale, which also had a 449 MHz profiler (no asterisk).
TRMM-PR, proved capable of reproducing orographic precipitation gradients in various mountain regions
over its lengthy temporal record (e.g., Anders et al., 2006). Given GPM-DPR’s improved ability to measure
solid-state precipitation, which is prevalent in the Sierra and Cascade Mountains during winter events
(Serreze et al., 2001), it is likely that additional data collection in the future will augment the representation
of regional precipitation patterns. While this section emphasizes the necessity for caution in interpreting
GPM-DPR observations over regions of infrequent precipitation or topographic complexity, it should be
noted that the manuscript’s conclusions were not sensitive to the number of cross-track bins used.
3.2. Bright-Band Height
The layer at which precipitation transitions from rain to snow can be identified by an increase in radar reflec-
tivity due to differences in the dielectric constant of ice and liquid water and the aggregation of ice particles
as they descend and melt (Austin & Bemis, 1950; Hooper & Kippax, 1950). Identification of a radar bright band
is important for discerning precipitation processes, which further influence reflectivity-derived precipitation
rate estimates on account of differing drop-size distributions for bright-band and non-bright-band precipita-
tion (Martner et al., 2008). Warm-type stratiform precipitation over Coastal California, which is driven by
coalescence in shallow precipitating clouds, typically does not exhibit a radar bright band and is defined
by smaller drop-size distributions, more numerous hydrometeors, and lighter precipitation rates than
cold-type stratiform precipitation, which does typically exhibit a radar bright band and arises from ice
processes in deep cloud layers associated with synoptic forcing (Martner et al., 2008). These differences
may also lead to significant variability in accumulated precipitation between ARs, irrespective of moisture

Figure 5. (a) A GPM-DPR swath through AR conditions on 4 February 2015 at 0:00 UTC and (b) the vertical profile of reflectivity from the Ka band along the center of
the GPM-DPR swath subset within the red box in Figure 5a.
flux conditions (Martner et al., 2008; White et al., 2002). Given the importance of these processes, GPM-DPR
bright-band measurements were compared with a network of 27 vertically pointing ground-based radars
(described in section 2). The principal characteristics of the radars—coordinates, altitude, profiling
frequency, installation data, and project name—are detailed in Table 1, and their locations, along with a
reference map of the study region, are shown in Figure 4a.
GPM-DPR recorded a bright band within 100 km of one of the network’s radars 241 times during the
2014–2017 study period. No single location accounted for more than 21 observations, though sites in
the northern and coastal extents of the domain accounted for proportionally more observations due to
the increased frequency of precipitation. A bright band was also observed at the colocated ground-based
radar within 1 h of the satellite measurement in 235 of the 241 instances, indicating that false positive
identifications of a bright band by GPM-DPR were uncommon. Additionally, GPM-DPR observations of preci-
pitation exceeding 1 mm h 1 failed to identify an existing bright band only 16 times. GPM-DPR bright-band
heights were compared with concomitant ground observations in Figure 4b. The different colored points in
the scatterplot indicate increasingly stringent spatiotemporal thresholds that are meant to ensure that each
sensor’s measurements were based on samples of the same features. Mean differences between GPM-DPR
bright-band height and individual ground-based radar sites, the number of observations, and the standard
deviation of the mean differences are detailed in Table 2.
The mean absolute error of GPM-DPR bright-band heights relative to the ground network was 284 m for the
235 comparable instances, and the correlation was 0.59 (p < 0.05). The discrepancy between these is primar-
ily attributable to rapid changes in bright-band height in the ground-based radars, which frequently occur

during frontal passages, and can exhibit subhourly decreases in

bright-band height of more than 1,000 m. Challengingly, precipitation
also frequently occurs during frontal passages. Excluding instances in
which the standard deviation of the ground-based radar’s bright-
band height for the date of the overpass exceeded 200 m (an empiri-
cally defined value intended to identify conditions that may lead to a
comparison of measurements from different air masses) not only
reduced the number of comparable overpasses to 105 but also
reduced the mean error to 152 m and improved the correlation to
0.92. Further constraining the data to only compare measurements
within a 50 km radius and 0.5 h of the radar observation decreased
the sample size to 49, decreased the mean error to 100 m, and
increased the correlation to 0.97. This error is smaller than the effec-
tive vertical resolution of the GPM-DPR data used in the study
(125 m). GPM-DPR appears to be capable of identifying bright-band
features and accurately estimating their elevation. Although the
dense network of ground-based snow-level radars is unique to the
West Coast, Matrosov et al. (2017) demonstrated significant agree-
ment between these data and NEXRAD WSR-88D bright-band height
estimates, indicating that a similar evaluation of GPM-DPR could be
performed across much of the U.S. using this large network of
operational radars.
4. GPM-DPR Over the Eastern North Pacific Ocean

4.1. GPM-DPR Oceanic Precipitation
An example GPM crossing of a landfalling AR over the eastern North
Pacific Ocean on 4 February 2015, and the corresponding along-track
Figure 6. (a) Accumulated GPM-DPR precipitation during winter seasons 2014– reflectivity retrieval, is demonstrated in Figure 5. This is the same AR
2017 and (b) mean precipitation rate, binned at 0.25° latitude/longitude feature that is depicted in Figure 1, 4 days prior. The integrated water
resolution. vapor (IWV) field, which identifies the position of the AR, was taken
from MERRA-2 data for the corresponding time slice. In this particular
case, precipitation was entirely confined to the plume of IWV greater than 20 mm, corresponding to AR
conditions (Neiman et al., 2008; Ralph et al., 2004). The reflectivity profile for a 4° subset of the GPM swath,
between 40.55 and 44.50°N, demonstrates an example bright-band retrieval, which characteristically
decreased with increasing latitude (Matrosov, 2013) (Figure 5b). This particular swath demonstrates good
agreement between the GPM-DPR bright-band and MERRA-2 freezing level, with the freezing level generally
several hundred meters higher. The maximum precipitation rate of 9.74 mm h 1 (not shown) in this swath
subset was found at 41.25°N, colocated with a maximum reflectivity of 35 dBZ and bright-band height of
2,618 m. The MERRA-2 time-averaged hourly precipitation amount for the corresponding grid point was
3.24 mm, and the MERRA-2 3-hourly instantaneous IWV was 31.4 mm. The mean bright-band height of this
swath subset was 1956 m, the mean MERRA-2 freezing elevation was 2202 m, and the mean absolute differ-
ence was 282 m. The analyses presented in the remainder of this manuscript are based on the aggregation of
precipitation and bright-band height statistics, similar to those presented in this example case, for all
GPM-DPR retrievals within the study period.
GPM-DPR total measured winter precipitation and average precipitation rate (Figure 6) and the proportion of
precipitation contributed by ARs (not shown) exhibited latitudinal dependence over the eastern North Pacific
Ocean during the study period. The highest precipitation totals were generally found between 40° and 50°N,
in a regime of frequent, moderate-intensity winter precipitation relative to the rest of the domain. Stratiform
precipitation, as defined by the GPM-DPR classification method based on the slope of the ratio of Ku band
and Ka band reflectivity, the intensity of reflectivity in the absence of a bright band, and the spatial distribu-
tion of reflectivity peaks (Awaka et al., 2016; Le & Chandrasekar, 2013a), accounted for 80.6% of total precipi-
tation. Convection, which is also identified by GPM-DPR’s precipitation classification module (Le &
Chandrasekar, 2013a), accounted for the remainder. Stratiform was the dominant precipitation type in all

regions except the southwestern portion of the domain, near Hawaii,

which exhibited infrequent convection during non-AR periods (not
shown). Bright-band precipitation accounted for 63% of total strati-
form precipitation and nonbright band accounted for 37%. The mean
precipitation rates for each category were 0.74 mm h 1 for stratiform
nonbright band, 1.86 mm h 1 for stratiform bright band, and
2.47 mm h 1 for convective.
A catalog of AR events, based on the tracking algorithm defined by
Rutz et al. (2014), was applied to investigate whether AR conditions
affected precipitation amount or precipitation processes within the
study domain. The AR catalog extends only until the end of 2015.
During this period, 83% of stratiform bright-band precipitation
occurred in AR conditions, compared to only 46% of non-bright-band
stratiform precipitation, a statistically significant difference. Overall,
75% of GPM-identified precipitation within AR features was stratiform
bright band. While individual AR features often exhibited both preci-
pitation types, similar to findings from Matrosov (2013) that CloudSat-
based study determined that the proportions of non-bright-band and
bright-band precipitation were nearly equal within ARs. The discre-
pancy between studies likely arises due differences in each sensor’s
sensitivity range. CloudSat has an approximate precipitation thresh-
old range between 0.02 and 5 mm h 1 (Stephens et al., 2010). By com-
parison, 5 mm h 1 is only the 65th percentile of measured GPM-DPR
precipitation rates in this study. Since bright-band stratiform precipi-
tation exhibited higher mean rain rates (Matrosov, 2013), it is possible
that GPM-DPR is biased toward measuring bright-band precipitation
and that Cloudsat is biased toward measuring non-bright-band preci-
pitation. The proportions of each rain type were also temperature
dependent (Matrosov, 2013; Ralph et al., 2004), indicating that lati-
tude, time in the season, interannual variability, and synoptic condi-
tions likely influenced these results.
4.2. GPM-DPR Oceanic Bright-Band Height

Example GPM-DPR bright-band height measurements within an AR in
the eastern North Pacific on 3 February 2015 were demonstrated
previously (Figure 5b). Here measured bright-band heights for all
instances in which GPM-DPR exhibited a bright band during the study
period were considered. Aggregated statistics from bright-band
height retrievals exhibit strong latitudinal dependence over the study
domain (Figure 7a). Mean bright-band height over oceanic portions of
the domain decreased more than 1,000 m (from approximately
Figure 7. (a) Mean GPM-DPR bright-band height measurements for all overpasses
3,000 m to 2,000 m) between 30° and 50°N. Mean bright-band height
that identified a bright band during the study period, binned at 0.25° latitude/
longitude resolution. (b) Bright-band height binned by 1° latitude and 2 mm also exhibited a statistically significant positive relationship with IWV
MERRA-2 IWV. (c) Total number of bright-band retrievals binned by 1° latitude and (Figure 7b), indicating that the latitude dependence of mean bright-
2 mm MERRA-2 IWV. band height is influenced by the effect of the latitudinal temperature
gradient on moisture content. Precipitation systems at lower latitudes
of the domain were characteristically warmer and contain more water
vapor, due to the Clausius Clapeyron relationship. Below 30°N, stratiform bright-band precipitation was
observed relatively infrequently and only for MERRA-2 IWV values in excess of 30 mm (Figure 7c).
Contrastingly, the most frequent stratiform bright-band precipitation retrievals occurred between 35° and
50°N and exhibited a steep gradient in mean bright-band height that depended upon moisture content.
The standard deviation of bright-band height measurements in this middle-latitude range was notably large
on account of the variability in temperature between individual AR events (not shown).

2
Figure 8. (a) GPM-DPR precipitation rate measurements for all overpasses within 1° latitude and 2 kg m MERRA-2 IWV bins. (b) Total number of GPM-DPR samples
with precipitation, (c) MERRA-2 precipitation rate, and (d) total number of GPM-DPR samples with colocated MERRA-2 precipitation. (e) Cumulative distribution
function of oceanic precipitation in the study domain, according to rain rate, and (f) an enlarged view of the cumulative distribution function for rain rates less than
1
6 mm h .
5. Comparison With Reanalysis

5.1. Comparison of GPM-DPR and MERRA-2 Precipitation
Comparisons of GPM-DPR precipitation with MERRA-2 equivalents yield insight about the sources of bias in
each product over the study domain. AR conditions accounted for 69% of GPM-DPR precipitation and 60%
of total MERRA-2 precipitation, inferring discrepancies in their frequency and intensity distributions.
Excluding MERRA-2 precipitation rates below 0.2 mm h 1, on account of GPM-DPR’s minimum sensitivity,
removed 21% of total MERRA-2 precipitation and increased the AR contribution to 65%. Subsequent compar-
isons between products used only MERRA-2 precipitation exceeding 0.2 mm h 1; however, discrepancies in
their distributions persisted. Scatterplots of the mean precipitation rate within 1° latitude × 2 mm precipitable
water bins indicate that more intense but less frequent GPM-DPR precipitation was recorded in conditions
with IWV > 20 mm (Figures 8a–8d).
A cumulative frequency distribution plot comparing oceanic precipitation between MERRA-2 and GPM-DPR
(Figures 8e and 8f) clearly demonstrates that low-intensity precipitation accounted for far more total precipi-
tation in MERRA-2 than GPM-DPR. In fact, 10% of all GPM-DPR data were observed at precipitation rates that

exceeded the highest value in MERRA-2, and 25% of GPM-DPR precipita-

tion fell at rain rates above MERRA-2’s 99th percentile. MERRA-2’s
comparative underestimation of precipitation rates in high precipitable
water conditions at low-to-middle latitudes likely results from several fac-
tors, including the following: the comparison of mean grid-scale precipita-
tion with instantaneous point observations (Wehner et al., 2010), well-
known light precipitation biases in coarse-resolution models
(Intergovernmental Panel on Climate Change, 2007), and GPM-DPR’s diffi-
culty measuring light precipitation (near or below its minimum sensitivity
of ~12 dBZ). Unlike Stephens et al. (2010), who also demonstrated discre-
pancies between Cloudsat and global model precipitation distributions,
the present study did not attempt to reconcile these differences through
statistical upscaling or downscaling of the independent precipitation
sources in order to achieve a direct comparison. Thus, differences shown
here are important considerations regarding uncertainty in each data
sources’ precipitation distribution but are not intended to quantify error.
Discriminating between convective and stratiform precipitation, as identi-
fied by GPM-DPR (Awaka et al., 2016), further substantiates the origin of
differences between products. MERRA-2 precipitation was evaluated
based on the GPM-DPR precipitation classification. Stratiform (convective)
precipitation accounted for 75% (25%) of GPM-DPR precipitation and colo-
cated MERRA-2 precipitation accounted for 93% (7%) of the MERRA-2
along-track total. It is important to note that only 50% of along-track
MERRA-2 precipitation was included in this analysis given that a GPM-
DPR precipitation measurement was necessary to define precipitation
type. The large difference in stratiform and convective rainfall between
precipitation products is explained by an overestimation of precipitation
by MERRA-2 during GPM-defined low-intensity non-bright-band stratiform
precipitation and an underestimation of precipitation during
moderate-to-high-intensity convective conditions. The lack of convective
precipitation is partially explained by the inability of coarse-resolution
models to explicitly resolve convection (Prein et al., 2015; Warner, 2011;
Warner & Hsu, 2000). Importantly, non-bright-band stratiform precipitation
was evenly distributed between AR and non-AR conditions, while AR
conditions produced >83% of bright-band stratiform and >68% of
convective precipitation in both GPM-DPR and MERRA-2. Thus, the
increased proportion of light-intensity MERRA-2 precipitation in
Figure 9. Scatterplot of along-track mean GPM-DPR bright-band height and
mean MERRA-2 freezing level height for each event during the study period non-bright-band stratiform conditions decreased the overall contribution
for (a) offshore and (b) onshore domains. The red line indicates the best fit of ARs to total simulated oceanic precipitation, relative to GPM-DPR.
linear relationship. The domains are shown in the red box within the map
inset.
5.2. Comparison of GPM-DPR Bright-Band Height and MERRA-2
Freezing Level
MERRA-2 freezing level height and GPM-DPR bright-band height were compared to verify each product’s
representation of temperature in precipitating systems. MERRA-2 freezing level is defined as the height of
the 0°C isotherm, which is typically several hundred meters above the observed bright band because the
melting of frozen hydrometeors falling through the 0°C isotherm is not instantaneous (Houze, 1997; Leary
& Houze, 1979; Minder & Kingsmill, 2013; Stewart et al., 1984). A case study comparison of GPM-DPR
bright-band height measurements and colocated MERRA-2 freezing level height was shown in Figure 5b
and discussed in section 4.2. Here similar analysis was applied for all GPM bright-band measurements near
the Central and Northern California coast (Figure 9a, inset). The mean along-track GPM-DPR bright-band
height and corresponding MERRA-2 freezing level height for all 83 valid winter overpasses had a mean
bright-band height of 2175 m and mean MERRA-2 freezing elevation of 2,512 m, although the event-to-event
variability was large. Despite a bright-band height range of more than 2,500 m and standard deviation of

Figure 10. (a) Number of overpasses with a GPM-DPR bright-band measurement for all 1° × 1° bins with at least three valid measurements. (b) Correlation between
the mean GPM-DPR bright-band height and concomitant mean MERRA-2 freezing level. (c) Average difference between GPM-DPR the mean bright-band heights and
concomitant mean MERRA-2 freezing level.

~600 m, the correlation with MERRA-2 freezing level was 0.93, indicating exceptional agreement in capturing
event-to-event temperature variability (Figure 9a). The mean absolute difference between bright-band
height and freezing level height was 356 m, with larger variance in events with higher bright-band heights.
Results of comparisons over the onshore domain were not appreciably different (Figure 9b).
Performing similar comparisons within 1° × 1° grid boxes covering the entire study domain, the mean corre-
lation between MERRA-2 freezing level height and all GPM-DPR overpasses was 0.93. High correlations were
relatively uniform across the domain (Figure 10b), confirming that GPM-DPR measurements were sensitive to
the primary drivers of freezing level height variability, including the seasonal cycle, latitude, and synoptic
weather events. The mean difference between bright-band height and colocated MERRA-2 freezing level
height for those overpasses was 288 m, which agrees with findings of an approximately 300 m difference
between airborne measurements of bright-band height and freezing level height within maritime stratiform
precipitation off the coast of California (Stewart et al., 1984; White et al., 2002). The difference between
MERRA-2 and GPM-DPR varies by latitude, with larger differences south of 40°N and for higher precipitable
water values (Figure 10c). Previous studies have also found that differences increase at lower latitude
(Houze, 1997; Harris et al., 2000; Shin et al., 2000). These results demonstrate that GPM-DPR captures
event-to-event freezing level variability with excellent skill and infer that its measurements could be used
to independently observe variability and trends in the temperature characteristics of ARs over the satellite’s
lifetime. GPM-DPR bright-band height will additionally benefit validation of freezing level initial conditions
and forecasts over the ocean and improve understanding of the offset between forecasted freezing level
and observed snow level in the western United States (White et al., 2010). These advancements have the
potential to significantly improve the skill of flood forecasting and hazard prediction associated with ARs
(Neiman et al., 2014).
6. Conclusions
Precipitation over midlatitude oceans is not well measured or simulated (Stephens et al., 2010) and is an
important source of uncertainty in the simulation of global water and energy fluxes in climate models
(L’Ecuyer et al., 2015; include (Li et al., 2008) reference). Decades of passive infrared and microwave satellite
observations over the eastern North Pacific Ocean have demonstrated that extratropical cyclones, and their
associated atmospheric river (AR) features, are the primary driver of regional precipitation (Ralph et al., 2004).
However, advancing understanding of precipitation processes in this region requires improved observations
(e.g., Matrosov, 2013; Stephens et al., 2010). In this research, measurements of AR precipitation rates, micro-
physical processes, and freezing level from the Dual-Frequency Precipitation Radar (DPR) on board the Global
Precipitation Measurement (GPM; Hou et al., 2014) satellite were evaluated relative to in situ observations and
reanalyses during winter seasons 2014–2017.
GPM-DPR did not effectively reproduce the observed spatial patterns of precipitation accumulation over the
western United States, due to sampling frequency and its minimum sensitivity. Accounting for these limita-
tions, distributions of GPM-DPR precipitation measurements were evaluated over the eastern North Pacific
Ocean and demonstrated that ARs accounted for nearly 70% of measured precipitation over the study
period. These observations identified moderate- to high-intensity precipitation over the eastern North
Pacific Ocean that was generally absent in global model simulation. MERRA-2 precipitation rates below the
minimum sensitivity of GPM-DPR accounted for a 20% increase in total precipitation, and 25% of radar-
derived precipitation rates were greater than the 99th percentile precipitation rate in reanalyses.
Comparison of GPM-DPR with reanalyses further illuminated discrepancies between the proportions of
non-bright-band stratiform, bright-band stratiform, and convective precipitation, which underscore the
propensity of reanalyses to overestimate light precipitation and underestimate heavy precipitation.
Notably, the higher proportion of low-intensity MERRA-2 precipitation in non-bright-band stratiform condi-
tions diminished the contribution of ARs to total simulated oceanic precipitation by 10%, relative to GPM-DPR.
GPM-DPR also demonstrated utility in providing an accurate and independent measurement of AR freezing
level height. Comparison of GPM-DPR-derived bright-band height with in situ radar in cold-type stratiform
conditions exhibited a mean absolute error that was smaller than the effective resolution of the satellite
product, and comparison with reanalysis freezing level height demonstrated exceptional agreement in
event-to-event variability over the entire study domain. These results additionally indicate that GPM-DPR is

potentially useful for independently investigating interannual precipitation and temperature variability in
oceanic precipitating systems.
Diagnosing inconsistencies between satellite measurements and reanalyses lends perspective to precipita-
tion and temperature biases in global models. This is necessary for interpreting model-based estimates of
future AR characteristics and impacts (e.g., Dettinger, 2011; Guan & Waliser, 2017; Ralph et al., 2016;
Warner et al., 2015), reducing uncertainty in numerical weather prediction forecasts (Okamoto et al., 2016;
Ralph et al., 2010) and constraining simulated global moisture and energy budgets (L’Ecuyer et al., 2015).
Although GPM-DPR exhibited several important limitations, chiefly related to sampling frequency and mini-
mum reflectivity sensitivity, it is clear that the advanced satellite product yields tremendous benefit for redu-
cing the data gap over the global oceans and supplementing current understanding of precipitation
processes in ARs.
Acknowledgments References
This research was supported by the
National Aeronautics and Space Adler, R. F., Huffman, G. J., Chang, A., Ferraro, R., Xie, P.-P., Janowiak, J., … Nelkin, E. (2003). The version-2 Global Precipitation Climatology
Administration (NASA) grant Project (GPCP) monthly precipitation analysis (1979–present). Journal of Hydrometeorology, 4(6), 1147–1167. https://doi.org/10.1175/
NNX14AD77G. Any opinions, findings, 1525-7541(2003)004%3C1147:TVGPCP%3E2.0.CO;2
and conclusions or recommendations Anders, A. M., Roe, G. H., Hallet, B., Montgomery, D. R., Finnegan, N. J., & Putkonen, J. (2006). Spatial patterns of precipitation and topography
expressed in this publication are those in the Himalaya. Geological Society of America, 398, 39–53.
of the authors and do not necessarily Austin, P. M., & Bemis, A. C. (1950). A quantitative study of the “bright band” in radar precipitation echoes. Journal of Meteorology, 7(2),
reflect the views of NASA. The MERRA-2 145–151. https://doi.org/10.1175/1520-0469(1950)007%3C0145:AQSOTB%3E2.0.CO;2
data used in this research were devel- Awaka, J., Le, M., Chandrasekar, V., Yoshida, N., Higashiuwatoko, T., Kubota, T., & Iguchi, T. (2016). Rain type classification algorithm module
oped by NASA’s Global Modeling and for GPM dual-frequency precipitation radar. Journal of Atmospheric and Oceanic Technology, 33(9), 1887–1898. https://doi.org/10.1175/
Assimilation Office (GMAO) and disse- JTECH-D-16-0016.1
minated by the Goddard Earth Sciences California Department of Water Resources (2016). Hydroclimate report water year 2015. Memorandum report.
Data and Information Services Center Daly, C., Neilson, R. P., & Phillips, D. L. (1994). A statistical-topographic model for mapping climatological precipitation over moun-
(GES DISC). Observational GPM data sets tainous terrain. Journal of Applied Meteorology, 33(2), 140–158. https://doi.org/10.1175/1520-0450(1994)033%3C0140:ASTMFM%
are made available through a joint col- 3E2.0.CO;2
laboration between JAXA and NASA, Dettinger, M. (2011). Climate change, atmospheric rivers, and floods in California—A multimodel analysis of storm frequency and magnitude
with data access also made available by changes. Journal of the American Water Resources Association, 47(3), 514–523. https://doi.org/10.1111/j.1752-1688.2011.00546.x
GES DISC. Radar profiler data were pro- Dettinger, M. (2013). Atmospheric rivers as drought busters on the U.S. West Coast. Journal of Hydrometeorology, 14(6), 1721–1732. https://
vided by the NOAA/OAR/ESRL PSD, doi.org/10.1175/JHM-D-13-02.1
Boulder, Colorado, USA, from their Dettinger, M., Redmond, K., & Cayan, D. (2004). Winter orographic precipitation ratios in the Sierra Nevada—Large-scale atmospheric cir-
website at www.esrl.noaa.gov/psd/. culations and hydrologic consequences. Journal of Hydrometeorology, 5(6), 1102–1116. https://doi.org/10.1175/JHM-390.1
PRISM data are provided by the PRISM Gimeno, L., Raquel, N., Vazquez, M., & Lavers, D. A. (2014). Atmospheric rivers: A mini-review. Frontiers in Earth Science, 2, 1–6.
Climate Group at Oregon State Guan, B., & Waliser, D. E. (2017). Atmospheric rivers in 20 year weather and climate simulations: A multi-model, global evaluation. Journal of
University and are available through Geophysical Research: Atmospheres, 122, 5556–5581. https://doi.org/10.1002/2016JD026174
www.prism.oregonstate.edu. Hamada, A., & Takaybu, Y. N. (2016). Improvements in detection of light precipitation with the Global Precipitation Measurement Dual-
Frequency Precipitation Radar (GPM-DPR). Journal of Atmospheric and Oceanic Technology, 33, 653–667.
Harris, G. N. Jr., Bowman, K. P., & Shin, D.-B. (2000). Comparison of freezing-level altitudes from the NCEP reanalysis with TRMM preci-
pitation radar bright band data. Journal of Climate, 13(23), 4137–4148. https://doi.org/10.1175/1520-0442(2000)013%3C4137:COFLAF%
3E2.0.CO;2
He, M., Russo, M., & Anderson, M. (2017). Hydroclimatic characteristics of the 2012–2015 California Drought from an operational perspective.
Climate, 5(1). https://doi.org/10.3390/cli5010005
Hirose, M., Shimizu, S., Oki, R., Iguchi, T., Short, D. A., & Nakamura, K. (2012). Incidence-angle dependency of TRMM PR rain estimates. Journal
of Atmospheric and Oceanic Technology, 29(2), 192–206. https://doi.org/10.1175/JTECH-D-11-00067.1
Hooper, J. E. N., & Kippax, A. A. (1950). The bright band, a phenomenon associated with radar echoes from falling rain. Quarterly Journal of the
Royal Meteorological Society, 76(328), 125–132. https://doi.org/10.1002/qj.49707632803
Hou, A. Y., Kakar, R. K., Neeck, S. A., Azarbarzin, A., Kummerow, C. D., Kojima, M., … Iguchi, T. (2014). The global precipitation measurement
mission. Bulletin of the American Meteorological Society, 95(5), 701–722. https://doi.org/10.1175/BAMS-D-13-00164.1
Houze, R. A. Jr. (1997). Stratiform precipitation in regions of convection: A meteorological paradox? Bulletin of the American Meteorological
Society, 78(10), 2179–2196. https://doi.org/10.1175/1520-0477(1997)078%3C2179:SPIROC%3E2.0.CO;2
Houze, R. A., Houze, R. A. Jr., McMurdie, L. A., Petersen, W. A., Schwaller, M. R., Baccus, W., … Barnes, H. C. (2017). The Olympic Mountains
Experiment (OLYMPEX). Bulletin of the American Meteorological Society, 98(10), 2167–2188. https://doi.org/10.1175/BAMS-D-16-0182.1
Huffman, G. J., Bolvin, D. T., Nelkin, E. J., Wolff, D. B., Adler, R. F., Gu, G., … Stocker, E. F. (2007). The TRMM multisatellite precipitation analysis
(TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. Journal of Hydrometeorology, 8(1), 38–55. https://
doi.org/10.1175/JHM560.1
Iguchi, T., Shinta, S., Menghini, R., Yoshida, N., Awaka, J., Le, M., … Kubota, T. (2010). GPM/DPR Level-2 algorithm theoretical basis document
(p. 68). Retrieved from https://pps.gsfc.nasa.gov/atbd.html
Intergovernmental Panel on Climate Change (2007). In S. Solomon, et al. (Eds.), Climate models and their evaluation. Climate change, 2007: The
physical science basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
(pp. 627–629). Cambridge, UK: Cambridge University Press.
Kingston, D. G., Lavers, D. A., & Hannah, D. M. (2016). Floods in the Southern Alps of New Zealand: The importance of atmospheric rivers.
Hydrological Processes, 30(26), 5063–5070. https://doi.org/10.1002/hyp.10982
Kubota, T., Iguchi, T., Kojima, M., Liao, L., Masaki, T., Hanado, H., … Oki, R. (2016). A statistical method for reducing sidelobe clutter for the Ku-
band precipitation radar on board the GPM Core Observatory. Journal of Atmospheric and Oceanic Technology, 33(7), 1413–1428. https://
doi.org/10.1175/JTECH-D-15-0202.1

L’Ecuyer, T. L., Beaudoing, H. K., Rodell, M., Olson, W., Lin, B., Kato, S., … Hilburn, K. (2015). The observed state of the energy budget in the
early twenty-first century. Journal of Climate, 28, 8320–8346.
Lavers, D. A., & Villarini, G. (2013). The nexus between atmospheric rivers and extreme precipitation across Europe. Geophysical Research
Letters, 40(12), 3259–3264. https://doi.org/10.1002/grl.50636
Le, M., & Chandrasekar, V. (2013a). Precipitation type classification method for Dual-Frequency Precipitation Radar (DPR) onboard the GPM.
Transactions on Geoscience and Remote Sensing, 51(3), 1784–1790. https://doi.org/10.1109/TGRS.2012.2205698
Le, M., & Chandrasekar, V. (2013b). Hydrometeor profile characterization method for dual-frequency precipitation radar onboard the GPM.
Transactions on Geoscience and Remote Sensing, 51, 3648–3658.
Leary, C. A., & Houze, R. A. Jr. (1979). The structure and evolution of convection in a tropical cloud cluster. Journal of the Atmospheric Sciences,
36(3), 437–457. https://doi.org/10.1175/1520-0469(1979)036%3C0437:TSAEOC%3E2.0.CO;2
Li, J.-L., Waliser, D., Woods, C., Teixeira, J., BAcmeister, J., Chern, J., … Köhler, M. (2008). Comparisons of satellites liquid water estimates to
ECMWF and GMAO analyses, 20th century IPCC AR4 climate simulations, and GCM simulations. Geophysical Research Letters, 35(19),
L19710. https://doi.org/10.1029/2008GL035427
Martner, E. B., Yuter, S. E., White, A. B., Matrosov, S. Y., Kingsmill, D. E., & Ralph, F. M. (2008). Raindrop size distributions and rain characteristics
in California coastal rainfall for periods without a radar bright band. Journal of Hydrometeorology, 9(3), 408–425. https://doi.org/10.1175/
2007JHM924.1
Matrosov, S. Y. (2007). Potential for attenuated-based estimations of rainfall rate from CloudSat. Geophysical Research Letters, 34(5), L05817.
https://doi.org/10.1029/2006GL029161
Matrosov, S. Y. (2013). Characteristics of landfalling atmospheric rivers inferred from satellite observations over the eastern North Pacific
Ocean. Monthly Weather Review, 141(11), 3757–3768. https://doi.org/10.1175/MWR-D-12-00324.1
Matrosov, S. Y., Cifelli, R., White, A., & Coleman, T. (2017). Snow-level estimates using operational polarimetric weather radar measurements.
Journal of Hydrometeorology, 18(4), 1009–1019. https://doi.org/10.1175/JHM-D-16-0238.1
Minder, J. R., & Kingsmill, D. E. (2013). Mesoscale variations of the atmospheric snow line over the northern Sierra Nevada: Multiyear statistics,
case study, and mechanisms. Journal of the Atmospheric Sciences, 70(3), 916–938. https://doi.org/10.1175/JAS-D-12-0194.1
Molod, A., Takacs, L., Suarez, M., & Bacmeister, J. (2015). Development of the GEOS-5 atmospheric general circulation model: Evolution from
MERRA-2 to MERRA2. Geoscientific Model Development, 8(5), 1339–1356. https://doi.org/10.5194/gmd-8-1339-2015
Neiman, P. J., Gottas, D. J., White, A. B., Schick, L. J., & Ralph, F. M. (2014). The use of snow-level observations derived from vertically profiling
radars to assess hydrometeorological characteristics and forecasts over Washington’s Green River basin. Journal of Hydrometeorology,
15(6), 2522–2541. https://doi.org/10.1175/JHM-D-14-0019.1
Neiman, P. J., Ralph, F. M., Wick, G. A., Lundquist, J., & Dettinger, M. D. (2008). Meteorological characteristics and overland precipitation
impacts of atmospheric rivers affecting the West Coast of North America based on eight years of SSM/I satellite observations. Journal of
Hydrometeorology, 9(1), 22–47. https://doi.org/10.1175/2007JHM855.1
Okamoto, K., Aonashi, K., Kubota, T., & Tashima, T. (2016). Experimental assimilation of the GPM Core Observatory DPR reflectivity profiles for
Typhoon Halong (2014). Monthly Weather Review, 144(6), 2307–2326. https://doi.org/10.1175/MWR-D-15-0399.1
Prein, A. F., Langhans, W., Fosser, G., Ferrone, A., Ban, N., Goergen, K., … Leung, R. (2015). A review on regional convection-permitting climate
modeling: Demonstrations, prospects, and challenges. Reviews of Geophysics, 53(2), 323–361. https://doi.org/10.1002/2014RG000475
Ralph, F., Dettinger, M., Lavers, D., Gorodetskaya, I., Martin, A., Viale, M., … Cordeira, J. (2017). Atmospheric rivers emerge as a global science
and applications focus. Bulletin of the American Meteorological Society, 98, 1969–1973. https://doi.org/10.1175/BAMS-D-16-0262.1
Ralph, F. M., Neiman, P. J., & Wick, G. A. (2004). Satellite and CALJET aircraft observations of atmospheric rivers over the eastern North Pacific
Ocean during the winter of 1997/98. Monthly Weather Review, 132(7), 1721–1745. https://doi.org/10.1175/1520-0493(2004)132%3C1721:
SACAOO%3E2.0.CO;2
Ralph, F. M., Prather, K. A., Cayan, D., Spackman, J. R., DeMott, P., Dettinger, M., … Intrieri, J. (2016). CalWater field studies designed to quantify
the roles of atmospheric rivers and aerosols in modulating U.S. West Coast precipitation in a changing climate. Bulletin of the American
Meteorological Society, 97, 1209–1228. https://doi.org/10.1175/BAMS-D-14-00043.1
Ralph, F. M., Sukovich, E., Reynolds, D., Dettinger, M., Weagle, S., Clark, W., & Neiman, P. J. (2010). Assessment of extreme quantitative pre-
cipitation forecasts and development of regional extreme event thresholds using data from HMT-2006 and COOP observers. Journal of
Hydrometeorology, 11, 1288–1306.
Rienecker, M. M., Suarez, M. J., Gelaro, R., Todling, R., Bacmeister, J., Liu, E., … Woollen, J. (2011). MERRA: NASA’s modern-era retrospective
analysis for research and applications. Journal of Climate, 24(14), 3624–3648. https://doi.org/10.1175/JCLI-D-11-00015.1
Roe, G. H. (2005). Orographic precipitation. Annual Review of Earth and Planetary Sciences, 33(1), 645–671. https://doi.org/10.1146/annurev.
earth.33.092203.122541
Rutz, J. J., Steenburgh, W. J., & Ralph, F. M. (2014). Climatological characteristics of atmospheric rivers and their inland penetration over the
western United States. Monthly Weather Review, 142(2), 905–921. https://doi.org/10.1175/MWR-D-13-00168.1
Serreze, M. C., Clark, M. P., & Frei, A. (2001). Characteristics of large snowfall events in the montane western United States as examined using
snowpack telemetry (SNOTEL) data. Water Resources Research, 37(3), 675–688. https://doi.org/10.1029/2000WR900307
Shin, D.-B., North, G. R., & Bowman, K. P. (2000). A summary of reflectivity profiles from the first year of TRMM radar data. Journal of Climate,
13(23), 4072–4086. https://doi.org/10.1175/1520-0442(2000)013%3C4072:ASORPF%3E2.0.CO;2
Skofronick-Jackson, G., Petersen, W. A., Berg, W., Kidd, C., Stocker, E. F., Kirschbaum, D. B., … Wilheit, T. (2016). The Global Precipitation
Measurement (GPM) mission for science and society. Bulletin of the American Meteorological Society, 98(8), 1679–1695. https://doi.org/
10.1175/BAMS-D-15-00306.1
Stephens, G. L., L’Ecuyer, T., Forbes, R., Gettlemen, A., Golaz, J.-C., Bodas-Salcedo, A., … Haynes, J. (2010). Dreary state of precipitation in
global models. Journal of Geophysical Research, 115, D24211. https://doi.org/10.1029/2010JD014532
Stephens, G. L., Vane, D. G., Boain, R. J., Mace, G. G., Sassen, K., Wang, Z., … the CloudSat Science Team (2002). The CloudSat mission and the
A-train. A new dimension of space-based observations of clouds and precipitation. Bulletin of the American Meteorological Society, 83(12),
1771–1790. https://doi.org/10.1175/BAMS-83-12-1771
Stewart, R., Marwitz, J., Pace, J., & Carbone, R. (1984). Characteristics through the melting layer of stratiform clouds. Journal of the Atmospheric
Sciences, 41(22), 3227–3237. https://doi.org/10.1175/1520-0469(1984)041%3C3227:CTTMLO%3E2.0.CO;2
Viale, M., & Nuñez, M. N. (2011). Climatology of Winter Orographic Precipitation over the Subtropical Central Andes and Associated Synoptic
and Regional Characteristics. Journal of Hydrometeorology, 12(4), 481–507.
Warner, M. D., Mass, C. F., & Salathe, E. P. Jr. (2015). Changes in winter atmospheric rivers along the North American West Coast in CMIP5
climate models. Journal of Hydrometeorology, 16(1), 118–128. https://doi.org/10.1175/JHM-D-14-0080.1
Warner, T. T. (2011). Numerical weather and climate prediction (pp. 129–139). New York: Cambridge University Press.

Warner, T. T., & Hsu, H. (2000). Nested-model simulation of moist convection: The impact of coarse-grid parameterized convection on fine-
grid resolved convection. Monthly Weather Review, 128(7), 2211–2231. https://doi.org/10.1175/1520-0493(2000)128%3C2211:NMSOMC%
3E2.0.CO;2
Wehner, F., Smith, R., Bala, G., & Duffy, P. (2010). The effect of horizontal resolution on simulation of very extreme US precipitation events in a
global atmosphere model. Climate Dynamics, 34(2-3), 241–247. https://doi.org/10.1007/s00382-009-0656-y
st
White, A. B., Anderson, M. L., Dettinger, M. D., Ralph, F. M., Hinojosa, A., Cayan, D. R., … Coleman, T. (2013). A 21 century California observing
network for monitoring extreme weather events. Journal of Atmospheric and Oceanic Technology, 30(8), 1585–1603. https://doi.org/
10.1175/JTECH-D-12-00217.1
White, A. B., Gottas, D. J., Henkel, A. F., Neiman, P. J., Ralph, F. M., & Gutman, S. I. (2010). Developing a performance measure for snow-level
forecasts. Journal of Hydrometeorology, 11(3), 739–753. https://doi.org/10.1175/2009JHM1181.1
White, A. B., Gottas, D. J., Strem, E. T., Ralph, F. M., & Neiman, P. J. (2002). Automated bright band height detection algorithm for use with
Doppler radar spectral moments. Journal of Atmospheric and Oceanic Technology, 19(5), 687–697. https://doi.org/10.1175/1520-
0426(2002)019%3C0687:AABHDA%3E2.0.CO;2
Zhu, Y., & Newell, R. (1998). A proposed algorithm for moisture fluxes from atmospheric rivers. Monthly Weather Review, 126(3), 725–735.
https://doi.org/10.1175/1520-0493(1998)126%3C0725:APAFMF%3E2.0.CO;2

JGR Atmospheres - 2017 - Cannon

Uploaded by

Copyright:

Available Formats

JGR Atmospheres - 2017 - Cannon

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

JGR Atmospheres - 2017 - Cannon

Uploaded by

Copyright:

Available Formats

PUBLICATIONS

Journal of Geophysical Research: Atmospheres

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,747

2. Data and Methodology

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,748

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,749

2.2. In Situ Precipitation

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,750

3. GPM-DPR In Situ Validation

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,751

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,752

1 Forks 47.97 124.40 95 449 MHz wind proﬁler 7/21/2015 DOE

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,753

ID Name ε (m) n σ (m) ε (m) n σ (m) ε (m) n σ (m)

1 Forks 299.2 15 436.7 171.2 6 157.5 177.5 5 162.3

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,754

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,755

during frontal passages, and can exhibit subhourly decreases in

4. GPM-DPR Over the Eastern North Paciﬁc Ocean

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,756

regions except the southwestern portion of the domain, near Hawaii,

4.2. GPM-DPR Oceanic Bright-Band Height

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,757

5. Comparison With Reanalysis

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,758

exceeded the highest value in MERRA-2, and 25% of GPM-DPR precipita-

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,759

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,760

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,761

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,762

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,763

CANNON ET AL. GPM-DPR EVALUATION IN ATMOSPHERIC RIVERS 12,764

You might also like