Abstract

This study analyzes 25 years of Special Sensor Microwave Imager (SSM/I) retrievals of rain rate and wind speed to assess changes in storminess over the open water of the Pacific Ocean. Changes in storminess are characterized by combining trends in both the statistically derived 95th percentile exceedance frequencies of rain rate and wind speed (i.e., extremes). Storminess is computed annually and seasonally, with further partitioning done by phase of the El Niño–Southern Oscillation (ENSO) index and the Pacific decadal oscillation (PDO) index. Overall, rain-rate exceedance frequencies of 6–8 mm h−1 cover most of the western and central tropical Pacific, with higher values present around the Philippines, Japan, Mexico, and the northwest coast of Australia. Wind speed exceedance frequencies are a strong function of latitude, with values less (greater) than 12 m s−1 equatorward (poleward) of 30°N/S. Statistically significant increasing trends in rain rate were found in the western tropical Pacific near the Caroline Islands and the Solomon Islands, and in the extratropics from the Aleutian Islands down the coast along British Columbia and Washington State. Statistically significant increasing trends in wind speed are present in the equatorial central Pacific near Kiribati and the Republic of the Marshall Islands (RMI), and in the extratropics along the west coast of the United States and Canada. Thus, while extreme rain and winds are both increasing across large areas of the Pacific, these areas are modulated according to the phase of ENSO and the PDO, and their intersection takes aim at specific locations.

1. Introduction

The Pacific Ocean is vast, covering more than 165 million km2, encompasses many nations consisting of small island atolls, and is host to a myriad of plant and animal life. By comparison, the combined landmass of the small islands makes up roughly 527 000 km2, or much less than 1% of the total area of the Pacific Ocean. The weather and climate of the Pacific region has a direct impact on these islands, impacts that are expected to gradually worsen with time as a result of climate change. As stated in the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (AR4; Parry et al. 2007, 7–22) small islands will be especially susceptible to accelerated beach erosion, increased damage to infrastructure from overtopping waves and storms, and changes in rainfall patterns affecting freshwater availability. Furthermore, sea surface heights and sea surface temperatures are expected to continue to increase, drawing elevated concerns over coral bleaching and sea level rise (Parry et al. 2007, 7–22; Keener et al. 2012). Recently, a U.S. Geological Survey report found that the Hawaiian atolls of Midway and Laysan may become inundated and uninhabitable during this century (Pyper 2013). The report also found that the Republic of the Marshall Islands (RMI) and the Federated States of Micronesia (FSM) face similar threats.

The concept of “storminess” is often used as a proxy for the patterns and trends of storm frequency, intensity, and impacts from high seas, strong winds, and heavy rains (Marra 2008; Marra et al. 2008; Marra et al. 2010; Kruk et al. 2013). Each one of these elements can occur either independently, such as long-period swell from an extratropical cyclone, or together, such as from a typhoon passing overhead. In this manuscript, the extremes of wind and rain are assessed both as individual and grouped impact-based metrics—storminess. The distributions of extremes in rainfall and wind speed are modulated by the phases of El Niño–Southern Oscillation (ENSO) and the Pacific decadal oscillation (PDO). For example, the location of the South Pacific convergence zone (SPCZ) shifts northeast or southwest of its climatological position depending on El Niño or La Niña [Kruk et al. (2014) and references therein]. Given the lack of land-based data across the open waters of the Pacific Ocean, can we assess changes in storminess using remotely sensed data, and if so, how are these changes influenced by the background states of ENSO and PDO?

Using land-based station data from 1981 to 2011, McGree et al. (2013) substantiated a strong relationship between the phase of ENSO and both total and extreme rainfall. McGree et al. (2013) found spatially heterogeneous and largely not statistically significant trends in rainfall for 1961–2011. They infer that a shift in the phase of the interdecadal Pacific oscillation around 1999 largely reversed earlier trends. Gu and Adler (2013) also find evidence for a climate regime shift around 1998/99. ENSO and PDO are different ocean–atmosphere teleconnections that are not completely independent of one another. This manuscript will show that wind and rain extremes encompassing storminess depend on the phases of both phenomena.

Assessing storminess over the ocean requires observations of rain and wind, which are traditionally measured by buoys or other similar floating platforms. However, there remain an inadequate number of buoys/platforms to perform a detailed climate analysis over the vast Pacific Ocean. Moreover, while there are many island-based data networks for rainfall and wind, they are typically spaced hundreds of kilometers apart, offering a relatively incomplete assessment of storminess over the region. Fortunately, passive microwave remote sensing provides a physically based technique for retrieving wind and rain over the ocean. The Special Sensor Microwave Imager (SSM/I) and Special Sensor Microwave Imager Sounder (SSMIS) operated by the Defense Meteorological Satellite Program (DMSP) provide the longest continuous record of passive microwave measurements of the earth. The measurements began in 1987 and continue to this day, with one more satellite launch scheduled in 2016. Microwave brightness temperatures at 19, 22, and 37 GHz are used to retrieve wind, vapor, cloud, and rain over the ocean. These frequencies are able to penetrate liquid and ice clouds, as well as atmospheric dust and smoke, to see the ocean surface. Dual-polarization (vertical and horizontal) measurements at 19 and 37 GHz allow accurate retrieval of wind speed in rain-free conditions. In rain, they are used to separate the emission and scattering signals of precipitation. Since wind speed estimates are not available in rain, nearby rain-free observations are used as a local proxy. For the time period of this study, SSM/I and SSMIS data come from nine sensors, each operating on its own spacecraft. These sensors have been intercalibrated and are suitable for development of a long time series of these variables (Wentz and Schabel 2000; Wentz et al. 2007; Wentz 2013).

Many studies have used SSM/I to develop climatological means of the parameters made available by the satellite (e.g., Ferraro et al. 1996; Chang et al. 1995, Wentz and Spencer 1998). Some of these past studies have been applied on a global scale (e.g., Young et al. 2012); though others like Allan et al. (2010) have focused only on the tropical regions of the globe. Young (1994) examined oceanic-scale wave statistics using remote sensing satellite sensors. Young (1999) constructed a global climatology of ocean wave and wind conditions finding that the Southern Ocean was the “roughest ocean on Earth.” Young and Holland (1996) created an atlas of ocean wave height conditions using the Geosat satellite and focused globally on international trade routes. Allan et al. (2010) found that the interannual mean variability in the SSM/I precipitation data over the tropical ocean can be explained by daily rainfall events above the 94th percentile. Similar strong proportional linkage between extreme rainfall and total rainfall is seen in Greene et al. (2007). Extreme value return periods using altimeter data were also developed (Alves and Young 2003; Chen et al. 2012; Panchang et al. 1998; Vinoth and Young 2011). Young et al. (2012) investigated trends in extreme values of wave heights and wind speeds over the entire globe. They found that 100-yr return intervals for the 99th percentile extreme wind event across the Pacific varied considerably, from 50 m s−1 near the Aleutian Islands, to only 12–15 m s−1 in the equatorial Pacific, and increasing again to 44–48 m s−1 south of New Zealand (Young et al. 2012). In addition, Kruk et al. (2013), in concert with the Pacific Storms Climatology Products (PSCP) suite, also defined extreme events as those values falling above the 95th percentile, but whose analysis was extended to encompass wind speed and direction, sea height, and waves. The primary goal of the PSCP project was to identify changes in storminess across the Pacific, where the essence of storminess was defined by patterns and trends of storm frequency and intensity from heavy rainfall, strong winds, and high seas. Each of these variables was obtained from in situ sites—both from land surface stations and buoy platforms.

A recent study by Kruk et al. (2014) focused on one element of storminess—–heavy rains—and documented both the primary atmospheric drivers of and the historical trends in rainfall extremes and related indices in the western and northern Pacific. While that study found mostly increasing trends in extreme rainfall between the subbasins of the Pacific Ocean, it also omitted the vast expanse of open water due to a lack of surface-based observing systems. However, given the widespread availability of the SSM/I data, its unique ability to collect subcloud layer data, and the supporting evidence lending confidence in the ability for the SSM/I to accurately detect extremes, this study aims to complement Kruk et al. (2014) by expanding the analysis to the overwater portion of the Pacific basin. The spatial size and geographic location of storminess and how it is modulated by the phase of the ENSO and the PDO are also discussed.

2. Methodology

Surface wind speed and rain-rate retrievals were gathered over the Pacific Ocean from 50°S to 60°N and from 110°E to 100°W from Remote Sensing Systems (RSS) version 7 SSM/I intercalibrated ocean retrievals. The version-7 SSM/I data include improvements to the radiative transfer model (Meissner and Wentz 2012) and the antenna temperature calibration (Wentz 2013). Previous versions of SSM/I have established its usage for studying climate trends and variability (Wentz et al. 2007; Trenberth et al. 2005; Wentz and Schabel 2000). The algorithm for retrieving wind speed is described in Wentz (1997) with an update by Wentz and Meissner (2000). The algorithm for retrieving rain is described in Wentz and Spencer (1998) with an update by Hilburn and Wentz (2008). SSM/I rain rates (Bowman et al. 2009) and wind speeds (Mears et al. 2001) have been shown to be in excellent agreement with buoys.

The physical basis for wind speed retrievals relies primarily upon the polarization signature. The horizontal polarization brightness temperature increases approximately 1 K for every 1 m s−1, while the change in vertical polarization is much smaller. Rain rate is also retrieved by its polarization signature, which reduces the contrast between vertically and horizontally polarized brightness temperatures on the order of 1 K per 0.2 mm h−1 at 19 GHz. The orthogonality of these brightness temperature changes with environmental parameter changes has been thoroughly studied and the retrieval algorithm has been formulated to minimize potential cross-talk between parameters (Wentz 1997; Wentz and Meissner 2000).

The number of available SSM/I satellites changes over time, but for this study the dataset is restricted to the “core” satellites listed in Table 1. This provides the most consistent sampling over time for the analysis of the temporal variability in extremes. These particular instruments were chosen over others based on their local time of day. They all take measurements near 0600 and 1800 local time and have the least drift in local time of day among the SSM/I data. This is especially important for rain rate, which has a diurnal cycle in the tropics of 16% of the mean (Hilburn and Wentz 2008).

Table 1.

List of all SSM/I satellites and the start date for data usage.

List of all SSM/I satellites and the start date for data usage.
List of all SSM/I satellites and the start date for data usage.

The accuracy of SSM/I wind retrievals degrades in the presence of rain because it lacks the lower-frequency (10 GHz and below) channels needed to see through the rain to the surface (Meissner and Wentz 2009). So, to study both wind and rain trends at a particular location, the SSM/I data must be spatially aggregated to capture neighboring rain-free pixels. The data are provided on a 0.25° × 0.25° latitude–longitude earth grid. A sensitivity study (not shown) indicated that the minimum grid size needed to reliably capture both wind and rain measurements is 1.5°. A slightly larger grid size of 2.5° was used for the benefit of additional noise suppression in spatial plots. One SSM/I sensor makes between 0 and 2 observations per day at a particular location, with a mean of 1.2 observations per day. Satellite swaths precess in longitude, so there is no east–west geographic pattern to the sampling. The sampling does have a north–south pattern, caused by overlapping swaths at latitudes poleward of 60° latitude, which is outside the region of study.

This study uses SSM/I-derived wind speeds; however, other wind speed data are available. For example, the Blended Sea Winds (Zhang et al. 2006a,b) and the Cross-Calibrated MultiPlatform (CCMP) Ocean Surface Wind Vector (Atlas et al. 2011) both provide a long-term record of wind speed over the ocean. Both products incorporate information from reanalysis datasets: Blended Sea Winds includes NCEP Reanalysis 2 and CCMP includes ERA-40. One large advantage of using RSS version 7 data over the Blended Sea Winds is that by using only certain satellites from the SSM/I and SSMIS series, the highest possible level of consistency is achieved by local time of day of the observation and spatiotemporal sampling patterns over the time record.

One concern about using satellite data to study extreme rainfall is that it may significantly underestimate extreme rainfall rates. The satellite rain rates in this study represent an average over a 625-km2 footprint, compared with an R. M. Young tipping-bucket rain gauge with a 200-cm2 catchment area. To evaluate these concerns, 10-min rain gauge data were obtained from Pacific Marine Environmental Laboratory (PMEL). The buoy data were then collocated with DMSP F13 satellite SSM/I and F17 SSMIS data allowing a 60-min maximum time difference between satellite and buoy. Cumulative distribution functions were produced for the collocated buoy and satellite rainfall data when both were nonzero (not shown). This comparison demonstrated that the buoy rain rates have greater fractions of both light and heavy rain compared with the satellite, consistent with the vastly different sample volumes. However, while the satellite data do show a loss of sensitivity at very low rain rates, this study is concerned with the 95th percentile, for which the buoy and satellite data are in excellent agreement. A separate sensitivity analysis using PMEL hourly rain gauge data showed very similar results. To that end, zero rain rates were excluded from the percentiles because the focus of this study is on the most extreme values (95th percentile), which aligns well with previous studies documented in section 1. For wind, a bin size of 0.2 m s−1 was assigned and for rain a bin size of 0.1 mm h−1 was used. From there, the cumulative distribution functions were used to determine the 95th percentiles. Additionally, the analysis of extreme values at the 95th percentile was further subdivided into seasons, where boreal winter is from 1 October to 30 April, and boreal summer is from 1 May to 30 September, consistent with Kruk et al. (2013, 2014). The percentiles were then used to calculate an exceedance frequency, defined as the number of observations in a 2.5° × 2.5° grid cell exceeding the percentile threshold value (Fig. 1) divided by the total number of observations. The exceedance frequencies were multiplied by the number of days per time period to express them in units of days.

Fig. 1.

The 95th percentile thresholds for (a) rain rate and (b) wind speed.

Fig. 1.

The 95th percentile thresholds for (a) rain rate and (b) wind speed.

Another altimeter, called the Archiving, Validation, and Interpretation of Satellite Oceanographic (AVISO), has been in operation since October 1992, but it is not a passive microwave satellite. Rather, the AVISO measures the satellite-to-surface, round-trip time of a radar pulse. The AVISO satellite is particularly useful at gathering information about the sea surface height (SSH), where the sea height is computed as the difference between the satellite height and the altimetric range (http://www.aviso.altimetry.fr/en/home.html). AVISO has been successfully used in the past to estimate sea surface heights (e.g., Merrifield et al. 2009; Young et al. 2012). In these studies, the western equatorial Pacific showed the largest increases in SSH, with positive trends of 5–10 mm yr−1. While not the focus of this manuscript, we did obtain and analyze the AVISO data to independently verify the trends in SSH over the Pacific Ocean. The results (not shown) perfectly mirrored those that were found by Merrifield et al. (2009). Since increasing trends in SSH across the western and northern Pacific have been previously documented using satellite data, all grid cells were assigned a binary “1” to denote increasing trends, which will be used later in conjunction with rain and wind event trends to determine changes in storminess across the domain (section 5).

3. Trend results

The first step toward assessing changes in storminess is to identify trends in the extremes of rain rate and wind speeds, following the definition of extremes from section 2. Trend results are presented by annual, boreal summer, and boreal winter. The resulting trends have spatial patterns that are regional in scale and often vary substantially over small distances. Thus, geographic capital cities are marked in the figures to help identify island-level impacts.

Figure 1a shows the 95th percentile threshold for rain rate and Fig. 1b shows the 95th percentile threshold for wind speed. The heaviest rain-rate thresholds are evident in the darker green shading from Japan and Korea southwest along the Asian continent, and through the west-central Pacific, including many islands nations and atolls such as the RMI, FSM, Fiji, and Tonga. Meanwhile, low thresholds, indicated by the brown shading, are evident across the far southeastern Pacific and along 45°S. There is also an area of reduced values from the west coast of North America westward through the Hawaiian Islands.

For wind speed (Fig. 1b), the highest 95th percentile values are located closer to the poles, generally along and north of 30°N and south of 30°S. The swath of lowest thresholds is aligned between 15°S and 15°N, with more moderate values elsewhere. The absolute lowest wind speed thresholds reside in the far west-central Pacific, including areas such as Koror, Papua New Guinea (PNG), and the eastern Philippines.

a. Annual

Annually, the trend in 95th percentile extreme rain-rate occurrences (Fig. 2a) was increasing across the far northern Pacific, decreasing along the 10th parallel, and increasing again diagonally from PNG to Fiji. Areas with statistically significant decreasing trends, enclosed by solid black lines, were noted across the far eastern Pacific to the southeast of the Hawaiian Islands and off the coast of Baja California. Meanwhile, statistically significant increasing trends are located along the SPCZ through the Solomon Islands, Vanuatu, and Tonga, with a secondary area of increasing trends found from Midway Island northeastward to the coast of British Columbia, Canada. The increasing trends across Micronesia are consistent with the findings from Kruk et al. (2014), which analyzed rainfall extremes from land-based rain gauges across the Pacific and found increasing trends over FSM and the Commonwealth of the Northern Mariana Islands (CNMI). There is a large region of annually increasing high-wind days (Fig. 3a) located along the equator from about 165°E to 135°W, with the area extending farther east in the Southern Hemisphere to about 105°W. The areas with positive high-wind event trends include the southern part of the RMI, Nauru, Kiribati, FSM, and the northern parts of Tuvalu, Tokelau, the Cook Islands, and French Polynesia. Other prominent increases were found along the west coast of North America and the east coast of China. The Hawaiian Islands were another area where high-wind events appear to be increasing on the annual time scale; however, these increases are not statistically significant. Negative trends stretch from Hawaii, through the northern parts of the RMI, across CNMI, Palau, FSM, PNG, Solomon Islands, Fiji, Tonga, Niue, and American Samoa. The strongest declining trends in 95th percentile events are located in RMI, the Solomon Islands, and PNG. The strongest positive trends are in Kiribati with high values extending west into Nauru and parts of the FSM.

Fig. 2.

Trend in rain-rate 95th percentile exceedance frequency by season for (a) annual, (b) summer, and (c) winter. Black contours enclose regions of trends that are statistically significant at the 0.95 level.

Fig. 2.

Trend in rain-rate 95th percentile exceedance frequency by season for (a) annual, (b) summer, and (c) winter. Black contours enclose regions of trends that are statistically significant at the 0.95 level.

Fig. 3.

Trend in wind speed 95th percentile exceedance frequency by season for (a) annual, (b) summer, and (c) winter. Black contours enclose regions of trends that are statistically significant at the 0.95 level.

Fig. 3.

Trend in wind speed 95th percentile exceedance frequency by season for (a) annual, (b) summer, and (c) winter. Black contours enclose regions of trends that are statistically significant at the 0.95 level.

b. Summer

During the boreal summer (Fig. 2b), extreme rain rates show a more muted response across the basin, but also with a few noteworthy areas. Once again the central and eastern Pacific had decreasing trends, with an area of statistically significant decreases in between Hawaii and the western United States. A ribbon of decreasing rain-rate trends was also found along 10°S, from 120°W all the way to 165°E. The Coral Sea also had minor decreasing trends. Meanwhile, near the equator in the western Pacific, statistically significant increasing rain-rate trends were found from PNG into the southern parts of FSM as well. Most other areas were neutral in trends or the trends were small and statistically insignificant.

During the boreal summer (Fig. 3b), the trend in wind speeds at or above the 95th percentile changes only slightly from the annual map. Statistically significant increases on the order of 4–6 days of more extreme wind speeds per decade were found near the equator just east of the date line. This is the same region where Gastineau and Soden (2011) found a bull’s-eye of negative SST correlation with wind speed in the middle of the equatorial Pacific. Statistically significant increases were also found over the waters between southern Australia and western New Zealand, where trends here are two to four windier days per decade. Meanwhile, farther north, statistically significant decreasing trends (2–4 days decade−1) were centered on 40°N and 165°E. Smaller decreases were located southwest of the U.S. coastline and over parts of Palau and Yap in the far western Pacific, but these decreases were not statistically significant at the 95% confidence interval.

c. Winter

The boreal winter trends (Fig. 2c) closely resemble those from the boreal summer season. Statistically significant decreasing rain rates of −0.4 to −0.6 days decade−1 were found over the eastern Pacific, with another stripe of decreasing trends from the RMI southeastward to French Polynesia. Large increasing rain rates of 0.4 days decade−1 were located across much of the Alaskan peninsula, and again between Midway Island and the coast of British Columbia.

The overall trend pattern changes a little more in the boreal winter (Fig. 3c) when the map is dominated by a larger area of decreasing trends in extreme wind speeds. The largest decreases of 4–8 days decade−1, which were also statistically significant, are located over the CNMI, including Guam, Saipan, and Rota. Smaller decreases extend outward from there to include Midway Island, and diagonally from northern PNG through the Solomon Islands and down toward Fiji. Many of these decreasing trends in this region of the Pacific are also statistically significant. Increasing trends were located off the northern coast of western North America, extending westward through the Gulf of Alaska all the way to Japan and Korea. The positive trends of 2–6 days decade−1 off the coast of North America are statistically significant. Another area of increasing trends, though not statistically significant, was found from near the equator and date line southeast through the Samoas into northern French Polynesia.

d. Generalized extreme value theory

The analysis of extreme events is often linked to the anticipation of return intervals, that is, how likely an event is to happen again based on history. The generalized extreme value (GEV) approach is often used to assess return intervals in extreme events, and was done so in this study. For each grid box, the annual maximum value in rain rate and wind speed was obtained. The data were then fit to the Gumbel distribution following the statistical guidance in Coles (2001). The resulting 100-yr return values for rain rates and wind speeds are shown in Figs. 4a and 4b, respectively.

Fig. 4.

GEV 100-yr return values for (a) rain rate and (b) wind speed.

Fig. 4.

GEV 100-yr return values for (a) rain rate and (b) wind speed.

For rain rates (Fig. 4a), the 100-yr values exceeding 30 mm h−1 are confined to the more northern latitudes, near 45°N, between 150°E and 150°W. The lowest rain-rate values of 0–5 mm h−1 for the 100-yr period are found off the coasts of South America and the western United States. Much of the Pacific domain is dominated by the 10–20 mm h−1 signal. A ribbon of higher return values (20–25 mm h−1) spans the southern latitudes, from 15° to 45°S between 150°E and 110°W, inclusive of the Tasman Sea, New Zealand, and the Cook Islands.

Wind speed return values at the 100-yr interval (Fig. 4b) were coincidentally highest in the same areas as those for rain rates, namely the far northern portion of the Pacific basin and again across the southernmost latitudes. The Southern Hemispheric maxima were more constrained to areas due south of Australia and into the Tasman Sea. Not surprisingly, the lowest wind speed values were found along the equator (10–20 m s−1) and especially in the far eastern Pacific, were local minima of 5–10 m s−1 were found. The highest values in the basin were located along 45°N, between 150°E and 150°W, in the same location as that for the highest return values for rain rates in the basin.

4. Teleconnections and trends

An underlying factor in rain event and wind event trends is that they are modulated by seasonal variability and ENSO and PDO teleconnections. Understanding how the trends noted in the previous section are influenced by the background climatology requires documenting the relationship between these trends and the larger-scale teleconnections. Importantly, to avoid aliasing the seasonal variability into estimates of ENSO and PDO variability, the seasonal cycle had to be removed before calculating the exceedance frequency by ENSO and/or PDO phase, where the boreal winter spans from 1 October to 30 April, and boreal summer is from 1 May to 30 September. To do this, maps of the exceedance frequencies for the boreal summer and winter seasons were produced over the period 1988–2012 (not shown). These seasonal climatology maps were subtracted from the exceedance frequency results for both ENSO and PDO teleconnections. For wind speeds, this removed the strong seasonal cycle associated with the greatest number of strong wind events occurring in each respective hemispheric winter. For rain rates, this removed a strong seasonal cycle in extratropical storm tracks and tropical convection. The following subsections document the relationship between trends and the larger-scale teleconnections (with seasonality now being removed).

a. El Niño–Southern Oscillation

Figure 5 shows the distribution of the frequency of extremes in wind and rain events for each phase of ENSO after subtracting out the seasonal normals. The ENSO data were obtained from the National Oceanic and Atmospheric Administration/Climate Prediction Center (NOAA/CPC) for the years 1988–2012. Not surprisingly, the results obtained herein line up well with previous studies on the linkages between seasonal rainfall and the phase of ENSO. Allan and Soden (2008) examined time series of the 90th percentile rain rate over the ocean and found that El Niño events coincided with increased frequency of heaviest precipitation. Curtis et al. (2007) showed that over the tropical oceans, El Niño–Southern Oscillation events result in a spatial redistribution and overall increase in extremes. During La Niña, an increase in the frequency of dry extremes and no change in wet extremes were observed. Because of the juxtaposition of tropical land areas with the ascending branches of the global Walker circulation, El Niño (La Niña) contributes to generally dry (wet) conditions in these land areas. Karnauskas et al. (2008) found that ENSO forces variability in the Papagayo gap winds over the east Pacific warm pool. The connection between convection to winds over the east Pacific warm pool has also been studied with satellite data by Maloney and Esbensen (2007). It is encouraging to see that the SSM/I data have captured this signal change in rainfall rate and are discussed in the following subsections.

Fig. 5.

The 95th percentile exceedance frequency by ENSO phase and deseasonalized for (a) rain rate, cool phase; (b) rain rate, neutral phase; (c) rain rate, warm phase; (d) wind speed, cool phase; (e) wind speed, neutral phase; and (f) wind speed, warm phase.

Fig. 5.

The 95th percentile exceedance frequency by ENSO phase and deseasonalized for (a) rain rate, cool phase; (b) rain rate, neutral phase; (c) rain rate, warm phase; (d) wind speed, cool phase; (e) wind speed, neutral phase; and (f) wind speed, warm phase.

1) Cool phase (La Niña)

During the cold phase of ENSO (i.e., La Niña; Fig. 5a), sea surface temperature anomalies within the Niño-3.4 region are defined to be less than −0.5°C according to the CPC. In this phase, a reduction in the frequency of extreme rainfall days was evident from along the equator eastward to the coast of South America. The largest decreases were found across the central and western Pacific, including Kiribati and most of Micronesia. Increases in extreme rainfall days were noted north of 30°N, extending from the coastal waters of British Columbia southwest to the international date line. A narrow ribbon of higher frequencies of extremes during La Niña was also noted from the Cook Islands westward through Vanuatu and into PNG.

Much of the Pacific experiences stronger than average wind speeds during La Niña (Fig. 5d). Across a large area of the central Pacific, the frequency of extreme winds is nearly twice what would be expected in a typical year. This area is centered across the equator and the international date line and includes the islands of Kiribati, Tuvalu, and much of the RMI. Additional but smaller increases were found over the Hawaiian Islands, most of the South China Sea, and northward toward Japan and the Gulf of Alaska. Meanwhile, less windy days were found mainly in the Southern Hemisphere and the far eastern equatorial Pacific, from New Zealand east to the coast of South America. Along the equator east of 120°W the frequency of extreme winds is about half what it would be in a typical year.

2) Neutral phase

During the neutral phase of ENSO, which is defined as sea surface temperature anomalies within the Niño-3.4 region falling between −0.5° and +0.5°C, the distribution of extreme rainfall event counts shows a more muted response than either phase of ENSO. The axes of extreme rainfall days (Fig. 5b) extend diagonally from Honolulu southwest through the RMI and into the northern Solomon Islands (+0.1 to +0.3 days yr−1). A second area of increasing heavy rain events is located along the alignment of the SPCZ from the equator diagonally to far southern South America at 45°S. Fewer extreme rainfall days (−0.3 to −0.5 days yr−1) were located along the equator, with another area of drier conditions (−0.1 to −0.3 days yr−1) from the coast of British Columbia southwest through Guam and into the South China Sea.

The distribution of extreme wind events shows a remarkably different pattern than either the warm or cold phases of ENSO (Fig. 5e). During the neutral phase, wind speed extremes are most likely to occur along and south of 30°S and between 15° and 30°N, with reduced frequencies north of 30°N in the Northern Hemisphere. Specifically, much of the Solomon Islands sees more occurrences of high-wind events during neutral ENSO phases (+1 to +5 days yr−1), along with parts of Tuvalu and in the Tasman Sea. To the north, fewer extremes were noted in the Gulf of Alaska, the Philippine Sea, and over the islands of Kiribati (−1 to −5 days yr−1). Over the Hawaiian Islands, parts of CNMI, and the far northern RMI, these areas experience one to three more windy days, while Samoa and the Cook Islands experience one to three fewer windy days during ENSO neutral conditions.

3) Warm phase (El Niño)

The warm phase of ENSO is characterized by sea surface temperatures in the Niño-3.4 region above +0.5°C. The magnitude of the distribution of extreme rainfall days appears to be more emphasized or exaggerated than the other phases (Fig. 5c). Specifically, a dry tongue became most pronounced from western PNG southeastward through northern Australia, Vanuatu, and the southern Cook Islands. Additional drier conditions were noted over the Gulf of Alaska southward through the Hawaiian Islands. Much of northern Australia, southeastward into the Tasman Sea, has historically seen many fewer heavy rain days during El Niño. Meanwhile, an incredible frequency of extreme rain events (greater than 1 day yr−1) was located along the equator, from west to east, and southward into Tuvalu, Kiribati, and the northern Cook Islands. Other areas of higher exceedances were found along the west coast of the United States and over far western Micronesia and CNMI.

Unlike the neutral phase of ENSO, the warm phase features an enhancement of the frequencies in extreme wind events (Fig. 5f), with most of the windier days occurring north of 23.5°N or south of 23.5°S. The most extreme wind speed days during El Niño were found between 30° and 45°N and 165°E and 120°W, just north of the Hawaiian Islands. A secondary maximum was also found over the far western Pacific, including FSM, CNMI, PNG, and the Solomon Islands. A third area of increasing wind speed days was located between 30° and 45°S, including areas around Tasmania, New Zealand, Tonga, and the southern Cook Islands. Minima were found over the Hawaiian Islands (3–5 fewer days yr−1) and along and just south of the equator and 165°W, inclusive of parts of Kiribati (7–11 fewer windy days yr−1).

b. Pacific decadal oscillation

The Pacific decadal oscillation is characterized using the PDO index produced by the Joint Institute for the Study of the Atmosphere and Ocean (JISAO). The index is derived as the leading principle component of monthly SST anomalies in the North Pacific Ocean (Zhang et al. 1997). The monthly values are used without additional smoothing, and are separated into two phases: positive or negative. The positive phase is associated with warm SST anomalies across the equatorial Pacific and northward along the west coast of North America, while the negative phase is the inverse pattern (warm western Pacific waters, cold eastern Pacific waters).

When the phase of the PDO becomes negative (Fig. 6a), trends are highest generally north of the equator across much of the central and western Pacific up to and including the Gulf of Alaska. In these regions, heavy rain extremes are increasing from +0.3 to +0.5 days yr−1. Meanwhile, off the coast of California and east of Hawaii, extreme rain events are decreasing by −0.1 to −0.3 days yr−1. The largest area of lower rainfall rates (dry) runs across the date line, and along the equator, with a secondary area of lower rainfall rates from Samoa southeast through the northern Cook Islands and into French Polynesia. Most of these areas are shaded with decreasing trends on the order of −0.3 to −0.5 days yr−1.

Fig. 6.

The 95th percentile exceedance frequency by PDO phase and deseasonalized for (a) rain rate, negative phase; (b) rain rate, positive phase; (c) wind speed, negative phase; and (d) wind speed, positive phase.

Fig. 6.

The 95th percentile exceedance frequency by PDO phase and deseasonalized for (a) rain rate, negative phase; (b) rain rate, positive phase; (c) wind speed, negative phase; and (d) wind speed, positive phase.

Heavy rain event trends during a positive PDO (Fig. 6b) are significantly different compared to those of a negative PDO. During a positive PDO, dry anomalies stretch from the Gulf of Alaska southward through the northern RMI and westward to CNMI. A secondary area of decreasing rainfall extreme events is exhibited from PNG southeast through Port Villa and the far southern Cook Islands. Meanwhile, increases in rainfall extreme event days are found along the west coast of the United States and along the equator from the Solomon Islands eastward through the Gilbert Islands and the northern Cook Islands.

However, the distribution of trends in wind speed extremes in a negative PDO (Fig. 6c) is nearly the inverse of its counterpart during a positive PDO (Fig. 6d). During a positive PDO phase, occurrences of extreme wind events are highest along and south of the equator to roughly 15°S. A secondary area of increased frequency of windy days is found from around the Hawaiian Islands west toward Guam and CNMI. Reduced frequencies of windy days are found just north of the equator inclusive of Kiribati and Micronesia southward through the Solomon Islands. Meanwhile, during a negative PDO (Fig. 6c), increasing trends in extreme wind events (five to seven more days per year) are noted from just south of the equator extending southeast toward South America, with another area of increasing trends in the Gulf of Alaska. At the same time, a negative PDO phase decreases the trends in windy days across all of Micronesia, much of PNG, the Solomon Islands, and Fiji, with another belt of decreasing frequencies between 30° and 45°N.

5. Discussion

To assess changes in storminess across the domain, a simple storminess index was developed. Using traffic light colors of red, yellow, and green, storminess maps were determined as follows:

  • Green = (rain trend < 0) AND (wind trend < 0).

  • Yellow = (rain trend > 0) OR (wind trend > 0).

  • Red = (rain trend > 0) AND (wind trend > 0).

Figure 7a shows the annual changes in storminess across the Pacific, while Figs. 7b and 7c show the changes in boreal winter and summer, respectively. Later in this section, the analysis is further subdivided by phase of ENSO and PDO.

Fig. 7.

Storminess index by season for (a) annual, (b) summer, and (c) winter. Red indicates “high risk,” yellow indicates “medium risk,” and green indicates “low risk.”

Fig. 7.

Storminess index by season for (a) annual, (b) summer, and (c) winter. Red indicates “high risk,” yellow indicates “medium risk,” and green indicates “low risk.”

Recall from section 2 that the trends in sea surface heights are increasing in each of the cities listed in Table 2. As such, it is the lowest common denominator in each of the stoplight categories; the inclusion of sea surface height trends does not significantly affect the overall result of the storminess index. The benefit of choosing a tricolor map is that it gives a quick “first look” at those areas that are essentially most “at risk” because of the recent trends in extremes of heavy rains and strong winds. Areas that are shaded in red have both increasing trends in the extremes of rain and wind, indicating that storminess is also increasing, while green-shaded areas could be considered relatively “low risk,” since both trends in heavy rains and strong winds are decreasing (only sea surface heights are increasing).

Table 2.

Trend summary by capital and island/island nation.

Trend summary by capital and island/island nation.
Trend summary by capital and island/island nation.

For the annual series (Fig. 7a) the red shading includes the grid boxes representing the RMI, parts of the Gulf of Alaska, the waters near Japan, and southwestern PNG to the northern coast of Australia. The area near New Zealand, especially the southern portions, is also highlighted in red. During the boreal summer months, the map looks similar, with the red-shaded storminess index continuing in the Gulf of Alaska, the waters of far northern Japan, and much of PNG and Malaysia. The RMI capital of Majuro also remains in the red during boreal summer, as does Nuku’alofa in Tonga (cf. Table 2). There was some relaxation of the red shading at 60°N and 180°, as well as across the coast of British Columbia. When the season shifts to the boreal winter (Fig. 7b), the region with the most storminess shifts to the polar latitudes above 45°N and below 45°S. The exception is the northern coast of Australia, through the grid box inclusive of Port Moresby, and along 5°–10°N from RMI east to 140°W.

Yellow areas have at least one increasing trend, either in heavy rains or strong winds, indicating a “moderate risk” of increasing trends in storminess. The yellow area encompasses the majority of the Pacific basin on the annual time scale and during the boreal summer season. Yellow grid boxes include parts of FSM, Hawaii, the Samoas, Fiji, and along the southwest coast of the United States.

It is worth noting that Table 2 provides a component-based assessment of the contributions from extreme winds and rains toward the overall storminess index. While the red areas in Fig. 7 highlight the overlap of increasing trends in both extremes of winds and rains, Table 2 shows specifically how much of an increase can be attributed to each component. For example, for Majuro, Marshall Islands, it is clear that its “red” storminess index assessment largely stems from the contribution due to extreme winds, rather than extreme rain events, on the summer, winter, and annual scales. However, such storminess assessments are subject to the current state of ENSO or the PDO, which can act to enhance or weaken extremes in wind and rain. The ensuing discussion examines the effect of both teleconnections on the storminess index across the region.

a. Storminess and teleconnections

Since the PDO is based on the first principle component of an empirical orthogonal function, it is not surprising that ENSO and PDO were found to be positively correlated, with a correlation coefficient r = 0.4456. However there are periods where the indices differ. For example, in 1991, 1995, and 2012, positive ENSO occurred with negative PDO. In 1988, 1996, and 2006, negative ENSO occurred with positive PDO.

Exceedance frequencies for heavy rain and wind events were broken down by ENSO and PDO phase (not shown). For heavy rain events during warm ENSO and positive PDO, the pattern resembles warm ENSO, with a strong signal in the central Pacific. However, for rain events under a warm ENSO and negative PDO regime, the wet anomaly is located farther west in the Pacific, while during both cold and neutral ENSO phases, the exceedance frequencies during negative PDO are not significantly different than those during a positive PDO. For wind events during the warm phase of ENSO and positive PDO, the signal resembles a warm ENSO phase (Fig. 5f). For the case of a warm ENSO and negative PDO, there is a large negative wind anomaly in the central Pacific and a positive wind anomaly along the Pacific Northwest coastline. For cold ENSO and negative PDO, the pattern resembles cold ENSO (Fig. 5d). For cold ENSO and positive PDO, there is a strong positive wind anomaly in the western Pacific Northern Hemisphere extratropics.

Keeping in mind the question posed in the introduction of this manuscript (Can we assess changes in storminess using remotely sensed data, and if so, how are these changes influenced by the background states of ENSO and PDO?), a storminess index map was produced for the overlap of each phase of ENSO and the PDO (Fig. 8). The left column in Fig. 8 represents positive PDO and the right column corresponds to negative PDO. The top row is cool ENSO, followed by neutral ENSO with the last row associated with warm ENSO. Figure 8 shows several key changes in the placement of the storminess axis (red areas) depending upon the overlapping phase of ENSO and the PDO, though it seems more dependent upon the state of ENSO than it does PDO. For example, the areas under the “worst” storminess in a cool ENSO (Figs. 8a,d) appear to be heavily dependent on the phase of the PDO. During a cool ENSO and positive PDO (Fig. 8a), higher risk areas include the north-central Pacific, parts of far western Micronesia, and much of PNG, Fiji, Samoa, and Tonga. Yet, during a cool ENSO with a negative PDO (Fig. 8d), much of those same areas are now shaded green, with the higher risk locations shifting to north of 45°N, west of 120°E, and south of 15°S, essentially the perimeter of the Pacific. Meanwhile, during ENSO neutral conditions (Figs. 8b,e), the storminess index is less dependent on the crossover with the phase of the PDO. Many areas shaded in red in Fig. 8b, including Honolulu, Hawaii, are also shaded in red in Fig. 8e, though the negative PDO seems to temper some of the extremes across the basin during a neutral ENSO phase. Finally, during warm ENSO and positive PDO (Fig. 8c), the overlap of high extremes in wind and rain are found along the west coast of the United States, along the equator through Kiribati, through parts of CNMI, and southeastward along the SPCZ from the Solomon Islands, through the Samoas, and into the Cook Islands. At the same time, areas near Alaska, New Zealand, PNG, and Hawaii are shaded green during warm ENSO and positive PDO. Yet, when the phase of the PDO shifts to negative during a warm ENSO, much of the green areas faded into yellow (“moderate risk”), but also the higher-risk red-shaded areas dissipated somewhat as well. Following Fig. 8, periods when cool ENSO overlaps with positive PDO (Fig. 8a), and when warm ENSO overlaps with negative PDO (Fig. 8f), give rise to the most elevated risk of storminess across the Pacific. Thus, the storminess maps are a reflection of the interactions between ENSO and PDO teleconnection states.

Fig. 8.

Storminess index by combined phases of ENSO and PDO for (a) cool ENSO, positive PDO; (b) neutral ENSO, positive PDO; (c) warm ENSO, positive PDO; (d) cool ENSO, negative PDO; (e) neutral ENSO, negative PDO; and (f) warm ENSO, negative PDO. Red is “high risk,” yellow “medium risk,” and green “low risk.”

Fig. 8.

Storminess index by combined phases of ENSO and PDO for (a) cool ENSO, positive PDO; (b) neutral ENSO, positive PDO; (c) warm ENSO, positive PDO; (d) cool ENSO, negative PDO; (e) neutral ENSO, negative PDO; and (f) warm ENSO, negative PDO. Red is “high risk,” yellow “medium risk,” and green “low risk.”

b. Tropical cyclones

The linear trend in the number of tropical storms per year over 1988–2012 was computed based on data from the International Best Track Archive for Climate Stewardship (IBTrACS) dataset (Knapp et al. 2010) to determine if the storminess results were influenced by historical tropical cyclone activity. In the east Pacific, around Guam, and the Coral Sea, the tropical storm frequency over the past 25 yr is decreasing in concert with decreasing rain-rate and wind speed trends. Over the Timor Sea, Arafura Sea, and Gulf of Carpentaria, the tropical storm frequency is increasing, and the rain and wind trends are both increasing. Over the Philippine Sea and around Niue, the tropical storm frequency is increasing, and while the rain trends are increasing, the wind trends are mixed. Increasing rain over FSM does not correspond to changes in tropical storm frequency. The increasing trend in wind speed in the equatorial central Pacific does not correspond to changes in tropical storm frequency. Thus, changes in tropical cyclone frequency do manifest themselves in storminess changes, especially rain, on basin-wide scales. However, the large-scale pattern in storminess cannot be explained by the frequency and intensity of tropical cyclones alone. The discussion of changes to tropical cyclone behavior is best dealt with from other sources in the literature (e.g., Knutson et al. 2010).

6. Summary

In this manuscript, 25 yr (1988–2012) of Special Sensor Microwave Imager (SSM/I) data were analyzed to assess trends in rain and wind events over the Pacific Ocean. SSM/I rain-rate and wind speed retrievals were analyzed on an annual and seasonal basis, with the boreal summer season extending from May to September, and the boreal winter season from October to April. The concept of “storminess” was used to capture the unique combination of impacts caused by high seas, strong winds, and heavy rains. Many of these factors are expected to worsen in a changing climate, that is, even higher seas, stronger winds, and heavier extreme rain events (Parry et al. 2007, 7–22; Keener et al. 2012). To that end, this study answers the question, given the lack of land-based data across the open waters of the Pacific Ocean, can we assess changes in storminess using remotely sensed data, and if so, how are these changes influenced by the background states of ENSO and PDO?

The influence of ENSO on the frequency and trend in rain and wind events over the ocean was examined by binning the data according to ENSO phase (cool, warm, and neutral). For high-wind events, results indicated that much of the Pacific sees stronger than average wind speeds during La Niña. In addition, the results also showed a strong drying signal across the eastern North Pacific, from California south toward Mexico and westward through the Hawaiian Islands and into parts of the Northern Mariana Islands including Guam during La Niña. However, during El Niño, an increase in frequency of extreme rain events was found along the equator, from west to east, and southward into the Solomon Islands, Tuvalu, Kiribati, and the Cook Islands. Additionally, the warm phase of ENSO featured a nearly exact opposite of the frequencies in extreme high-wind events, with most of the windier days occurring in the Northern Hemisphere, with many fewer windy days in the Southern Hemisphere (Fig. 5).

Though we recognize the period of record is only 25 years (due to the SSM/I launching in 1987), a statistical generalized extreme value (GEV) analysis was completed to derive 100-yr return period values for rain rate and wind speed (Fig. 4). Much of the Pacific domain is dominated by the 10–20 mm h−1 rain-rate distribution, while the highest values for wind speed return values are located along 45°N, between 150°E and 150°W, generally in the same location as that for the highest return values for rain rates in the basin.

Finally, a storminess index was created and mapped (Figs. 7 and 8). The storminess map gives a quick “first look” at those areas that are essentially most “at risk” as a result of the recent trends in heavy rains, high seas, and strong winds. The areas where the overlap of increasing trends in both wind speed and rain rates were shaded as red, indicating the “higher” risk potential. In all seasons, both PNG and RMI were highlighted in red (Fig. 7). Meanwhile, yellow areas were designated when at least one trend was increasing and are here labeled as “moderate” risk locations. The yellow area encompasses the majority of the Pacific basin on the annual time scale and during the boreal summer season. Green areas, or “low” risk locations, were highlighted when both trends in heavy rain and high-wind events were decreasing. The most green shading occurs during boreal winter, with a large area of lower risk depicted in Hawaii, CNMI, northern FSM, the Samoas, Fiji, and Vanuatu. For interested user groups, such as the decision-maker, water manager, or agroforestry planner, given a predicted state of ENSO and PDO, Fig. 8 can be used as a tool to assess seasonal risk and inform their resilience and adaptation planning strategies.

There are several areas of future work related to assessing trends in rain and wind events across the Pacific. A logical next step would be to examine climate model projections, such as those from the fifth phase of the Coupled Model Intercomparison Project (CMIP5) and determine what a storminess index and trends in rain and wind events may look like in 2050 or 2100. In addition, an explorative study into the impact of a downscaling technique on the SSM/I data would be prudent to determine the feasibility of obtaining more specific-island-based results.

Finally, not all impacts are created equal, and the storminess index presented here does not address the social, economic, or humanitarian impacts from increasing inundation events, drought occurrences and associated windy days, or too much rainfall and increasing tropical cyclone activity. To that end, an idealistic future study should incorporate satellite data, and specifically SSM/I data, in conjunction with a digital elevation model (DEM), wave heights, and predominant wind directions, to address specific impacts on the island/atoll level to inform emergency managers, water managers, and other decision-makers.

Acknowledgments

The authors thank members of the Remote Sensing Applications Division of the National Climatic Data Center and staff at the Remote Sensing Systems office for their help in identifying available microwave imagery products. We also thank Michael Palecki and two anonymous reviewers for their insightful guidance, which helped improve this manuscript.

REFERENCES

REFERENCES
Allan
,
R. P.
, and
B. J.
Soden
,
2008
:
Atmospheric warming and the amplification of precipitation extremes
.
Science
,
321
,
1481
1484
, doi:.
Allan
,
R. P.
,
B. J.
Soden
,
V. O.
John
,
W.
Ingram
, and
P.
Good
,
2010
: Current changes in tropical precipitation. Environ. Res. Lett.,5, 025205, doi:.
Alves
,
J. H. G. M.
, and
I. R.
Young
,
2003
:
On estimating extreme wave heights using combined Geosat, Topex/Poseidon and ERS-1 altimeter data
.
Appl. Ocean Res.
,
25
, 167–186, doi:.
Atlas
,
R.
,
R. N.
Hoffman
,
J.
Ardizzone
,
S. M.
Leidner
,
J. C.
Jusem
,
D. K.
Smith
, and
D.
Gombos
,
2011
:
A cross-calibrated, multiplatform ocean surface wind velocity product for meteorological and oceanographic applications
.
Bull. Amer. Meteor. Soc.
,
92
,
157
174
, doi:.
Bowman
,
K. P.
,
C. R.
Homeyer
, and
D. G.
Stone
,
2009
:
A comparison of oceanic precipitation estimates in the tropics and subtropics
.
J. Appl. Meteor. Climatol.
,
48
,
1335
1344
, doi:.
Chang
,
A. T. C.
,
L. S.
Chiu
, and
G.
Yang
,
1995
:
Diurnal cycle of oceanic precipitation from SSM/I data
.
Mon. Wea. Rev.
,
123
,
3371
3380
, doi:.
Chen
,
C.-A.
,
C.
Chou
, and
C.-T.
Chen
,
2012
:
Regional perspective on mechanisms for tropical precipitation frequency and intensity under global warming
.
J. Climate
,
25
,
8487
8501
, doi:.
Coles
,
S.
,
2001
: An Introduction to Statistical Modeling of Extreme Values. Springer-Verlag, 208 pp.
Curtis
,
S.
,
A.
Salahuddin
,
R. F.
Adler
,
G. J.
Huffman
,
G.
Gu
, and
Y.
Hong
,
2007
:
Precipitation extremes estimated by GPCP and TRMM: ENSO relationships
.
J. Hydrometeor.
,
8
,
678
689
, doi:.
Ferraro
,
R. R.
,
F.
Weng
,
N. C.
Grody
, and
A.
Basist
,
1996
:
An eight-year (1987–1994) time series of rainfall, clouds, water vapor, snow cover, and sea ice derived from SSM/I measurements
.
Bull. Amer. Meteor. Soc.
,
77
,
891
905
, doi:.
Gastineau
,
G.
, and
B. J.
Soden
,
2011
:
Evidence for a weakening of tropical surface wind extremes in response to atmospheric warming
.
Geophys. Res. Lett.
,
38
,
L09706
, doi:.
Greene
,
J. S.
,
B.
Paris
, and
M.
Morrissey
,
2007
:
Historical changes in extreme precipitation events in the tropical Pacific region
.
Climate Res.
,
34
,
1
14
, doi:.
Gu
,
G.
, and
R. F.
Adler
,
2013
:
Interdecadal variability/long-term changes in global precipitation patterns during the past three decades: Global warming and/or Pacific decadal variability
.
Climate Dyn.
, 40, 3009–3022, doi:.
Hilburn
,
K. A.
, and
F. J.
Wentz
,
2008
:
Intercalibrated passive microwave rain products from the unified microwave ocean retrieval algorithm (UMORA)
.
J. Appl. Meteor. Climatol.
,
47
,
778
794
, doi:.
Karnauskas
,
K. B.
,
A. J.
Busalacchi
, and
R.
Murtugudde
,
2008
:
Low-frequency variability and remote forcing of gap winds over the east Pacific warm pool
.
J. Climate
,
21
,
4901
4918
, doi:.
Keener
,
V. W.
,
J. J.
Marra
,
M. L.
Finucane
,
D.
Spooner
, and
M. H.
Smith
, Eds.,
2012
: Climate Change and Pacific Islands: Indicators and Impacts. Report for the 2012 Pacific Islands Regional Climate Assessment. National Climate Assessment Regional Technical Input Report Series, Island Press, 170 pp.
Knapp
,
K. R.
,
M. C.
Kruk
,
D. H.
Levinson
,
H. J.
Diamond
, and
C. J.
Neumann
,
2010
:
The International Best Track Archive for Climate Stewardship (IBTrACS): Unifying tropical cyclone best track data
.
Bull. Amer. Meteor. Soc.
,
91
,
363
376
, doi:.
Knutson
,
T. R.
, and Coauthors
,
2010
:
Tropical cyclones and climate change
.
Nat. Geosci.
,
3
,
157
163
, doi:.
Kruk
,
M. C.
,
J. J.
Marra
,
P.
Ruggiero
,
D.
Atkinson
,
M.
Merrifield
,
D.
Levinson
, and
M.
Lander
,
2013
:
Pacific Storms Climatology Products (PSCP): Understanding extreme events
.
Bull. Amer. Meteor. Soc.
,
94
,
13
18
, doi:.
Kruk
,
M. C.
,
A.
Lorrey
,
G.
Griffiths
,
M.
Lander
,
E. J.
Gibney
,
H. J.
Diamond
, and
J. J.
Marra
,
2014
:
On the state of the knowledge of rainfall extremes in the western and northern Pacific basin
.
Int. J. Climatol.
, 35, 321–336, doi:.
Maloney
,
E. D.
, and
S. K.
Esbensen
,
2007
:
Satellite and buoy observations of boreal summer intraseasonal variability in the tropical northeast Pacific
.
Mon. Wea. Rev.
,
135
,
3
19
, doi:.
Marra
,
J. J.
,
2008
: Pacific Region Integrated Climatology Information Products (PRICIP) derived-data products. Eos, Trans. Amer. Geophys. Union,89 (Fall Meeting Suppl.), Abstract A53E-0324.
Marra
,
J. J.
,
U.
Kari
, and
T. A.
Sabbatelli
,
2008
: Anatomies of historical storm events in the Pacific. Proc. Solutions to Coastal Disasters, Oahu, HI, American Society of Civil Engineers, 222–228, doi:.
Marra
,
J. J.
,
E.
Shea
,
U.
Kari
, and
E.
Wong
,
2010
: Pacific Region Integrated Climatology Information Products (PRICIP). Pacific Disaster Net, accessed 29 May 2015. [Available online at http://www.pacificdisaster.net/pdnadmin/data/original/NOAA_2010_PAC_Intergrated_climatology_info.pdf.]
McGree
,
S.
, and Coauthors
,
2013
:
An updated assessment of trends and variability in total and extreme rainfall in the western Pacific
.
Int. J. Climatol.
, 34, 2775–2791, doi:.
Mears
,
C. A.
,
D. K.
Smith
, and
F. J.
Wentz
,
2001
:
Comparison of Special Sensor Microwave Imager and buoy-measured wind speeds from 1987 to 1997
.
J. Geophys. Res.
,
106
,
11 719
11 729
, doi:.
Meissner
,
T.
, and
F. J.
Wentz
,
2009
:
Wind vector retrievals under rain with passive satellite microwave radiometers
.
IEEE Trans. Geosci. Remote Sens.
,
47
,
3065
3083
, doi:.
Meissner
,
T.
, and
F. J.
Wentz
,
2012
:
The emissivity of the ocean surface between 6 and 90 GHz over a large range of wind speeds and Earth incidence angles
.
IEEE Trans. Geosci. Remote Sens.
,
50
,
3004
3026
, doi:.
Merrifield
,
M. A.
,
S. T.
Merrifield
, and
G. T.
Mitchum
,
2009
: An anomalous recent acceleration of global sea level rise. J. Climate,22, 5772–5781, doi:.
Panchang
,
V.
,
L.
Zhao
, and
Z.
Demirbilek
,
1998
: Estimation of extreme wave heights using GEOSAT measurements. Ocean Eng.,26, 205–225, doi:.
Parry
,
M. L.
,
O. F.
Canziani
,
J. P.
Palutikof
,
P. J.
van der Linden
, and
C. E.
Hanson
, Eds.,
2007
: Climate Change 2007: Impacts, Adaptation and Vulnerability. Cambridge University Press, 976 pp.
Pyper
,
J.
,
2013
:
Storm surges, rising seas could doom Pacific islands this century
.
Sci. Amer.
, 12 April. [Available online at https://www.scientificamerican.com/article/storm-surges-rising-seas-could-doom-pacific-islands-this-century/.]
Stephen
,
M.
,
2011
: On Nauru, a sinking feeling. New York Times, 18 July 2011.
Trenberth
,
K. E.
,
J.
Fasullo
, and
L.
Smith
,
2005
:
Trends and variability in column-integrated atmospheric water vapor
.
Climate Dyn.
,
24
,
741
758
, doi:.
Vinoth
,
J.
, and
I. R.
Young
,
2011
: Global estimates of extreme wind speed and wave height. J. Climate,24, 1647–1665, doi:.
Wentz
,
F. J.
,
1997
:
A well calibrated ocean algorithm for special sensor microwave/imager
.
J. Geophys. Res.
,
102
,
8703
8718
, doi:.
Wentz
,
F. J.
,
2013
: SSM/I cersion-7 calibration report. RSS Rep. 011012, Remote Sensing Systems, Santa Rosa, CA, 46 pp. [Available online at http://images.remss.com/papers/tech_reports/2012_Wentz_011012_Version-7_SSMI_Calibration.pdf.]
Wentz
,
F. J.
, and
R. W.
Spencer
,
1998
:
SSM/I rain retrievals within a unified all-weather ocean algorithm
.
J. Atmos. Sci.
,
55
,
1613
1627
, doi:.
Wentz
,
F. J.
, and
T.
Meissner
,
2000
: AMSR ocean algorithm, version 2. RSS Rep. 121599A-1, Remote Sensing Systems, Santa Rosa, CA, 66 pp. [Available online at www.remss.com/papers/rsstech/2000_121599A-1_Wentz_AMSR_Ocean_ATBD_V2.pdf.]
Wentz
,
F. J.
, and
M. C.
Schabel
,
2000
:
Precise climate monitoring using complementary satellite data sets
.
Nature
,
403
,
414
416
, doi:.
Wentz
,
F. J.
,
L.
Ricciardulli
,
K. A.
Hilburn
, and
C. A.
Mears
,
2007
:
How much more rain will global warming bring?
Science
,
317
,
233
235
, doi:.
Young
,
I. R.
,
1994
:
Global ocean wave statistics obtained from satellite observations
.
Appl. Ocean Res.
,
16
,
235
248
, doi:.
Young
,
I. R.
,
1999
:
Seasonal variability of the global ocean wind and wave climate
.
Int. J. Climatol.
,
19
,
931
950
, doi:.
Young
,
I. R.
, and
G. J.
Holland
,
1996
: Atlas of the Oceans: Wind and Wave Climate. Pergamon Press, 241 pp.
Young
,
I. R.
,
J.
Vinoth
,
S.
Zieger
, and
A. V.
Babanin
,
2012
:
Investigation of trends in extreme value wave height and wind speed
.
J. Geophys. Res.
,
117
, C00J06, doi:.
Zhang
,
H.-M.
,
J. J.
Bates
, and
R. W.
Reynolds
,
2006a
:
Assessment of composite global sampling: Sea surface wind speed
.
Geophys. Res. Lett.
,
33
,
L17714
, doi:.
Zhang
,
H.-M.
,
R. W.
Reynolds
, and
J. J.
Bates
,
2006b
: Blended and gridded high resolution global sea surface wind speed and climatology from multiple satellites: 1987–present. 14th Conf. on Satellite Meteorology and Oceanography, Atlanta, GA, Amer. Meteor. Soc., P2.23. [Available online at https://ams.confex.com/ams/pdfpapers/100004.pdf.]
Zhang
,
Y.
,
J. M.
Wallace
, and
D. S.
Battisti
,
1997
: ENSO-like interdecadal variability: 1900–93. J. Climate,10, 1004–1020, doi:.