Understanding the Response of Tropical Cyclone Structure to the Assimilation of Synthetic Wind Profiles

Lisa R. Bucci aNOAA/Atlantic Oceanographic and Meteorological Laboratory/Hurricane Research Division, Miami, Florida

Search for other papers by Lisa R. Bucci in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0002-2706-3795
,
Sharanya J. Majumdar bUniversity of Miami, Miami, Florida

Search for other papers by Sharanya J. Majumdar in
Current site
Google Scholar
PubMed
Close
,
Robert Atlas cNSF Center for Accelerated Real Time Analytics, University of Maryland, Baltimore County, Baltimore, Maryland

Search for other papers by Robert Atlas in
Current site
Google Scholar
PubMed
Close
,
G. David Emmitt dSimpson Weather Associates, Charlottesville, Virginia

Search for other papers by G. David Emmitt in
Current site
Google Scholar
PubMed
Close
, and
Steve Greco dSimpson Weather Associates, Charlottesville, Virginia

Search for other papers by Steve Greco in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

This study examines how varying wind profile coverages in the tropical cyclone (TC) core, near environment, and broader synoptic environment affects the structure and evolution of a simulated Atlantic Ocean hurricane through data assimilation. Three sets of observing system simulation experiments are examined in this paper. The first experiment establishes a benchmark for the case study specific to the forecast system used by assimilating idealized profiles throughout the parent domain. The second presents how TC analyses and forecasts respond to varying the coverage of swaths produced by polar-orbiting satellites of idealized wind profiles. The final experiment assesses the role of TC inner-core observations by systematically removing them radially from the center. All observations are simulated from a high-resolution regional “nature run” of a hurricane and the tropical atmosphere, assimilating with an ensemble square root Kalman filter and using the Hurricane Weather and Research Forecast regional model. Results compare observation impact with the analyses, domainwide and TC-centric error statistics, and TC structural differences among the experiments. The study concludes that the most accurate TC representation is a result of the assimilation of collocated and uniform thermodynamic and kinematics observations. Intensity forecasts are improved with increased inner-core wind observations, even if the observations are only available once daily. Domainwide root-mean-square errors are significantly reduced when the TC is observed during a period of structural change, such as rapid intensification. The experiments suggest the importance of wind observations and the role of inner-core surveillance when analyzing and forecasting realistic TC structure.

Atlas: Retired.

© 2021 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Lisa Bucci, lisa.r.bucci@noaa.gov

Abstract

This study examines how varying wind profile coverages in the tropical cyclone (TC) core, near environment, and broader synoptic environment affects the structure and evolution of a simulated Atlantic Ocean hurricane through data assimilation. Three sets of observing system simulation experiments are examined in this paper. The first experiment establishes a benchmark for the case study specific to the forecast system used by assimilating idealized profiles throughout the parent domain. The second presents how TC analyses and forecasts respond to varying the coverage of swaths produced by polar-orbiting satellites of idealized wind profiles. The final experiment assesses the role of TC inner-core observations by systematically removing them radially from the center. All observations are simulated from a high-resolution regional “nature run” of a hurricane and the tropical atmosphere, assimilating with an ensemble square root Kalman filter and using the Hurricane Weather and Research Forecast regional model. Results compare observation impact with the analyses, domainwide and TC-centric error statistics, and TC structural differences among the experiments. The study concludes that the most accurate TC representation is a result of the assimilation of collocated and uniform thermodynamic and kinematics observations. Intensity forecasts are improved with increased inner-core wind observations, even if the observations are only available once daily. Domainwide root-mean-square errors are significantly reduced when the TC is observed during a period of structural change, such as rapid intensification. The experiments suggest the importance of wind observations and the role of inner-core surveillance when analyzing and forecasting realistic TC structure.

Atlas: Retired.

© 2021 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Lisa Bucci, lisa.r.bucci@noaa.gov

1. Introduction

The prediction of tropical cyclones (TCs) has evolved rapidly over the past few decades. Numerical forecasts have steadily improved corresponding to increased observation coverage (Weng and Zhang 2016), advanced data assimilation techniques (Poterjoy et al. 2014), smaller numerical model grid spacing with larger domains (Alaka et al. 2017; Goldenberg et al. 2015), upgraded physics parameterizations (Zhang et al. 2015), and ocean coupling (Bender et al. 2019). However, the predictability of TCs is still an active and important area of research in the meteorology community because of the difficult nature of capturing the many scales of motion important in TC evolution.

A key limitation to TC predictability is the inability to accurately initialize the TC inner-core structure. This is partially due to a lack of adequate data coverage. One important reason for this is the availability of remotely sensed wind observations in and near a TC is particularly limited. The Advanced Scatterometer (ASCAT; Bentamy et al. 2012; Brennan et al. 2009), a space-based polar-orbiting scatterometer, occasionally measures a relatively coarse snapshot of the surface wind velocities during a coincidental TC overpass. The Cyclone Global Navigation Satellite System (CYGNSS) provides wind speed estimates (Ruf et al. 2016) at the surface and validation of these data are still in progress. Satellite-derived atmospheric motion vectors (AMVs) provide the majority of the wind observations near TCs but are often limited to the upper troposphere (Sears and Velden 2012) due to the presence of central dense overcast. Uncertainties in the AMV height assignments, AMV derivation method, and use of quality control can also lead to the utilization of fewer AMV observations by data assimilation systems (Velden et al. 2017). This combination of available remote measurement capabilities leaves the majority of the inner-core wind field unobserved, except when and where aircraft reconnaissance data are available.

Numerical models depend on data assimilation (DA) techniques to blend available observations with an existing model background to create an analysis. The recent operational DA systems for regional hurricane models are often based on an ensemble Kalman filter (EnKF) or a hybrid of an EnKF with a variational method (3DVar or 4DVar). Several studies have demonstrated how tropical cyclone track and intensity forecasts were improved through the use of an EnKF with increased wind observation coverage from hourly or rapid-scan AMVs (Li et al. 2020; Lim et al. 2019; Velden et al. 2017; Wu et al. 2014). Several studies have also tested how an EnKF or 3DVar systems can initialize a TC by ingesting observations from aircraft reconnaissance to produce improved TC forecasts (Nystrom and Zhang 2019; Christophersen et al. 2018; Aksoy et al. 2012; Pu et al. 2016). In the absence of the substantial datasets collected by aircraft reconnaissance, regional numerical models like the Hurricane Weather and Research Forecast (HWRF) model rely on a synthetic vortex in order to initialize a TC forecast (Biswas et al. 2017). This method has been shown to correct surface characteristics of the TC such as intensity, size, and location (Liu et al. 2020). However, the modifications have been shown to generate an unrealistic vertical structure of the TC that worsens in mature hurricanes (Tong et al. 2018).

To produce better forecasts, it is important to determine what observations will be needed when designing future observing systems. An effective way to examine potential options is to use observing system simulation experiments (OSSEs; Atlas 1997). Recent OSSE studies have quantified the impact of specific wind observing systems on TC forecasts. Several studies tested the forecast effect of simulated CYGNSS observations on TC prediction (Cui et al. 2019; Leidner et al. 2018; Annane et al. 2018; McNoldy et al. 2017; Zhang et al. 2017). They found that the data not only improved the short-term maximum sustained surface wind speed (referred to as intensity) forecast but also the overall structure of the surface wind field. Atlas et al. (2015b) explored the potential forecast impact of a Doppler wind lidar (DWL) on board the International Space Station. A small portion of the study was dedicated to regional TC forecasts and showed improvement in intensity error statistics with the addition of DWL winds. Atlas et al. (2015a) showed how the analysis of a mature TC generated from a grid of “error free” observations drawn directly from a nature run and assimilated by 3DVAR resulted in a 6-h forecast that did not weaken as a result of dynamical imbalances (known as “spindown”). No other past OSSE studies have explored the use of creating TC analyses using a grid of error-free observation profiles and DA outside of the context of a specific observing system.

This study begins with the assimilation of a domainwide grid of error-free profiles to understand how observation type and coverage influence the prediction of a hypothetical TC in a cycled EnKF data assimilation system. The aim of this paper is to investigate how wind observations modify the initial structure of a TC and its subsequent forecasts. The evaluation explores variations of the spatial and temporal wind profile coverage to take the initial steps toward a more realistic, traditional OSSE. However, for this study the observation coverage is idealized, and future work will add more realism in simulated observation coverage and error.

Section 2 describes the OSSE system used in this study. Section 3 presents idealized experiments that set a benchmark for the OSSE system by which subsequent experiments are compared. Section 4 discusses the experiments and sensitivity tests when assimilating idealized wind profiles. Section 5 presents conclusions and future work.

2. OSSE system description

The five key components of the OSSE system used in this study, that is, the Nature Run, synthetic observations sampled from the Nature Run, the data assimilation system, the forecast model, and the analysis and forecast verification, are described in this section.

a. Nature run

A Nature Run is typically generated using a global forecast model that is integrated in time for an extended period (Hoffman and Atlas 2016). It is used as the “truth” that all experiments aim to proxy through the assimilation of observations generated from the Nature Run. However, for high-impact, convective to mesoscale phenomena such as tropical cyclones, computational constraints have not yet permitted a global, convection-resolving simulation with frequent storage (every few minutes). To overcome this deficiency, a regional Nature Run uses initial and boundary conditions from the global Nature Run to simulate finer scales of motion. This study uses a regional WRF-ARW Hurricane Nature Run (HNR1; Nolan et al. 2013) with the global ECMWF T511 Nature Run (Reale et al. 2007) initial and boundary conditions. The HNR1 simulation employs a fixed parent domain over the Atlantic basin with 27-km grid spacing and three nested grids that follow the TC, with the innermost nest of grid spacing 1 km. There are 61 hybrid levels in the vertical. The HNR1 simulates the genesis, rapid intensification (RI), recurvature, and subsequent decay of an Atlantic Ocean TC over a 13-day period (Fig. 1). Other specifics and the parameterization schemes used in HNR1 are listed in Table 1. In this study, our attention is focused on a 4-day period (1–5 August), during which the TC intensifies from a strong tropical storm into a category-4 hurricane.

Fig. 1.
Fig. 1.

Hurricane regional OSSE domain configuration. The track of the hurricane is colored by intensity (kt) and is labeled at 0000 UTC of each date. The black outline represents the 27-km parent domain of the HNR1. The gray boundaries represent the HWRF parent domain (9 km) and an instance of the HWRF storm-following nest (3 km).

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

Table 1.

Comparison of model setup between the HNR1 and HWRF, OSSE framework forecast model.

Table 1.

b. Simulated observations

Observations are simulated from the HNR1 Nature Run. The first set of OSSEs in section 3 extracts “perfect” soundings of wind, temperature and moisture from the 27-km parent domain, at all grid points and 61 model levels. Although no errors are added directly to the simulated sounding data, a small error is assigned during the assimilation process as a requirement of the system. The second set of OSSEs, presented in section 4, simulates perfect wind vector profiles in a more realistic configuration of swaths from a polar-orbiting satellite. These profiles, created through bilinear interpolation, have a vertical resolution of 1 km (beginning at an altitude of 0.5 km) and a horizontal spacing of 70 km. The coverage over the domain repeats every 24 h. The swath of the polar-orbiter yielded full or partial coverage of the TC at 1200 UTC 1 August, 0600 UTC 2 August, 0600 UTC 3 August, and 0600 and 1800 UTC 4 August.

c. Data assimilation

A square root EnKF (Whitaker and Hamill 2002) is employed in the HWRF analysis–forecast framework. The EnKF contains 30 ensemble members and assimilates observations on only the 9-km HWRF parent domain shown in Fig. 1. The EnKF is adapted from the operational HWRF system (Biswas et al. 2017) and has been tuned for this domain size and resolution. Details on the tunable parameters of the EnKF include an inflation factor of 0.9 that is used on the posterior ensemble variance, a horizontal localization radius of 900 km, and a vertical localization of 1 scale length (in units of lnp). In the OSSE system, the data assimilation cycle begins with a 6-h forward integration from NCEP Global Forecast System (GFS) ensemble initial conditions at 0000 UTC 1 August. This is followed by 6-hourly EnKF cycling ending at 0000 UTC 5 August for a total of 16 analysis cycles.

d. HWRF forecast model

The forecast model used in this study is Version 3.4 of the HWRF atmospheric model (Biswas et al. 2017), which uses the Nonhydrostatic Mesoscale Model (NMM) dynamic core. It comprises a parent domain of 9-km grid spacing and a vortex-following nested domain of 3-km grid spacing. There are 61 vertical levels, with a model top of 50 hPa. A different set of parameterization schemes to the Nature Run is used to reduce unnecessary similarities with the simulated truth (Table 1) so that an imperfect model scenario is achieved to avoid the “identical twin” problem (Atlas 1997). Boundary and initial conditions for HWRF are provided by the GFS model that assimilates standard observations simulated from the global ECMWF T511 Nature Run. In this study, 120-h HWRF forecasts are generated from ensemble mean analyses every 6 h.

3. Near-perfect observational network: An estimated bound to analysis and forecast improvements

The OSSE framework offers the ability to evaluate different configurations of wind (and other) observations. To place this in the context of the maximum gain to HWRF analyses and forecasts that would be possible if a “near perfect” observational network were available, this section first examines highly idealized observations. Such a near-perfect network can be created by directly sampling the HNR1 without interpolating or adding any errors to the extracted values and then assimilating them in the EnKF to create an analysis. Deterministic forecasts are then generated every 6 h between 0600 UTC 1 August and 0000 UTC 5 August. The variables to be assimilated in this highly idealized configuration are temperature, specific humidity, and the zonal and meridional wind components (Table 2). Three combinations of these variables are assimilated in parallel cycled analysis-forecast experiments: 1) Wind + Mass, 2) Wind Only, and 3) Mass Only. “Wind” comprises the two components of the horizontal velocity, and “Mass” comprises temperature and moisture. A systematic analysis of each experiment offers the opportunity to examine the interaction between these two fundamentally different variable types.

Table 2.

List of the variable type and resolution assimilated with each OSSE experiment discussed in this study.

Table 2.

Each analysis and 6-h forecast (referred to as a background) is evaluated against the HNR1 truth. The biases and root-mean-square (RMS) errors of the different variables are averaged over the entire three-dimensional fixed outer HWRF domain for each cycle to evaluate how the OSSE system performed through the assimilation cycles (Fig. 2). As expected, the background RMS errors grow during each 6-h integration and are then reduced after each EnKF analysis is performed. The averaged analysis errors are gradually reduced over the first four cycles (until 0600 UTC 2 August), suggesting that this DA system needed a 24-h period to stabilize (after being initialized from the global ensemble). This is consistent for all three combinations of wind and mass assimilation, and for all verification variables. Following this initial 24-h period, the RMS errors stabilize for a few cycles until the TC begins its RI on 3 August. The averaged domainwide errors are largest during the period when the TC intensifies rapidly into a steady-state major hurricane (0600 UTC 3 August–1800 UTC 4 August).

Fig. 2.
Fig. 2.

(a) The volumetrically averaged RMS errors (solid lines) and bias (dashed lines) for the (a) u component of the wind (m s−1) and (b) temperature (°C) on the HWRF parent domain. Shown are the background (filled circles) and analyses (open circles) for each cycle for the Wind+Mass (purple), Wind Only (blue), and Mass Only (green) experiments. The period of rapid intensification is highlighted in orange.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

Even though observations are assimilated everywhere, the domainwide errors are found to be strongly dependent on the combination of variables assimilated. For example, in the Mass Only experiment the RMS errors of the volumetrically averaged zonal wind remain 0.5–1 m s−1 larger than in the Wind Only experiment. Similarly, the RMS errors of temperature are 0.1°C larger when only winds are assimilated. Limiting the assimilated variable to only winds also introduces a domainwide 0.1°–0.2°C average temperature bias. Comparable results are seen in other model variables, such as meridional wind and specific humidity (not shown). This result implies that the background error covariance matrix used in this particular EnKF is not providing comprehensive modifications to the mass variables in the analysis domain when wind observations are being assimilated (and vice versa).

The domain-averaged errors provide context when investigating differences in the structure of the analyzed TC. Particular attention is given to the RI period, given the noticeable deviation in the average domainwide errors between the three experiments. During this period, the azimuthally averaged tangential and radial wind fields from each of the three experiments are compared against each other and HNR1 (Figs. 3 and 4 ). The structural representation of the azimuthally averaged tangential winds that compares most favorably with HNR1 comes from both the Wind+Mass and the Wind Only experiments. They best capture the location and intensity of the tangential wind maxima in the eyewall. The Wind+Mass and Wind Only also create the best analysis of the radial wind field, with a similar representation of the strength and depth of the inflow (negative radial wind) and outflow (positive radial wind). In contrast, the Mass Only experiment is unable to capture the tangential wind maximum in the boundary layer. Instead, it creates a broader, weaker, and slightly shallower structure. The secondary circulation also contains an inflow layer that is too weak and diffuse. The Wind+Mass experiment displays a blend of the structures obtained in the separate Winds Only and Mass Only experiments.

Fig. 3.
Fig. 3.

The azimuthally averaged (left) tangential and (right) radial winds for the (a),(b) HNR1; (c),(d) Wind+Mass; (e),(f) Wind Only; and (g),(h) Mass Only experiments.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

Fig. 4.
Fig. 4.

Differences in the azimuthally averaged (left) tangential and (right) radial winds for 0600 UTC 3 Aug between the HNR1 and the (a),(b) Wind+Mass; (c),(d) Wind Only; and (e),(f) Mass Only experiments.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

Although the azimuthally averaged structures and differences in Figs. 3 and 4 provide a useful analysis of the general initial vortex, they are computed from the ensemble mean analysis. This averaging may yield a TC that is more diffuse than reality, smoothing important structural information especially if there is a large ensemble spread in the size, strength or location of the vortex. We therefore also examine how each analysis ensemble member captures the TC structure, by comparing the horizontal extent of 34-kt (tropical-storm force; 1 kt = 0.514 m s−1) surface winds against the truth in HNR1 (Fig. 5). The Wind+Mass members consistently capture the asymmetric horizontal distribution of the 34-kt winds and the correct center of the vortex (Fig. 5a). On the other hand, the Wind Only members underestimate the extent of the tropical storm force winds in the southern portion of the TC, although some ensemble members do capture the winds associated with the southern rainband feature that is not identified by the 34-kt contours in any of the Wind+Mass members (Fig. 5b). The Mass Only experiment is the least effective at capturing the southern portion of the TC and has the largest spread in the placement of the center of the vortex (Fig. 5c).

Fig. 5.
Fig. 5.

The 34-kt contours of wind speed at the surface (1000 hPa) for the (top) background and (bottom) analysis from the 0600 UTC 3 Aug cycle. HNR1 (black) is compared with all 30 ensemble members for the (a),(b) Wind+Mass (purple); (c),(d) Wind Only (blue); and (e),(f) Mass Only (green) experiments.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

The 5-day HWRF forecasts are initialized from each of these three sets of analysis fields, over the period of interest. The forecasts derived from the Wind+Mass experiment possess the lowest averaged intensity errors (Fig. 6a) for up to 72 h. This is closely followed by the Wind Only experiment. Although the verification of only intensity shows a difference between the three OSSEs out to 96 h and decreasing intensity errors at longer lead times, substantial differences in the results at all lead times are evident when other intensity metrics are used. For example when the overall kinetic energy at the surface of the TC is evaluated using the integrated kinetic energy (IKE), the Wind+Mass and the Wind Only experiments demonstrate consistently superior performances throughout the 5-day forecast (Fig. 6b). The Mass Only experiment is inferior at all forecast hours. This suggests that the Mass Only experiment is able to forecast intensity at longer lead times but does not accurately forecast the full extent of the area coverage of the surface wind field.

Fig. 6.
Fig. 6.

Homogeneous (same number of cases per forecast lead time) average (a) intensity error (m s−1), (b) total integrated kinetic energy error (TJ), and (d) error for the radius of 34-kt wind (n mi) over the 120-h forecast period for the Wind+Mass (purple), Wind Only (blue), and Mass Only (green) experiments. Also shown is the (c) normalized distribution of 10-m surface wind speeds at the 96-h forecast from the 0000 UTC 3 Aug cycle for the HNR1 1-km nest (black) and 3-km nest (gray) in addition to the three previously listed experiments.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

A further analysis of the 96-h HWRF forecast reveals that while the surface wind speed maximum is generally predicted well in all three experiments, the distributions of wind speeds within 200 km of the center do not closely resemble the corresponding distribution in HNR1 (Fig. 6c). The curves usually reflect a similar Gaussian-shaped distribution, but their means and modes are weaker by 10–15 m s−1 for all forecast hours and all cycles (not shown). This is likely due in part to the smaller horizontal grid spacing used in the HNR1 simulation (1 vs 3 km). A comparison with the 3-km nest from the HNR1 (gray line in Fig. 6c) reveals a closer resemblance to the HWRF forecasts. However, this is not a completely accurate comparison since the 1-km nest in the HNR1 feeds back to the 3-km nest, thus influencing the forecast intensity. Differences in the distributions among the three experiments are more subtle, but the Mass experiment distribution is consistently shifted further toward weaker winds (and hence away from HNR1).

The example of the wind distribution highlights how the HWRF does not closely replicate the size of the TC in HNR1. This is partially by design in the OSSE since an imperfect model will never reproduce nature. Figure 6d shows the average error in the radius of tropical-storm-force (34 kt) winds. The positive values indicate that the extent of that particular wind field is on average significantly smaller than the truth. The Mass Only experiment yields the smallest wind radii compared with the other two experiments, and hence the largest errors.

The differences in the analyzed and forecast vortex structure among the three experiments also has an impact on the average HWRF track forecast skill (Fig. 7). Overall, the Wind+Mass and the Wind Only experiments have the lowest track forecast errors through nearly the entire forecast period (Fig. 7). This general superiority over the Mass Only experiment is largely due to two reasons. First, in the earliest cycles, the vortex is weaker and shallower, and its center is more difficult to analyze compared to the other experiments. The Mass Only experiment therefore has much higher errors at shorter forecast times (0–48 h), but as the vortex strengthens and deepens, the track errors diminish and the TCs in all three experiments become collocated. Second, during the later cycles, the TCs in all three experiments are initially in the same location, but the differences in vortex depth and outflow strength lead to larger track errors in the Mass Only experiment at later forecast times (72–120 h; not shown).

Fig. 7.
Fig. 7.

Average track error (km) over the 120-h forecast period for the Wind+Mass (purple), Wind Only (blue), and Mass Only (green) experiments.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

4. Reduced observational network: Idealized wind profiler

One objective of the previous section was to establish a threshold for analysis and forecast skill, by assimilating near-perfect wind and thermodynamic observations at all grid points throughout the domain. This section investigates how spatial and temporal reductions in the coverage of wind observations modifies the analyzed TC structure and forecast statistics. These types of degradations were chosen as small steps toward a more realistic set of observations collected from a space-based platform. Two additional sets of OSSEs are designed to achieve this (Table 2). One set comprises three experiments that temporally vary the swath coverage from an idealized polar-orbiting profiler with perfect wind observations (Fig. 8a). The other set comprises four experiments that use an identical swath coverage of idealized wind profiles, but with wind profiles removed over the TC region (if present) at varying radii from the center (Fig. 8b). In particular, the swaths cover the TC at 0600 UTC 3 August and 0600 UTC 4 August. It is important to note that this work is also idealized in the sense that there are no competing wind observations from other hypothetical observing systems.

Fig. 8.
Fig. 8.

(a) The 24-period of wind profiler observation locations. Colors indicate the center time of the assimilation window: Red is 0000 UTC, yellow is 0600 UTC, purple is 1200 UTC, and green is 1800 UTC. The HNR1 track is color coded in the same colors. (b) Sample wind profile coverage from the 0600 UTC 3 Aug assimilation cycle window. Colors indicate distance from the center of the storm center. Red is more than 500 km, purple is 300–500 km, pink is 200–300, orange is 100–200, and gray is within 100 km of the center.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

a. Varying temporal coverage of wind profiler

The goal of this experiment set is to simulate the randomness in which polar-orbiting satellites occasionally observe TCs and understand how altered coverage changes the results. Swaths, with an approximate width of 850 km, of wind profiles are produced to imitate coverage from polar-orbiting satellites. Both the zonal and meridional wind components are directly interpolated from the parent domain (27-km grid) of the HNR1 without any errors added, and there is no removal of data due to cloud or rain attenuation. Each wind profile is approximately 70 km apart and the orbital pattern repeats every 24 h. Three sensitivity experiments are performed. The first assimilates the observations as simulated and the second two shift the timing of the coverage by 6 and 12 h. Therefore different swaths are observing the TC 6 and 12 h later into the storm evolution. The three configurations are referred to as Swath, Swath6, and Swath12.

Similar to the previous section, domainwide RMS errors and biases are first calculated for the background and analysis of each individual cycle for each of the three configurations (Fig. 9). After the first 24 h, the errors stabilize and are indistinguishable among all three experiments for a few cycles until the TC begins RI at 0600 UTC 3 August (Fig. 9a). When the swath of wind profiles samples even a portion of the inner core, the domainwide RMS error reduces substantially. In the absence of inner-core observations (Fig. 9c) over the next 4 cycles, the RMS errors grow until the next swath coincides with the TC and its immediate environment. This is in substantial contrast to the domainwide Winds Only experiment (Fig. 2a) where no error growth occurs on the domain scale when wind observations are available everywhere at each cycle time. The error growth was largest making the reduction greatest in the Swath12 experiment, which collected observations late in the RI period (on 1800 UTC 3 August as compared with 0600 UTC 3 August). When inner-core observations are assimilated in the early stages of a TC structural change (i.e., onset of RI), the errors in the subsequent cycles remain lower throughout the following cycles. This is in contrast to experiments in which inner-core observations are introduced later in the RI process and the model background cannot reproduce the rapid reorganization of the TC, which leads to larger errors. The smallest changes between the background and analysis occurred when the observations were only in the northeast portion of the domain, far from the TC environment (red swath in Fig. 8a). When compared to the temporal influence of the wind profile swaths on the wind analyses, there is little difference in the analyses and short-term forecasts of the temperature (Fig. 9b).

Fig. 9.
Fig. 9.

As in Fig. 2, but for the Wind Only (blue), Swath (light blue), Swath6 (light green), and Swath12 (light purple) experiments. Also shown is the (c) total number of observations assimilated within 300 km of the TC center for each of the Swath experiments.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

Several structural differences result from inconsistent wind coverage over the TC. When the TC reaches hurricane strength, it is observed by the wind profiler. The assimilation of these wind profiler data creates stronger azimuthally averaged tangential wind analysis, compared against the corresponding background field (Figs. 10b,c,f,g,i,j). However, the TC is still too weak and broad when compared to the HNR1. The secondary circulation is also strengthened when inner-core observations are assimilated but much weaker than the HNR1 (not shown). The weaker analyses could be caused by a small set of observations located in the inner core of the TC that were rejected by the assimilation system due to large innovations.

Fig. 10.
Fig. 10.

Azimuthally averaged tangential wind (m s−1) for (left) HNR1, (center) model background, and (right) analysis. Plots are the cycles in which inner-core observations were present for the Swath experiment at (a)–(c) 0600 UTC 3 Aug, (e)–(g) Swath6 experiment at 1200 UTC 3 Aug, and (h)–(j) Swath12 experiment at 1800 UTC 3 Aug. Blue contour lines are the radii of 34-, 50-, and 64-kt winds representing the gale force, damaging, and destructive wind locations, respectively.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

Next, how the analysis of each ensemble member captures the horizontal extent of 34-kt surface winds is evaluated in comparison with HNR1. The Swath experiment, representing the onset of RI on 0600 UTC 3 August, centers its members on HNR1 in the northern and eastern quadrants of the analysis instead of members straying as far as 1000 km from the truth in the background (Figs. 11a,b). However, this analysis has substantial spread in the vortex placement and the southwestern portion of the surface wind field. The Swath6 experiment demonstrates similar results with the difference that a larger fraction of ensemble members tend to underestimate, or place the 34-kt winds at smaller radii, the northern and eastern quadrants (Figs. 11c,d). The Swath12 experiment, containing the strongest vortex near the end of RI, shows the least spread in the analyzed extent of the wind field in most quadrants. However, there is still uncertainty in the placement of the TC center (Figs. 11e,f).

Fig. 11.
Fig. 11.

As in Fig. 5, but for the (a),(b) Swath (blues); (c),(d) Swath6 (greens); and (e),(f) Swath12 (purples) experiments.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

The averaged TC metrics in all swath experiments (Swath, Swath6, and Swath12) generally show little sensitivity to the timing of the observation coverage (Fig. 12). The original swath configuration that introduced inner-core observations earlier in the RI period produced the lowest maximum wind speed (intensity) errors in the 12–72-h forecast period (Fig. 12a). The broader wind field, evaluated through both the averaged errors of the IKE and radius of 34-kt wind, does not reflect this slight improvement (Figs. 12c,d). The track forecasts of the three experiments showed no significant differences (Fig. 12b).

Fig. 12.
Fig. 12.

Homogeneous average (a) intensity error (m s−1), (b) track error (km), (c) total integrated kinetic energy error (TJ), and (d) error for the radius of 34-kt wind (n mi) over the 120-h forecast period for the Swath (light blue), Swath6 (light green), and Swath12 (light purple) experiments.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

b. Influence of TC inner-core wind observations

The final OSSE investigates the importance of inner-core wind observations. The synthetic idealized wind profiler observations from the previous experiments are used; however, all vertical wind profiles in the swath are removed within varying radii from the TC center (100, 200, 300, and 500 km). The purpose of this OSSE is to test an idealized and rudimentary simulation of the presence of dense high-altitude cloud coverage often associated with hurricanes, which could limit the retrieval of vertical wind profiles from a space-based wind profiling instrument. Another goal is to understand the contributions of observations in the near-storm environment on the analysis and forecast of the TC. This set of sensitivity experiments uses the results presented in the previous section as a benchmark.

Domainwide RMS and bias errors are again calculated and compared for each experiment (Fig. 13). The first partial swath coverage occurs in the tropical storm stage at 0600 UTC 2 August. At this time, the experiments with fewer inner-core observations (noTC300 and noTC500) have RMS errors that grow larger with subsequent cycles than the other experiments. This trend continues during the RI phase where the errors progressively grow with each observational radial limit as the TC strengthens. Errors from the noTC500 experiment reach their peak for the entire experiment period. This is in contrast to the experiments that include observations 200 km or closer to the storm center. The noTC300 errors appear to fall on a critical gradient during the 0600 UTC 3 August cycle. The noTC300 background error behaves similarly to the noTC500 experiment but as observations are introduced in the near storm environment errors are reduced and are comparable to those of noTC200. On 0600 UTC 4 August there is another pass providing partial inner-core and near-storm observations that substantially reduces the domainwide RMSE in all but noTC500 to nearly the values computed in the full swath experiment.

Fig. 13.
Fig. 13.

As in Fig. 2a, but for the noTC100 (orange), noTC200 (pink), noTC300 (purple), noTC500 (maroon), Swath (light blue), and Wind Only (blue) experiments

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

The reduction in error at the onset of RI (0600 UTC 3 August) corresponds to considerable differences in the analyzed TC structure (Fig. 14). The azimuthally averaged tangential wind produced by each of the sensitivity experiments at the onset of RI show weaker, broader, and shallower circulations when observations were only available farther from the TC center. The structural differences between the noTC100 and noTC200 are small, with the vortex becoming slightly shallower as more observations are removed (Figs. 14c,d). However, the differences become more apparent as the radius is increased to 300 and 500 km (Figs. 14e,f). The vortex is so broad and weak in the noTC500 experiment that it is only analyzing a tropical-depression-like feature when the HNR1 is showing a hurricane.

Fig. 14.
Fig. 14.

Azimuthally averaged tangential wind (m s−1) at 0600 UTC 3 Aug for (a) HNR1, and the analyses from the (b) Swath, (c) noTC100, (d) noTC200, (e) noTC300, and (f) noTC500 experiments.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

As in Figs. 5 and 11, the contours of 34-kt wind at 1000 hPa in each of the background and analysis ensemble members are again used to illustrate differences between the respective ensembles in each experiment (Fig. 15). As the radius of exclusion of the wind profiles is increased, the difference between the background and analysis diminishes. As with the regular Swath experiment, the noTC100 experiment generally shows an improvement and reduction of spread in both the TC center and the exterior 34-kt radius (Figs. 15a,b). The noTC300 experiment shifts the ensemble members to a more accurate location when comparing the background to the analysis, but the extent of the wind field still varies greatly creating a large spread in vortex structures (Figs. 15c,d). In particular, it underestimates the wind speeds in the southwest quadrant of the storm. The noTC500 corrects neither the location nor the size of the vortex. A larger portion of the TC on the south and west side is also underestimated in most ensemble members. In summary, the TC inner-core structure is modified less as observations are restricted farther from the center. This could be a result of the choice of the horizontal length scale (900 km).

Fig. 15.
Fig. 15.

As in Fig. 5, but for the (a),(b) noTC100 (oranges); (c),(d) noTC300 (dark purples); and (e),(f) noTC500 (reds) experiments.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

The differences in analyzed TC structures produced a substantial impact on TC error statistics depending on the metric (Fig. 16). For example, the significantly weaker noTC500 took much longer to spin up and began with an average TC intensity error that was 14.85 m s−1 larger than the noTC100 at 0 h (Fig. 16a). After 72 h the average errors from each experiment fell within 3.5 m s−1 of each other. Larger storm-scale metrics, such as total IKE and radius of 34-kt wind errors, became similar across all experiments later into the forecast period. The average radius of 34-kt winds errors begin with a difference of 46 n mi (1 n mi = 1.852 km) between noTC500 and noTC100 (Fig. 16d). All experiments fall within a 2.5 n mi error by forecast hour 108. The one exception to the error statistic trend occurs in the averaged track errors (Fig. 16c). Although the initial position error becomes larger as more data are omitted from the storm, the result is reversed in the 24–120-h track errors. This is likely the product of the shallow steering flow in the experiments being more representative of the deep steering flow in the HNR1, thus placing the TC in the correct position despite having a vastly different structure.

Fig. 16.
Fig. 16.

As in Fig. 12, but for the noTC100 (orange), noTC300 (dark purple), and noTC500 (maroon) experiments.

Citation: Monthly Weather Review 149, 6; 10.1175/MWR-D-20-0153.1

5. Conclusions and discussion

This study uses an OSSE approach to understand the response of TC analysis and forecasts to the assimilation of different configurations of vertical wind profiles. Using a state-of-the-art Nature Run (HNR1), an operational-quality forecast model (HWRF) and assimilation system (EnKF), three different sets of experiments are designed. The first set of experiments assimilated near-perfect observations of winds, temperature, and moisture throughout the entire parent domain and defined a threshold for the subsequent sets of experiments. Using a domainwide verification first, it was found that when a variable type was not assimilated, the background error covariance matrix did not adequately compensate for the lack of the removed observation type. This highlights the importance of the direct measurement and assimilation of a variable. Since TC intensity and structure metrics are most frequently derived from wind observations, it is critical to regularly observe the wind field to make a more accurate analysis and forecast. However, the overall best results come with complete coverage of the most variables available.

Previous studies using observations collected by hurricane reconnaissance aircraft (Tong et al. 2018; Lu et al. 2017; Poterjoy and Zhang 2011) have shown that the cross-variable EnKF update was sufficient given one observation type (i.e., winds, temperature or pressure). These studies relied on partial coverage of the TC inner core provided by onboard instrumentation. The coverage becomes especially limited in the thermodynamic fields throughout the storm and portions of the kinematic fields as described in the introduction. Our study, like the previous studies, has found that with sufficient observational coverage of the TC winds (provided by the Tail Doppler Radar; Gamache 1997), an EnKF can produce a reasonable analysis of the TC wind structure and improve TC prediction. However, our study proposes that the cross-variable update may not be as sufficient as previously suggested. A potential shortcoming of our study that could result in the inadequate cross-variable update is the use of a smaller ensemble resulting in a larger sampling error (Zhang and Weng 2015; Poterjoy et al. 2014; Torn and Hakim 2009). While the ensemble size for this study has been used for other TC data assimilation studies, it would be useful to repeat this experiment with larger ensembles in a future study.

It is important to note that the TC forecast metrics of track and intensity alone portrayed deceiving conclusions. In this study, the TC was found to spin up to the approximate intensity of the HNR1 by the end of the forecast. This is likely related to the somewhat predictable nature of Atlantic recurving TCs and metrics used to classify the storms (Judt et al. 2016; Kieu and Moon 2016). Metrics that represented the behavior of larger-scale features displayed more robustness. The assimilation of perfect wind profile observations over the domain yielded improved size and IKE statistics, and analyzed azimuthally averaged surface wind structure. These results highlight the importance of extending “standard” TC metrics to regularly include structural verification.

The second set of experiments investigates how a spatial reduction in observation coverage into more realistic space-based swaths, together with the timing of this coverage, modified the TC structure and forecast statistics. Reduction in spatial coverage results in increased errors and variation in cycle-to-cycle error growth. The timing of the coverage was found to be important in the context of the TC life cycle, especially during periods of significant structural change such as rapid intensification. Even sporadic coverage during the RI phase provided the necessary observations to reduce domainwide error statistics and create a more realistic azimuthally averaged circulation. This error reduction was even maintained for several cycles until the next coincidental inner-core observations. However, the timing of the observational coverage did not have a noteworthy influence on the overall forecast metrics.

The third set of experiments demonstrated the importance of assimilating observations within the inner core of the TC. When all observations within 500 km of the TC center were removed, the combination of the localization length and error covariance matrix in the DA system did not provide a sufficient replacement for direct inner-core measurements. The averaged domainwide analysis and forecast errors grew much more than in the experiments with partial inner-core coverage (r ≤ 300 km), and the azimuthally averaged structure became too broad and weak. The 0–96-h forecasts of wind structure and intensity were also significantly degraded. The only benefit of removing the inner-core observations was the much shallower vortex that followed steering currents that more closely resembled the truth, resulting in an improved track forecast. This, however, is a classic example of the right result for the wrong reason. Overall, we conclude here that for a state-of-the-art data assimilation framework, wind profile measurements need to be sampled as close to the TC inner core as possible to reduce analysis and forecast errors. This set of experiments does not account for variations in size of the TC with time (Ryan et al. 2019).

The intention of this study was to lay the groundwork for more advanced OSSEs, aimed at the understanding of how the assimilation of wind profiles from space-based Doppler Wind lidar would affect TC analyses and forecasts. Future planned experiments include a systematic evaluation of idealized vertical coverage of wind observations in and around a TC to determine the most impactful observations (Torn 2014). Further refinements of existing experiments include the assimilation of Line-of-Sight observations in various spatial coverage patterns and the addition of exisiting observations that are assimilated operationally. By reducing the observation coverage in cloudy areas, the influence of attenuation due to clouds also can be quantified. Similarly, a reduction from a swath coverage to a single line of profiles could be studied to understand the cost benefit of future potential instrument configurations.

Synergies in the wind observations could be introduced in the assimilation process to enhance other wind observing technologies, such as improved height assignments for atmospheric motion vectors. It could also be used to stage a more complete three-dimensional analysis of the TC winds throughout its life cycle. For example, TC genesis studies (Zawislak and Zipser 2014) suffer from a lack of wind observations because of the lack of precipitation over the extent of the system. A swath of high-vertical-resolution wind profiles could inform forecasters and numerical models about the development of a TC’s incipient circulation over the depth of the mid- to lower troposphere. This could then be followed by an airborne radar later in the TC development process to achieve regular observational coverage of the wind field. As new wind observing systems are developed, it is important to understand what their role may be in our analysis and forecast systems.

Acknowledgments

This work was supported by NASA Grant NNX13AQ36G and by NOAA’s Quantitative Observing System Assessment Program (QOSAP). We greatly appreciate Javier Delgado, Kelly Ryan, and Bachir Annane for their work in setting up the OSSE framework used for this study. We thank Altug Aksoy, Kelly Ryan, and Matthew Onderlinde for their useful insight and helpful suggestions in communicating our science.

REFERENCES

  • Aksoy, A., S. Lorsolo, T. Vukicevic, K. J. Sellwood, S. D. Aberson, and F. Zhang, 2012: The HWRF Hurricane Ensemble Data Assimilation System (HEDAS) for high-resolution data: The impact of airborne Doppler radar observations in an OSSE. Mon. Wea. Rev., 140, 18431862, https://doi.org/10.1175/MWR-D-11-00212.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Alaka, G. J., X. Zhang, S. G. Gopalakrishnan, S. B. Goldenberg, and F. D. Marks, 2017: Performance of basin-scale HWRF tropical cyclone track forecasts. Wea. Forecasting, 32, 12531271, https://doi.org/10.1175/WAF-D-16-0150.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Annane, B., B. McNoldy, S. M. Leidner, R. Hoffman, R. Atlas, and S. J. Majumdar, 2018: A study of the HWRF analysis and forecast impact of realistically simulated CYGNSS observations assimilated as scalar wind speeds and as VAM wind vectors. Mon. Wea. Rev., 146, 22212236, https://doi.org/10.1175/MWR-D-17-0240.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Atlas, R., 1997: Atmospheric observations and experiments to assess their usefulness in data assimilation. J. Meteor. Soc. Japan, 75, 111130, https://doi.org/10.2151/jmsj1965.75.1B_111.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Atlas, R., L. Bucci, B. Annane, R. Hoffman, and S. Murillo, 2015a: Observing system simulation experiments to assess the potential impact of new observing systems on hurricane forecasting. Mar. Technol. Soc. J., 49, 140148, https://doi.org/10.4031/MTSJ.49.6.3.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Atlas, R., and Coauthors, 2015b: Observing system simulation experiments (OSSEs) to evaluate the potential impact of an optical autocovariance wind lidar (OAWL) on numerical weather prediction. J. Atmos. Oceanic Technol., 32, 15931613, https://doi.org/10.1175/JTECH-D-15-0038.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bender, M. A., T. Marchok, R. E. Tuleya, I. Ginis, V. Tallapragada, and S. J. Lord, 2019: Hurricane model development at GFDL: A collaborative success story from a historical perspective. Bull. Amer. Meteor. Soc., 100, 17251736, https://doi.org/10.1175/BAMS-D-18-0197.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bentamy, A., S. A. Grodsky, J. A. Carton, D. Croizé-Fillon, and B. Chapron, 2012: Matching ASCAT and QuikSCAT winds. J. Geophys. Res., 117, C02011, https://doi.org/10.1029/2011JC007479.

    • Search Google Scholar
    • Export Citation
  • Biswas, M. K., and Coauthors, 2017: Hurricane Weather Research and Forecasting (HWRF) model: 2016 scientific documentation. Developmental Testbed Center Doc., 28 pp., https://dtcenter.org/HurrWRF/users/docs/scientific_documents/HWRFv3.8a_ScientificDoc.pdf.

  • Brennan, M. J., C. C. Hennon, and R. D. Knabb, 2009: The operational use of QuikSCAT ocean surface vector winds at the National Hurricane Center. Wea. Forecasting, 24, 621645, https://doi.org/10.1175/2008WAF2222188.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Christophersen, H., A. Aksoy, J. Dunion, and S. Aberson, 2018: Composite impact of Global Hawk unmanned aircraft dropwindsondes on tropical cyclone analyses and forecasts. Mon. Wea. Rev., 146, 22972314, https://doi.org/10.1175/MWR-D-17-0304.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cui, Z., Z. Pu, V. Tallapragada, R. Atlas, and C. Ruf, 2019: A preliminary impact study of CYGNSS ocean surface wind speeds on numerical simulations of hurricanes. Geophys. Res. Lett., 46, 29842992, https://doi.org/10.1029/2019GL082236.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gamache, J. F., 1997: Evaluation of a fully three-dimensional variational Doppler analysis technique. Preprints, 28th Conf. on Radar Meteorology, Austin, TX, Amer. Meteor. Soc., 422–423.

    • Search Google Scholar
    • Export Citation
  • Goldenberg, S. B., S. G. Gopalakrishnan, V. Tallapragada, T. Quirino, F. Marks, S. Trahan, X. Zhang, and R. Atlas, 2015: The 2012 triply nested, high-resolution operational version of the Hurricane Weather Research and Forecasting Model (HWRF): Track and intensity forecast verifications. Wea. Forecasting, 30, 710729, https://doi.org/10.1175/WAF-D-14-00098.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hoffman, R. N., and R. Atlas, 2016: Future observing system simulation experiments. Bull. Amer. Meteor. Soc., 97, 16011616, https://doi.org/10.1175/BAMS-D-15-00200.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Judt, F., S. S. Chen, and J. Berner, 2016: Predictability of tropical cyclone intensity: Scale-dependent forecast error growth in high-resolution stochastic kinetic-energy backscatter ensembles. Quart. J. Roy. Meteor. Soc., 142, 4357, https://doi.org/10.1002/qj.2626.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kieu, C. Q., and Z. Moon, 2016: Hurricane intensity predictability. Bull. Amer. Meteor. Soc., 97, 18471857, https://doi.org/10.1175/BAMS-D-15-00168.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Leidner, S. M., B. Annane, B. McNoldy, R. Hoffman, and R. Atlas, 2018: Variational analysis of simulated ocean surface winds from the Cyclone Global Navigation Satellite System (CYGNSS) and evaluation using a regional OSSE. J. Atmos. Oceanic Technol., 35, 15711584, https://doi.org/10.1175/JTECH-D-17-0136.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, J., J. Li, C. Velden, P. Wang, T. J. Schmit, and J. Sippel, 2020: Impact of rapid-scan-based dynamical information from GOES-16 on HWRF hurricane forecasts. J. Geophys. Res. Atmos., 125, e2019JD031647, https://doi.org/10.1029/2019JD031647.

    • Search Google Scholar
    • Export Citation
  • Lim, A. H. N., J. A. Jung, S. E. Nebuda, J. M. Daniels, W. Bresky, M. Tong, and V. Tallapragada, 2019: Tropical cyclone forecasts impact assessment from the assimilation of hourly visible, shortwave, and clear-air water vapor atmospheric motion vectors in HWRF. Wea. Forecasting, 34, 177198, https://doi.org/10.1175/WAF-D-18-0072.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, Q., and Coauthors, 2020: Vortex initialization in the NCEP operational hurricane models. Atmosphere, 11, 968, https://doi.org/10.3390/atmos11090968.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lu, X., X. Wang, Y. Li, M. Tong, and X. Ma, 2017: GSI-based ensemble-variational hybrid data assimilation for HWRF for hurricane initialization and prediction: Impact of various error covariances for airborne radar observation assimilation. Quart. J. Roy. Meteor. Soc., 143, 223239, https://doi.org/10.1002/qj.2914.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McNoldy, B. D., B. Annane, S. Majumdar, J. Delgado, L. Bucci, and R. Atlas, 2017: Impact of assimilating CYGNSS data on tropical cyclone analyses and forecasts in a regional OSSE framework. Mar. Technol. Soc. J., 51, 715, https://doi.org/10.4031/MTSJ.51.1.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nolan, D. S., R. Atlas, K. T. Bhatia, and L. R. Bucci, 2013: Development and validation of a hurricane nature run using the joint OSSE nature run and the WRF model. J. Adv. Model. Earth Syst., 5, 382405, https://doi.org/10.1002/jame.20031.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nystrom, R. G., and F. Zhang, 2019: Practical uncertainties in the limited predictability of the record-breaking intensification of Hurricane Patricia (2015). Mon. Wea. Rev., 147, 35353556, https://doi.org/10.1175/MWR-D-18-0450.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Poterjoy, J., and F. Zhang, 2011: Dynamics and structure of forecast error covariance in the core of a developing hurricane. J. Atmos. Sci., 68, 15861606, https://doi.org/10.1175/2011JAS3681.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Poterjoy, J., F. Zhang, and Y. Weng, 2014: The effects of sampling errors on the EnKF assimilation of inner-core hurricane observations. Mon. Wea. Rev., 142, 16091630, https://doi.org/10.1175/MWR-D-13-00305.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pu, Z., S. Zhang, M. Tong, and V. Tallapragada, 2016: Influence of the self-consistent regional ensemble background error covariance on hurricane inner-core data assimilation with the GSI-based hybrid system for HWRF. J. Atmos. Sci., 73, 49114925, https://doi.org/10.1175/JAS-D-16-0017.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reale, O., J. Terry, M. Masutani, E. Andersson, L. P. Riishojgaard, and J. C. Jusem, 2007: Preliminary evaluation of the European Centre for Medium-Range Weather Forecasts’ (ECMWF) nature run over the tropical Atlantic and African monsoon region. Geophys. Res. Lett., 34, L22810, https://doi.org/10.1029/2007GL031640.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ruf, C. S., and Coauthors, 2016: New ocean winds satellite mission to probe hurricanes and tropical convection. Bull. Amer. Meteor. Soc., 97, 385395, https://doi.org/10.1175/BAMS-D-14-00218.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ryan, K., L. Bucci, J. Delgado, R. Atlas, and S. Murillo, 2019: Impact of Gulfstream-IV dropsondes on tropical cyclone prediction in a regional OSSE system. Mon. Wea. Rev., 147, 29612977, https://doi.org/10.1175/MWR-D-18-0157.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sears, J., and C. S. Velden, 2012: Validation of satellite-derived atmospheric motion vectors and analyses around tropical disturbances. J. Appl. Meteor. Climatol., 51, 18231834, https://doi.org/10.1175/JAMC-D-12-024.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tong, M., and Coauthors, 2018: Impact of assimilating aircraft reconnaissance observations on tropical cyclone initialization and prediction using operational HWRF and GSI ensemble–variational hybrid data assimilation. Mon. Wea. Rev., 146, 41554177, https://doi.org/10.1175/MWR-D-17-0380.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Torn, R. D., 2014: The impact of targeted dropwindsonde observations on tropical cyclone intensity forecasts of four weak systems during PREDICT. Mon. Wea. Rev., 142, 28602878, https://doi.org/10.1175/MWR-D-13-00284.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Torn, R. D., and G. J. Hakim, 2009: Ensemble data assimilation applied to RAINEX observations of Hurricane Katrina (2005). Mon. Wea. Rev., 137, 28172829, https://doi.org/10.1175/2009MWR2656.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Velden, C., W. E. Lewis, W. Bresky, D. Stettner, J. Daniels, and S. Wanzong, 2017: Assimilation of high-resolution satellite-derived atmospheric motion vectors: Impact on HWRF forecasts of tropical cyclone track and intensity. Mon. Wea. Rev., 145, 11071125, https://doi.org/10.1175/MWR-D-16-0229.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Weng, Y., and F. Zhang, 2016: Advances in convection-permitting tropical cyclone analysis and prediction through EnKF assimilation of reconnaissance aircraft observations. J. Meteor. Soc. Japan, 94, 345358, https://doi.org/10.2151/jmsj.2016-018.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Whitaker, J. S., and T. M. Hamill, 2002: Ensemble data assimilation without perturbed observations. Mon. Wea. Rev., 130, 19131924, https://doi.org/10.1175/1520-0493(2002)130<1913:EDAWPO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wu, T., H. Liu, S. J. Majumdar, C. S. Velden, and J. L. Anderson, 2014: Influence of assimilating satellite-derived atmospheric motion vector observations on numerical analyses and forecasts of tropical cyclone track and intensity. Mon. Wea. Rev., 142, 4971, https://doi.org/10.1175/MWR-D-13-00023.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zawislak, J., and E. J. Zipser, 2014: Analysis of the thermodynamic properties of developing and nondeveloping tropical disturbances using a comprehensive dropsonde dataset. Mon. Wea. Rev., 142, 12501264, https://doi.org/10.1175/MWR-D-13-00253.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, F., and Y. Weng, 2015: Predicting hurricane intensity and associated hazards: A five-year real-time forecast experiment with assimilation of airborne Doppler radar observations. Bull. Amer. Meteor. Soc., 96, 2533, https://doi.org/10.1175/BAMS-D-13-00231.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, J. A., D. S. Nolan, R. F. Rogers, and V. Tallapragada, 2015: Evaluating the impact of improvements in the boundary layer parameterization on hurricane intensity and structure forecasts in HWRF. Mon. Wea. Rev., 143, 31363155, https://doi.org/10.1175/MWR-D-14-00339.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, S., Z. Pu, D. J. Posselt, and R. Atlas, 2017: Impact of CYGNSS ocean surface wind speeds on numerical simulations of a hurricane in observing system simulation experiments. J. Atmos. Oceanic Technol., 34, 375383, https://doi.org/10.1175/JTECH-D-16-0144.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save
  • Aksoy, A., S. Lorsolo, T. Vukicevic, K. J. Sellwood, S. D. Aberson, and F. Zhang, 2012: The HWRF Hurricane Ensemble Data Assimilation System (HEDAS) for high-resolution data: The impact of airborne Doppler radar observations in an OSSE. Mon. Wea. Rev., 140, 18431862, https://doi.org/10.1175/MWR-D-11-00212.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Alaka, G. J., X. Zhang, S. G. Gopalakrishnan, S. B. Goldenberg, and F. D. Marks, 2017: Performance of basin-scale HWRF tropical cyclone track forecasts. Wea. Forecasting, 32, 12531271, https://doi.org/10.1175/WAF-D-16-0150.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Annane, B., B. McNoldy, S. M. Leidner, R. Hoffman, R. Atlas, and S. J. Majumdar, 2018: A study of the HWRF analysis and forecast impact of realistically simulated CYGNSS observations assimilated as scalar wind speeds and as VAM wind vectors. Mon. Wea. Rev., 146, 22212236, https://doi.org/10.1175/MWR-D-17-0240.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Atlas, R., 1997: Atmospheric observations and experiments to assess their usefulness in data assimilation. J. Meteor. Soc. Japan, 75, 111130, https://doi.org/10.2151/jmsj1965.75.1B_111.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Atlas, R., L. Bucci, B. Annane, R. Hoffman, and S. Murillo, 2015a: Observing system simulation experiments to assess the potential impact of new observing systems on hurricane forecasting. Mar. Technol. Soc. J., 49, 140148, https://doi.org/10.4031/MTSJ.49.6.3.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Atlas, R., and Coauthors, 2015b: Observing system simulation experiments (OSSEs) to evaluate the potential impact of an optical autocovariance wind lidar (OAWL) on numerical weather prediction. J. Atmos. Oceanic Technol., 32, 15931613, https://doi.org/10.1175/JTECH-D-15-0038.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bender, M. A., T. Marchok, R. E. Tuleya, I. Ginis, V. Tallapragada, and S. J. Lord, 2019: Hurricane model development at GFDL: A collaborative success story from a historical perspective. Bull. Amer. Meteor. Soc., 100, 17251736, https://doi.org/10.1175/BAMS-D-18-0197.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bentamy, A., S. A. Grodsky, J. A. Carton, D. Croizé-Fillon, and B. Chapron, 2012: Matching ASCAT and QuikSCAT winds. J. Geophys. Res., 117, C02011, https://doi.org/10.1029/2011JC007479.

    • Search Google Scholar
    • Export Citation
  • Biswas, M. K., and Coauthors, 2017: Hurricane Weather Research and Forecasting (HWRF) model: 2016 scientific documentation. Developmental Testbed Center Doc., 28 pp., https://dtcenter.org/HurrWRF/users/docs/scientific_documents/HWRFv3.8a_ScientificDoc.pdf.

  • Brennan, M. J., C. C. Hennon, and R. D. Knabb, 2009: The operational use of QuikSCAT ocean surface vector winds at the National Hurricane Center. Wea. Forecasting, 24, 621645, https://doi.org/10.1175/2008WAF2222188.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Christophersen, H., A. Aksoy, J. Dunion, and S. Aberson, 2018: Composite impact of Global Hawk unmanned aircraft dropwindsondes on tropical cyclone analyses and forecasts. Mon. Wea. Rev., 146, 22972314, https://doi.org/10.1175/MWR-D-17-0304.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cui, Z., Z. Pu, V. Tallapragada, R. Atlas, and C. Ruf, 2019: A preliminary impact study of CYGNSS ocean surface wind speeds on numerical simulations of hurricanes. Geophys. Res. Lett., 46, 29842992, https://doi.org/10.1029/2019GL082236.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gamache, J. F., 1997: Evaluation of a fully three-dimensional variational Doppler analysis technique. Preprints, 28th Conf. on Radar Meteorology, Austin, TX, Amer. Meteor. Soc., 422–423.

    • Search Google Scholar
    • Export Citation
  • Goldenberg, S. B., S. G. Gopalakrishnan, V. Tallapragada, T. Quirino, F. Marks, S. Trahan, X. Zhang, and R. Atlas, 2015: The 2012 triply nested, high-resolution operational version of the Hurricane Weather Research and Forecasting Model (HWRF): Track and intensity forecast verifications. Wea. Forecasting, 30, 710729, https://doi.org/10.1175/WAF-D-14-00098.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hoffman, R. N., and R. Atlas, 2016: Future observing system simulation experiments. Bull. Amer. Meteor. Soc., 97, 16011616, https://doi.org/10.1175/BAMS-D-15-00200.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Judt, F., S. S. Chen, and J. Berner, 2016: Predictability of tropical cyclone intensity: Scale-dependent forecast error growth in high-resolution stochastic kinetic-energy backscatter ensembles. Quart. J. Roy. Meteor. Soc., 142, 4357, https://doi.org/10.1002/qj.2626.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kieu, C. Q., and Z. Moon, 2016: Hurricane intensity predictability. Bull. Amer. Meteor. Soc., 97, 18471857, https://doi.org/10.1175/BAMS-D-15-00168.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Leidner, S. M., B. Annane, B. McNoldy, R. Hoffman, and R. Atlas, 2018: Variational analysis of simulated ocean surface winds from the Cyclone Global Navigation Satellite System (CYGNSS) and evaluation using a regional OSSE. J. Atmos. Oceanic Technol., 35, 15711584, https://doi.org/10.1175/JTECH-D-17-0136.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, J., J. Li, C. Velden, P. Wang, T. J. Schmit, and J. Sippel, 2020: Impact of rapid-scan-based dynamical information from GOES-16 on HWRF hurricane forecasts. J. Geophys. Res. Atmos., 125, e2019JD031647, https://doi.org/10.1029/2019JD031647.

    • Search Google Scholar
    • Export Citation
  • Lim, A. H. N., J. A. Jung, S. E. Nebuda, J. M. Daniels, W. Bresky, M. Tong, and V. Tallapragada, 2019: Tropical cyclone forecasts impact assessment from the assimilation of hourly visible, shortwave, and clear-air water vapor atmospheric motion vectors in HWRF. Wea. Forecasting, 34, 177198, https://doi.org/10.1175/WAF-D-18-0072.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, Q., and Coauthors, 2020: Vortex initialization in the NCEP operational hurricane models. Atmosphere, 11, 968, https://doi.org/10.3390/atmos11090968.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lu, X., X. Wang, Y. Li, M. Tong, and X. Ma, 2017: GSI-based ensemble-variational hybrid data assimilation for HWRF for hurricane initialization and prediction: Impact of various error covariances for airborne radar observation assimilation. Quart. J. Roy. Meteor. Soc., 143, 223239, https://doi.org/10.1002/qj.2914.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McNoldy, B. D., B. Annane, S. Majumdar, J. Delgado, L. Bucci, and R. Atlas, 2017: Impact of assimilating CYGNSS data on tropical cyclone analyses and forecasts in a regional OSSE framework. Mar. Technol. Soc. J., 51, 715, https://doi.org/10.4031/MTSJ.51.1.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nolan, D. S., R. Atlas, K. T. Bhatia, and L. R. Bucci, 2013: Development and validation of a hurricane nature run using the joint OSSE nature run and the WRF model. J. Adv. Model. Earth Syst., 5, 382405, https://doi.org/10.1002/jame.20031.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nystrom, R. G., and F. Zhang, 2019: Practical uncertainties in the limited predictability of the record-breaking intensification of Hurricane Patricia (2015). Mon. Wea. Rev., 147, 35353556, https://doi.org/10.1175/MWR-D-18-0450.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Poterjoy, J., and F. Zhang, 2011: Dynamics and structure of forecast error covariance in the core of a developing hurricane. J. Atmos. Sci., 68, 15861606, https://doi.org/10.1175/2011JAS3681.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Poterjoy, J., F. Zhang, and Y. Weng, 2014: The effects of sampling errors on the EnKF assimilation of inner-core hurricane observations. Mon. Wea. Rev., 142, 16091630, https://doi.org/10.1175/MWR-D-13-00305.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pu, Z., S. Zhang, M. Tong, and V. Tallapragada, 2016: Influence of the self-consistent regional ensemble background error covariance on hurricane inner-core data assimilation with the GSI-based hybrid system for HWRF. J. Atmos. Sci., 73, 49114925, https://doi.org/10.1175/JAS-D-16-0017.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reale, O., J. Terry, M. Masutani, E. Andersson, L. P. Riishojgaard, and J. C. Jusem, 2007: Preliminary evaluation of the European Centre for Medium-Range Weather Forecasts’ (ECMWF) nature run over the tropical Atlantic and African monsoon region. Geophys. Res. Lett., 34, L22810, https://doi.org/10.1029/2007GL031640.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ruf, C. S., and Coauthors, 2016: New ocean winds satellite mission to probe hurricanes and tropical convection. Bull. Amer. Meteor. Soc., 97, 385395, https://doi.org/10.1175/BAMS-D-14-00218.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ryan, K., L. Bucci, J. Delgado, R. Atlas, and S. Murillo, 2019: Impact of Gulfstream-IV dropsondes on tropical cyclone prediction in a regional OSSE system. Mon. Wea. Rev., 147, 29612977, https://doi.org/10.1175/MWR-D-18-0157.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sears, J., and C. S. Velden, 2012: Validation of satellite-derived atmospheric motion vectors and analyses around tropical disturbances. J. Appl. Meteor. Climatol., 51, 18231834, https://doi.org/10.1175/JAMC-D-12-024.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tong, M., and Coauthors, 2018: Impact of assimilating aircraft reconnaissance observations on tropical cyclone initialization and prediction using operational HWRF and GSI ensemble–variational hybrid data assimilation. Mon. Wea. Rev., 146, 41554177, https://doi.org/10.1175/MWR-D-17-0380.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Torn, R. D., 2014: The impact of targeted dropwindsonde observations on tropical cyclone intensity forecasts of four weak systems during PREDICT. Mon. Wea. Rev., 142, 28602878, https://doi.org/10.1175/MWR-D-13-00284.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Torn, R. D., and G. J. Hakim, 2009: Ensemble data assimilation applied to RAINEX observations of Hurricane Katrina (2005). Mon. Wea. Rev., 137, 28172829, https://doi.org/10.1175/2009MWR2656.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Velden, C., W. E. Lewis, W. Bresky, D. Stettner, J. Daniels, and S. Wanzong, 2017: Assimilation of high-resolution satellite-derived atmospheric motion vectors: Impact on HWRF forecasts of tropical cyclone track and intensity. Mon. Wea. Rev., 145, 11071125, https://doi.org/10.1175/MWR-D-16-0229.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Weng, Y., and F. Zhang, 2016: Advances in convection-permitting tropical cyclone analysis and prediction through EnKF assimilation of reconnaissance aircraft observations. J. Meteor. Soc. Japan, 94, 345358, https://doi.org/10.2151/jmsj.2016-018.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Whitaker, J. S., and T. M. Hamill, 2002: Ensemble data assimilation without perturbed observations. Mon. Wea. Rev., 130, 19131924, https://doi.org/10.1175/1520-0493(2002)130<1913:EDAWPO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wu, T., H. Liu, S. J. Majumdar, C. S. Velden, and J. L. Anderson, 2014: Influence of assimilating satellite-derived atmospheric motion vector observations on numerical analyses and forecasts of tropical cyclone track and intensity. Mon. Wea. Rev., 142, 4971, https://doi.org/10.1175/MWR-D-13-00023.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zawislak, J., and E. J. Zipser, 2014: Analysis of the thermodynamic properties of developing and nondeveloping tropical disturbances using a comprehensive dropsonde dataset. Mon. Wea. Rev., 142, 12501264, https://doi.org/10.1175/MWR-D-13-00253.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, F., and Y. Weng, 2015: Predicting hurricane intensity and associated hazards: A five-year real-time forecast experiment with assimilation of airborne Doppler radar observations. Bull. Amer. Meteor. Soc., 96, 2533, https://doi.org/10.1175/BAMS-D-13-00231.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, J. A., D. S. Nolan, R. F. Rogers, and V. Tallapragada, 2015: Evaluating the impact of improvements in the boundary layer parameterization on hurricane intensity and structure forecasts in HWRF. Mon. Wea. Rev., 143, 31363155, https://doi.org/10.1175/MWR-D-14-00339.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, S., Z. Pu, D. J. Posselt, and R. Atlas, 2017: Impact of CYGNSS ocean surface wind speeds on numerical simulations of a hurricane in observing system simulation experiments. J. Atmos. Oceanic Technol., 34, 375383, https://doi.org/10.1175/JTECH-D-16-0144.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fig. 1.

    Hurricane regional OSSE domain configuration. The track of the hurricane is colored by intensity (kt) and is labeled at 0000 UTC of each date. The black outline represents the 27-km parent domain of the HNR1. The gray boundaries represent the HWRF parent domain (9 km) and an instance of the HWRF storm-following nest (3 km).

  • Fig. 2.

    (a) The volumetrically averaged RMS errors (solid lines) and bias (dashed lines) for the (a) u component of the wind (m s−1) and (b) temperature (°C) on the HWRF parent domain. Shown are the background (filled circles) and analyses (open circles) for each cycle for the Wind+Mass (purple), Wind Only (blue), and Mass Only (green) experiments. The period of rapid intensification is highlighted in orange.

  • Fig. 3.

    The azimuthally averaged (left) tangential and (right) radial winds for the (a),(b) HNR1; (c),(d) Wind+Mass; (e),(f) Wind Only; and (g),(h) Mass Only experiments.

  • Fig. 4.

    Differences in the azimuthally averaged (left) tangential and (right) radial winds for 0600 UTC 3 Aug between the HNR1 and the (a),(b) Wind+Mass; (c),(d) Wind Only; and (e),(f) Mass Only experiments.

  • Fig. 5.

    The 34-kt contours of wind speed at the surface (1000 hPa) for the (top) background and (bottom) analysis from the 0600 UTC 3 Aug cycle. HNR1 (black) is compared with all 30 ensemble members for the (a),(b) Wind+Mass (purple); (c),(d) Wind Only (blue); and (e),(f) Mass Only (green) experiments.

  • Fig. 6.

    Homogeneous (same number of cases per forecast lead time) average (a) intensity error (m s−1), (b) total integrated kinetic energy error (TJ), and (d) error for the radius of 34-kt wind (n mi) over the 120-h forecast period for the Wind+Mass (purple), Wind Only (blue), and Mass Only (green) experiments. Also shown is the (c) normalized distribution of 10-m surface wind speeds at the 96-h forecast from the 0000 UTC 3 Aug cycle for the HNR1 1-km nest (black) and 3-km nest (gray) in addition to the three previously listed experiments.

  • Fig. 7.

    Average track error (km) over the 120-h forecast period for the Wind+Mass (purple), Wind Only (blue), and Mass Only (green) experiments.

  • Fig. 8.

    (a) The 24-period of wind profiler observation locations. Colors indicate the center time of the assimilation window: Red is 0000 UTC, yellow is 0600 UTC, purple is 1200 UTC, and green is 1800 UTC. The HNR1 track is color coded in the same colors. (b) Sample wind profile coverage from the 0600 UTC 3 Aug assimilation cycle window. Colors indicate distance from the center of the storm center. Red is more than 500 km, purple is 300–500 km, pink is 200–300, orange is 100–200, and gray is within 100 km of the center.

  • Fig. 9.

    As in Fig. 2, but for the Wind Only (blue), Swath (light blue), Swath6 (light green), and Swath12 (light purple) experiments. Also shown is the (c) total number of observations assimilated within 300 km of the TC center for each of the Swath experiments.

  • Fig. 10.

    Azimuthally averaged tangential wind (m s−1) for (left) HNR1, (center) model background, and (right) analysis. Plots are the cycles in which inner-core observations were present for the Swath experiment at (a)–(c) 0600 UTC 3 Aug, (e)–(g) Swath6 experiment at 1200 UTC 3 Aug, and (h)–(j) Swath12 experiment at 1800 UTC 3 Aug. Blue contour lines are the radii of 34-, 50-, and 64-kt winds representing the gale force, damaging, and destructive wind locations, respectively.

  • Fig. 11.

    As in Fig. 5, but for the (a),(b) Swath (blues); (c),(d) Swath6 (greens); and (e),(f) Swath12 (purples) experiments.

  • Fig. 12.

    Homogeneous average (a) intensity error (m s−1), (b) track error (km), (c) total integrated kinetic energy error (TJ), and (d) error for the radius of 34-kt wind (n mi) over the 120-h forecast period for the Swath (light blue), Swath6 (light green), and Swath12 (light purple) experiments.

  • Fig. 13.

    As in Fig. 2a, but for the noTC100 (orange), noTC200 (pink), noTC300 (purple), noTC500 (maroon), Swath (light blue), and Wind Only (blue) experiments

  • Fig. 14.

    Azimuthally averaged tangential wind (m s−1) at 0600 UTC 3 Aug for (a) HNR1, and the analyses from the (b) Swath, (c) noTC100, (d) noTC200, (e) noTC300, and (f) noTC500 experiments.

  • Fig. 15.

    As in Fig. 5, but for the (a),(b) noTC100 (oranges); (c),(d) noTC300 (dark purples); and (e),(f) noTC500 (reds) experiments.

  • Fig. 16.

    As in Fig. 12, but for the noTC100 (orange), noTC300 (dark purple), and noTC500 (maroon) experiments.