Future Changes in Typhoon-Related Precipitation in Eastern Hokkaido

From 16 to 23 August 2016, typhoons T1607, T1609, and T1611 hit eastern Hokkaido in northern Japan and caused heavy rainfall that resulted in severe disasters. To understand future changes in typhoon-related precipitation (TRP) in midlatitude regions, climate change experiments on these three typhoons were conducted using a high-resolution three-dimensional atmosphere– ocean coupled regional model in current and pseudo-global warming (PGW) climates. All PGW simulations projected decreases in precipitation frequency with an increased frequency of strong TRP and decreased frequency of weak TRP in eastern Hokkaido. In the current climate, snow-dominant precipitation systems start to cause precipitation in eastern Hokkaido about 24 hours before landfall. In the PGW climate, increases in convective available potential energy (CAPE) developed tall and intense updrafts and the snow-dominant precipitation systems turned to have more convective property with less snow mixing ratio (QS). Decreased QS reduced precipitation area, although strong precipitation increased or remained almost the same. Only TRP of T1607 increased the amounts before landfall. In contrast, all typhoons projected to increase TRP amount associated with landfall, because in addition to increased CAPE, the PGW typhoon and thereby its circulations intensified, and a large amount of rain was produced in the core region. (Citation: Kanada, S., H. Aiki, K. Tsuboki, and I. Takayabu, 2019: Future changes of typhoon-related precipitation in eastern Hokkaido. SOLA, 15, 244−249, doi:10.2151/sola.2019-044.)


Introduction
Extreme weather events such as tropical cyclones (TCs) and heavy rainfall events have become more violent as the surface air temperature has increased in the current climate Imada et al. 2019). The frequency of extremely intense precipitation has increased in the past several decades (e.g., Trenberth 2011;Fujibe 2015). Furthermore, Kossin et al. (2014) have found that the average latitude where TCs reach their peak intensity has been shifting poleward over the past 30 years and suggested that the past migration in the western North Pacific Ocean (WNP) coincided with increased TC exposure in the regions of the East China Sea, including Japan (Kossin et al. 2016).
By the late 21st century, sea surface temperature (SST) in the vicinity of Japan is projected to increase by 4°C under the RCP8.5 scenario . A high SST is a condition that favors intense TCs (e.g., Emanuel 1986). Most climate change studies of TCs based on state-of-the-art global and cloud-resolving models have therefore indicated that the maximum intensity and precipitation rate of TCs will increase in the future climate (e.g., IPCC 2012; Murakami et al. 2012;Kanada et al. 2013;Tsuboki et al. 2015;Yoshida et al. 2017;Kitoh and Endo 2019).
A typhoon (Northwest Pacific TC) transports a large amount of water vapor from the subtropical ocean and often causes heavy rainfall events in Japan (Fujita and Sato 2017;Nayak and Takemi 2019), even when the typhoon is several hundred kilometers away (Galarneau et al. 2010;Kamahori and Arakawa 2018). These results suggest that a large number of people living in mid-to-high latitudes may be exposed to unusually intense TCs and associated precipitation in the future warmer climate.
From 16 to 23 August 2016, a succession of three typhoons, T1607, T1611, and T1609, hit eastern Hokkaido in northern Japan and caused record-breaking heavy rainfalls that caused severe disasters in the region. On the basis of climate change experiments using high-resolution regional models, Kanada et al. (2017a) have shown that significant predecessor rainfall events (PREs) developed about 24 h before T1607 made landfall. However, it is still unclear whether all typhoons will be associated with increased PRE activity in the warming climate.
The purpose of the present study is to understand the impact of global warming on typhoon-related precipitation in midlatitude regions. To achieve that understanding, numerical experiments were conducted of all three of the typhoons that made landfall in eastern Hokkaido in August 2016 in the current and pseudo-global warming (PGW) climates. Eastern Hokkaido is located at 42°N− 46°N latitude, a region that is the northern limit of where typhoons travelling across the NWP directly make landfall. An atmosphereocean coupled regional model with a grid spacing of 0.04° by 0.04° was used to reproduce the inner-core and precipitation systems of the TCs (e.g., Gentry and Lackmann 2010). Here, we focus on precipitation changes in eastern Hokkaido as an example of a midlatitudinal eastern coastal region and explore how and why the amount, frequency, and characteristics of typhoon-related precipitation would change under the given future environmental conditions.

Models and methodology
The model used in the present study is a high-resolution, three-dimensional atmosphere-ocean coupled regional model composed of the Cloud Resolving Storm Simulator version 3.4 (CReSS; Tsuboki and Sakakibara 2002) for the atmospheric part and the Non-Hydrostatic Ocean model for the Earth Simulator (NHOES; Aiki et al. 2006Aiki et al. , 2011 for the oceanic part. The coupled model is referred to as the CReSS-NHOES (Aiki et al. 2015). The horizontal domain of the coupled model spans 132°E−155°E and 25°N−50°N (Fig. 1), and is discretized with a grid spacing of 0.04° by 0.04°.
The climate change simulations were conducted using the same PGW method used by Kanada et al. (2017a,b). First, we performed control simulations in the current climate (the CNTL simulations) of typhoons T1607 (Chanthu), T1609 (Mindulle), and T1611 (Kompasu). Initial and lateral boundary conditions for the atmosphere and ocean models were provided by the Japan Meteorological Agency 55-year Reanalysis dataset (Kobayashi et al. 2015) and the Japan Coastal Ocean Predictability Experiment reanalysis product (Miyazawa et al. 2009), respectively. The PGW climate simulations were conducted using the results of climate runs by the MRI-AGCM version 3.2 with a horizontal resolution of 20 km under the RCP8.5 scenario (Mizuta et al. 2012(Mizuta et al. , 2014. More detailed information for the models, data and the PGW method are found in Supplement 1. Following Kanada et al. (2017a), we focused on precipitation in eastern Hokkaido, which we defined as the land east of a straight line between Wakkanai (141.678°E, 45.415°N) and Cape Erimo (143.243°E, 41.925°N) (Fig. 1a). There were 110 Auto- The CNTL simulations captured the temporal evolutions and mean amounts of TRP in eastern Hokkaido associated with T1607, T1609, and T1611 (Figs. 2a, 2b, and 2c). In the PGW climate, the peak amounts of precipitation associated with the typhoon landfalls increased and appearance times of the peaks tended to be delayed by a few hours, respectively, compared with the current climate (Figs. 2d, 2e, and 2f). The delays in their northward movement occurred during TRP remote periods (Table 1). Despite the longer TRP remote period due to the slower movement in the PGW climate, the amount of TRP remote increased only for T1607, whereas the amounts of TRP direct increased for all typhoons ( Table 2).

Future Changes in Typhoon-Related Precipitation in Eastern Hokkaido
The frequency of precipitation, defined as the number of model grid points with hourly precipitation amounts not smaller than 0.5 mm, decreased during the TRP remote period for all the typhoons (Figs. 2g, 2h, and 2i). Figure 2j showed that frequency of weak TRP, defined as hourly precipitation smaller than 20 mm, decreased and those of strong TRP, defined as hourly precipitation not smaller than 20 mm, increased for all typhoons. The ratio of strong TRP frequency to the precipitation frequency increased in all typhoons, although the simulations overestimated the ratio of strong TRP frequency (Fig. 2j). Considerable decreases in the frequency of weak precipitation occurred in TRP remote for all typhoons (Fig. 2k). Kanada et al. (2017a) have attributed the increase in PRE in the PGW climate to the increase in convective available potential energy (CAPE) due to increases in water vapor in the low levels from the southern sea. We therefore investigated the temporal evolution of CAPE for air masses transported to eastern Hokkaido from the southern sea (Fig. 3). In the current climate, all typhoon had larger CAPE during TRP remote period, compared with those during TRP direct period. Under the PGW conditions, the CAPE increased for all typhoons. The maximum CAPE exceeded 1000 J kg −1 during TRP remote period for T1607 and T1611 in the PGW climate. The maximum CAPE approached to 2000 J kg −1 during TRP remote period for T1609.
Figures 3d, 3e, 3f, 3g, 3h, and 3i showed the temporal evolution of vertical profile of 99.9th percentile of vertical winds. Vertical winds of 3070 grid points over the eastern Hokkaido in the simulations were used. In the current climate, intense updrafts of 6−7 m s −1 appeared during TRP remote period. Under the PGW conditions with increased CAPE, taller and more intense updrafts developed for all typhoons; the maximum updrafts exceeded mated Meteorological Data Acquisition System (AMeDAS) observation stations in this region, and it contained 3070 grid points in the simulations. In the present study, typhoon-related precipitation (TRP) was defined as precipitation during the period that the storm center was located between 32°N and 46°N. The TRP was divided into two categories: precipitation when the center was located between 32°N and 40°N (TRP remote ) and between 40°N and 46°N (TRP direct ). TRP remote and TRP direct were assumed to be precipitation that occurred during the approach and landfall of the storm, respectively. The times when the center traveled over 32°N, 40°N, and 46°N were defined as it 32 , it 40 , and it 46 , respectively (Table 1).

Results
The CNTL simulations represented well the tracks and minimum central pressures (MCPs) of T1607, T1609 and T1611 compared with the Regional Specialized Meteorological Center Tokyo (RSMC) best-track dataset, although the MCP for T1609 was higher in the simulation than in the best-track dataset (Table  1 and Fig. 1). Under PGW conditions, the simulated MCPs of all the typhoons decreased. Furthermore, the PGW typhoons tended to travel northward at slower translation speeds than they did in the CNTL simulations. The weakening of the jet streak due to the reduction of baroclinicity around northern Japan (Kanada et al. 2017a;Ito et al. 2016) was a possible factor for the slower translation speeds.
The temporal evolutions of area-mean precipitation observed in eastern Hokkaido were examined (Fig. 2). The peak amounts  For all PGW typhoons, CAPE increased and tall and intense updrafts developed as mentioned in Kanada et al. (2017). However, the TRP remote amounts for T1611 and T1609 showed no increase in the PGW climate, while the TRP direct amounts increased for all typhoons (Table 2).
To understand changes in precipitation systems under the PGW conditions, the temporal evolutions of 50th percentile (the medians) of vertical winds, mean relative humidity (RH), and mixing ratios of snow (QS), graupel (QG), and rain (QR) were investigated (Fig. 4). The medians were made of 3070 grid points over the eastern Hokkaido in the simulations. The positive median values indicate that updrafts are predominant. Model grid points with QS not smaller than 0.05 kg kg −1 were defined as snowcoverage grids (Figs. 4a, 4b, 4c, 4g, 4h, and 4i). The temporal evolutions of vertically integrated QS (IQS), QG (IQG), and QR (IQR) are also shown in Fig. 5.
In the current climate, regions with QS larger than 0.1 g kg −1 spread widely over eastern Hokkaido for all typhoons (Figs. 4d, 4e, 4f, 4j, 4k, and 4l). IQS was twice larger than IQG for most of periods (Figs. 5a, 5b, and 5c). For T1607, more than half of eastern Hokkaido was covered by the snow-coverage grids (Fig. 4a). Two regions with updrafts (> 5 cm s −1 ) and downdrafts appeared above and below the 0°C isothermal level around 5 km until 0400 UTC on 17 August (Fig. 4a). The regions with updrafts on the order of centimeters per second corresponded to regions with QS larger than 0.1 g kg −1 . Widespread snow-coverage regions and a pair of weak updrafts and downdrafts below are typical of the stratiform region of organized precipitation systems (Houze 1989a, b).
In the PGW climate, snow-coverage grids during TRP remote periods decreased for all typhoons (Figs. 4a,4b,4c,4g,4h,and 4i); IQS during the periods also decreased for all typhoons (Fig.  5). For T1607, large amounts of QG were produced in the thick layers between altitudes of 5 km and 10 km from 2000 UTC on 16 to 1100 UTC on 17 August 2016 in the PGW climate (Fig. 4j).   Table 1. (j) Precipitation frequency smaller than 20 mm (Weak) and not smaller than 20 mm (Strong) (bars; left axis) and ratio of strong precipitation frequency to total precipitation frequency (crosses: AMeDAS, open circles: models; right axis) for T1607, T1611, and T1609. (k) Changes in precipitation frequency in TRP remote for weak precipitation (remoteW) and strong precipitation (remoteS), and in TRP direct for weak precipitation (directW) and strong precipitation (directS).  During the period, IQG became comparable to IQS (Fig. 5) and produced large amounts of precipitation, although the precipitation areas decreased in the PGW simulation (Figs. 2d and 2g). Decreases in the weak precipitation frequency and precipitation areas in TRP remote could be attributed to the reduction of QS under the PGW conditions.
On the other hand, snow-coverage grids remained during TRP direct period in the PGW climate for all typhoons. In the PGW climate, all typhoons intensified (Table 1) and thereby the typhoon circulations of the core regions intensified. In addition to increased water vapor in the low levels (e.g. Kanada et al. 2017a) under the PGW conditions, the intensified typhoon circulations produced large amount of IQS, IQG, and IQR in the typhoon core regions and led increases in TRP direct amounts (Fig. 5).

Discussion: The potential reasons for decreases in QS
According to Houze (1989a, b), stratiform precipitation systems with widespread QS cause relatively weak precipitation over wide areas, whereas convective systems with large amounts of QG and less QS cause strong precipitation. In the present study, updrafts and QS were predominant above the melting level during TRP remote periods in the current climate (Figs. 4a,4b,4c,4d,4e,and 4f). In the PGW climate, tall and intense updrafts developed and ratios of IQG to IQS increased during TRP remote periods. The results indicated that the snow-dominant precipitation systems in the current climate turned to have more convective property with vigorous updrafts with less QS. Indeed, snow-coverage grids during TRP remote periods decreased for all typhoons in the PGW (Figs. 4a,4b,4c,4g,4h,and 4i).
Another potential cause for the QS reduction is stabilization of atmospheric conditions in the future climate (Hibino et al. 2018).
Most studies have pointed out that the increase in air temperature is large in the upper troposphere (e.g., Hill and Lackmann 2011). In the present study, the 0°C isothermal level and the top of snowcoverage grids increased by 1 km in the PGW climate (Fig. 4). Because increases in air temperature exceeds 4°C above an altitude of 7 km (Kanada et al. 2017b), convection and formation of QS above the altitudes could be suppressed under the warming conditions. During TRP direct period, however, the increase in water vapor in the lower level could overcome the enhanced stability as suggested by Watanabe et al. (2019).

Summary
Future changes of typhoon-related precipitation (TRP) in eastern Hokkaido were investigated based on the results of climate change experiments of three typhoons that hit the region in August of 2016: T1607, T1609, and T1611. The current and pseudo-global warming (PGW) climate experiments were conducted by using a high-resolution, three-dimensional, atmosphere-ocean coupled regional model with a horizontal resolution of approximately 4 km.
All PGW simulations projected decreases in TRP frequency with an increased frequency of strong TRP and decreased frequency of weak TRP in eastern Hokkaido (Fig. 2j). The results indicated that strong precipitation concentrated in the small precipitation area. Considerable decreases in weak precipitation frequency and increases in ratio of strong precipitation frequency to the total precipitation frequency were found in TRP remote for all typhoons (Fig. 2k).
Changes in precipitation systems under the PGW conditions were investigated based on the simulation results. The precipitation systems for TRP remote in the current climate tend to show wide snow-coverage and amounts of QS is about twice larger than those of QG (Figs. 4 and 5). In the PGW simulations, the snow-dominant precipitation systems for TRP remote turned to have more convective property with intense updrafts and less QS. Decreases in QS resulted in decreases in the frequency of weak precipitation. Since the frequency of strong precipitation increased or remained almost the same, ratio of strong precipitation frequency to the precipitation frequency increased in the PGW simulations.
On the other hand, during landfall, in addition to the increased CAPE, all typhoons intensified and thereby the intensified typhoon circulations produced large amounts of rain in the core region (Fig. 5).
It should be noted that the enhancement of PRE (TRP remote ) suggested by Kanada et al. (2017a) only occurred in T1607. The simulations in the present study also overestimated the ratio of strong TRP frequency. Improvements of models and studies using a large ensemble simulation with higher-resolution models should be required in future work. Furthermore, no significant change was found in the translation speed of the typhoons in Watanabe et al. (2019), although the analysis region of Watanabe et al. (2019) differed from the present study. To study changes in track and frequency of tropical cyclones (TCs), a large ensemble of climate simulations with global circulation models, such as d4PDF , will be necessary to reduce the uncertainty.

Supplement
Supplement 1 describes detailed information for the models and methodology.