1. Introduction
Heavy precipitation often causes severe natural disasters such as floods or landslides that can have considerable socioeconomic consequences. Furthermore, heavy precipitation can be produced and enhanced by orographic effects. Houze (2012) classified orographic precipitation by several mechanisms in Fig. 3 of his review with complex interplay—those are the dynamic response of convective systems such as diurnal heating, stability of the upward current, and wind shear; the shape of terrain causing blocking or unblocking of the current; microphysical processes with topography; and so on. This study explores shallow orographic rainfall causing warm-rain processes enhanced by low-level orographic lifting.
Recently, satellite rainfall products that combine high-resolution spatial (from 0.l° × 0.1° to 0.25° × 0.25° latitude–longitude) and temporal (0.5–3 h) data from microwave radiometers (MWRs) in low-Earth orbit and infrared radiometers have become available. Additionally, the great success of the Tropical Rainfall Measuring Mission (TRMM) has accelerated the development of rain retrieval algorithms such as the Goddard profiling algorithm (GPROF; Kummerow et al. 2015) and the Global Satellite Mapping of Precipitation (GSMaP) algorithm (Aonashi et al. 2009). The MWR algorithms used for estimating the rainfall rate over land are based on the scattering effect in the high-frequency channels. This is due to vertically integrated solid hydrometeors above the freezing level. However, the estimates of rainfall from such procedures tend to be underestimated over mountainous regions (e.g., Kwon et al. 2008; Kubota et al. 2009; Sohn et al. 2010, 2013). This underestimation is due to the occurrence of shallow but heavy precipitation related to warm-rain processes as a possible cause.
To mitigate the severe underestimation of orographic rainfall, an orographic/nonorographic rainfall classification scheme has been developed (Shige et al. 2013, 2015, hereafter S13 and S15; Taniguchi et al. 2013, hereafter T13). This scheme has been implemented in the GSMaP algorithm, version 6 (V6), for MWRs (Yamamoto and Shige 2015, hereafter Y15) since September 2014. This scheme introduces orographically forced upward motion w and moisture flux convergence Q at the surface using objective analysis data to detect the area of orographic rainfall conditions. When the conditions of orographic rainfall are met, a lookup table (LUT), the relationship between rain rate and brightness temperatures (Tbs) estimated a priori by a radiative transfer model, is switched from the standard rain type to the orographic one. Then, the surface rainfall rate is estimated using the observed Tbs. This scheme improves estimates of rainfall over the entire Asian region; however, problems of misdetection and overestimation remain. Some of the factors that contribute to these problems include 1) false alarm of orographic rainfall conditions, which can be subdivided as overdetection in moderate thresholds and unsatisfied assumptions of the overall conditions; 2) overestimation of orographic rainfall, whereby estimates of rainfall reach the specified upper limit; and 3) the application of a single vertical-profile model, because of sampling problems, even though a variety of orographic rainfall scenarios could be expected.
Houze (2012) summarized the response of upstream airflow to topography as the strength of the cross-barrier upstream wind component, the degree of thermodynamic stability, and the height of the mountain barrier. Moreover, the characteristics of the airflow over the terrain, in conjunction with the terrain size and microphysical time scales, determine whether the precipitation will ultimately fall on the windward or the leeward slope of the mountain. Based on observations, Yu and Cheng (2008) found that the degree of downstream shift of the heaviest precipitation depends on the intensity of the oncoming flow and the mountain width. Shige and Kummerow (2016) focused on the thermodynamic characteristics of the atmospheric environment associated with shallow orographic heavy rainfall and mentioned the effect of wind shear on orographic rainfall (e.g., Yu and Cheng 2014). However, the current scheme assumes only the case in which warm-rain processes are enhanced by low-level orographic lifting of maritime air (S13), irrespective of various environmental circumstances described above.
The objective of this study was to improve the orographic rainfall retrievals for the GSMaP MWR algorithm. This study represents GSMaP estimates from the TRMM Microwave Imager (TMI) for three cases in a shallow coastal heavy orographic rainfall referred to in the previous studies, in comparison with rainfall amounts simultaneously observed by the TRMM Precipitation Radar (PR). These may be rare cases to be captured even for long-term (17 years) TRMM observations because of their narrow swaths. Although the Global Precipitation Measurement (GPM) core satellite, launched as a successor to the TRMM in 2014 (Hou et al. 2014; Skofronick-Jackson et al. 2017), carries the GPM Microwave Imager (GMI) and the Dual-frequency Precipitation Radar (DPR), there are not enough observation periods and the DPR rainfall retrieval algorithm is in development. Although such a rare case of heavy orographic rainfall contribution is small in terms of seasonal or annual mean rainfall, it is important to quantitatively evaluate those that would cause great disasters if they occur. The orographic/nonorographic classification scheme and the problems of false alarms and overestimation are summarized briefly in section 2. Two major improvements are presented in the following sections. The first is the false alarm of orographic rainfall. A developed conceptual model in the orographic/nonorographic rainfall classification scheme considering the upstream flow that determines the area of orographic enhancement on the windward slopes of the mountains is presented in section 3. The second is the improvement in the estimation of orographic rainfall amount. Validation of the improvement in the detection of orographic rainfall for three cases of heavy rain is also discussed in section 4. The verification scores for the cases and for the Asian region obtained from TMI are compared with TRMM PR data. Section 5 provides the concluding remarks.
2. Orographic/nonorographic rainfall classification scheme
This section briefly introduces the GSMaP MWR algorithm and orographic/nonorographic rainfall classification scheme implemented in the GSMaP V6 algorithm for MWRs by Y15. This algorithm is implemented not only for TMI but for all the MWRs including microwave sounder instruments (Shige et al. 2009; Kida et al. 2017).
The GSMaP MWR algorithm comprises two parts. The first part is a forward calculation that processes LUTs for the relationship between rainfall rate and Tbs by a radiative transfer model (Liu 1998). The radiative transfer model uses inputs of atmospheric field variables provided by the Japan Meteorological Agency (JMA) global analysis (GANAL) data and precipitation-related variables such as the vertical-profile model of precipitation accumulated by TRMM PR data. Takayabu (2008) classified precipitation profiles observed by TRMM PR into six categories for land (severe thunderstorm, afternoon shower, shallow, extratropical frontal systems, organized, and high land) and constructed the precipitation-type database. The convective and stratiform precipitation profiles from the PR data, which are relative to the freezing-level height to exclude the effects of atmospheric temperature variations (Kubota et al. 2007), are averaged trimonthly for each precipitation type in 5° × 5° boxes. The second part is the retrieval part in which the rain rate is estimated based on the LUTs and measured Tbs mainly at 35 and 85 GHz using the rain/no-rain classification method developed by Seto et al. (2005, 2008) for land and that of Kubota et al. (2007) and Mega and Shige (2016) for coasts. The LUTs are assigned depending on the dominant precipitation type, and the convective and the stratiform types are mixed according to weighting from the statistical frequency distribution of PR data.
The vertical-profile model is constructed using the same process as in the original model, except when the conditions of orographic rainfall are satisfied. Orographic precipitation profiles are used over the Indian subcontinent (15°–20°N, 70°–75°E) assumed in S15. In addition, precipitation-size ice particle density for orographic rainfall is set at 0.1 g cm−3 for snow and that for nonorographic rainfall is set at 0.4 g cm−3 for graupel, while the precipitation-size ice particle density was assumed empirically before the GSMaP, version 5 (V5), algorithm. The scheme is switched off for certain regions (e.g., the Sierra Madre in the United States and Mexico) where strong lightning activity occurs in the precipitation-type database because deep convective systems in such regions are involved in the orographic rain conditions. Hereafter, rainfall amounts estimated using the GSMaP V6 MWR algorithms, except when the thresholds of the orographic rain conditions proposed in Eqs. (3) and (4) are satisfied, are referred to as GSMaP0 and GSMaP1, respectively.
Horizontal distributions of the amount of rain derived from GSMaP0 and GSMaP1 are compared with that from the TRMM PR in Figs. 1 and 2, together with an illustration of the orographic rain conditions for GSMaP0 and GSMaP1 to ascertain where the conditions were detected. Each rain rate datum is aggregated over 0.1° × 0.1° to fit the standard GSMaP products provided by JAXA. Figure 1a shows the horizontal distribution of rain from the TRMM PR for the case of heavy rainfall associated with Typhoon Morakot on 8 August 2009 over Taiwan, as presented in T13. It can be seen that high rates of surface rainfall > 25 mm h−1 were detected on the western side of the Central Mountain Range. This heavy rainfall was detected well for GSMaP0 (Figs. 1b,c), as in T13. For GSMaP1, however, the area of the heavy rainfall (Fig. 2a) under the orographic rain condition (Fig. 2b) was detected not only over the Central Mountain Range but also over the western foothills of the range. This misclassification of orographic rain pixels over the foothill was because of the strong wind associated with Typhoon Morakot that led to large values of w and Q in excess of the thresholds.
For Typhoon Morakot (a) horizontal distribution of near-surface rainfall rate from the TRMM PR (orbit number 66 832) and horizontal distributions of (b) surface rainfall rate and (c) flags of the orographic rain condition (red) for GSMaP0 on 8 Aug 2009. (d)–(f) As in (a)–(c), but for the Indian MCS on 30 Jun 2007 (54 821). (g)–(i) As in (a)–(c), but for Typhoon Maemi on 12 Sep 2003 (33 207).
Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0332.1
As in Fig. 1, but for horizontal distributions of near-surface rainfall rate and flags of the orographic rainfall condition for GSMaP1.
Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0332.1
Figures 1d–f and Figs. 2c and 2d show the case of an MCS that occurred over the west coast of India on 30 June 2007, as discussed in S15. Surface rain rates > 30 mm h−1 were detected over the coastal areas at around 18°–19°N in the PR data (Fig. 1d). For GSMaP0, the heavy rainfall around 18°–18.5°N was not detected (Fig. 1e) because the orographic rain condition was not selected because of the conservative thresholds (Fig. 1f). For GSMaP1, the underestimation of the rainfall at around 18°–19°N disappeared (Fig. 2c), although the area of orographic rainfall conditions was more extensive (Fig. 2d). S15 confirmed that the conservative thresholds used by T13 for the heavy rainfall case could not detect the orographic rainfall conditions because the surface wind speed was weaker.
Figures 1g–i and Figs. 2e and 2f show the case of heavy rainfall over the Korean peninsula caused by Typhoon Maemi on 12 September 2003, as presented in Y15. An area of heavy rainfall (>30 mm h−1) was detected over southeastern parts of the Korean peninsula by the TRMM PR (Fig. 1g). Kwon et al. (2008) reported that rain gauge measurements showed not only the heavy rainfall (up to 30 mm h−1) over the south coast but also heavy rain (up to 25 mm h−1) over the east coast. Although GSMaP0 (Fig. 1h) estimated moderate rainfall (15–20 mm h−1) across a wide area, including western inland parts of the peninsula, orographic rainfall was not detected (Fig. 1i) because the amplitude of w was too conservative. When the lenient threshold of w was applied (GSMaP1), an increase in the rainfall amount was found not only in the area of actual heavy rainfall around the southeastern coast but also over western inland parts (Fig. 2e). The flag of the orographic rain condition (Fig. 2f) showed that orographic rainfall pixels extended widely over the eastern coastal and western inland areas of the peninsula.
3. Improvement of orographic rainfall detection
The fundamental concepts of design for the orographic rain conditions are that w may be estimated approximately from the lower boundary conditions of the flow over mountains (Lin 2007) and that warm-rain processes are enhanced by low-level orographic lifting of maritime air (S13). The previous schemes have used values of w and Q derived from near-surface atmospheric data (i.e., the JCDAS and GANAL) such as wind speed, temperature, and relative humidity. Recently, Shige and Kummerow (2016) examined the dynamic and thermodynamic characteristics of the atmospheric environment over tropical coastal mountains in Asia. They set the average wind direction over the target region (the pixel showing shallow orographic rainfall in that place and adjacent pixels) and used the average horizontal wind at a height of <1.5 km over the upstream region in the calculation of w, because low-level jets with a wind speed maximum in the lowest few kilometers of the atmosphere can occur even when surface winds are weak (Stensrud 1996). Indeed, ground data are influenced considerably by surface conditions such as friction. The wind field at the position showing upward motion does not directly represent the cause of orographic rainfall but might be the result of orographic rainfall. Therefore, this study adopts the low-level upstream atmospheric information.
The strength of the upstream current should be taken into consideration. Yu and Cheng (2008) documented the heavy orographic precipitation associated with Typhoon Xangsane (2000) using Doppler radar observation data. The intensity of the upstream oncoming flow modulates not only the precipitation intensity but also the location and structure. The heavy precipitation became deeper and tended to shift downstream with wider area, and the precipitation intensity increased as the low-level oncoming flow intensified. Schematics illustrating the concept of orographic rainfall in relation to weak and strong upstream winds are shown in the upper and middle panels of Fig. 3, respectively. For simplicity, this study considered rainfall enhancement only by the topographically forced upward motion over windward slopes, although several other mechanisms of orographic rainfall might be also active near mountains under different synoptic conditions (Yu et al. 2007; Houze 2012).
Schematics illustrate downward shift of area of orographic enhancement due to (top) strong and (middle) weak low-level upstream flow (gray arrows). Black dotted (black solid) arrows indicate the trajectory of hydrometeors in the weak (strong) oncoming flow. (bottom) The orographically forced upward motion with strong (solid line) and weak (dashed line) upstream flow and the area of orographic rainfall condition for GSMaP0 and GSMaP1 (gray) and real (black) conditions.
Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0332.1
Under a weak upstream current, condensation and the growth of hydrometeors might commence, and then rain reaches the foothills of the windward mountain slope because the advected distance would be shorter, because of sufficient time for precipitation enhancement by the topography. Conversely, under a strong upstream current, the primary region of precipitation enhancement might occur nearer to the mountain peak. This is because the upstream current flows so quickly that there is insufficient time for the enhancement of precipitation over the foothills of the windward mountain slope. The advected distance causing condensation and falling precipitation will be longer. The areas of orographic rainfall conditions of w for weak and strong upstream flows, corresponding to GSMaP0 [Eq. (3)], GSMaP1 [Eq. (4)], and our concept of orographic rainfalls, are shown in the lower panel of Fig. 3. For a weak upstream current, the orographic rain condition for GSMaP0 cannot be detected since w does not reach the threshold while GSMaP1 detects the orographic rain condition properly. On the other hand, a strong upstream current leads to the detection of areas with weak upward motion at the expense of the misclassification of nonorographic rain pixels for GSMaP1 and leads to proper detection of areas for GSMaP0.
Figure 4 shows the relationship between w and wind speed U in the low-level troposphere for the above three cases of heavy orographic rainfall. Pixels of heavy rainfall (>10 mm h−1) from the TRMM PR near-surface rainfall data, averaged over 0.05° × 0.05° grids shown in Fig. 1, can be identified. The values of w and U were calculated from GANAL data, and the low-level troposphere was defined as averaging the wind field up to 2.5 km every 500 m from the surface (i.e., 6 points). The “upstream region” in this study is assumed to be within a distance of 0.5°, chosen to follow the horizontal length scale shown in T13 for simplicity, from the target point. Heavy orographic rainfall frequently occurs under small w (~0.05) for the case of weak upstream currents (<10 m s−1) and favors large w with stronger upstream currents. Some pixels of strong upstream currents but small w correspond to heavy rainfall over the foothills of the mountain associated with Typhoon Morakot rainbands (Fig. 1a) rather than orographic enhancement. These results imply that the value of the threshold should be determined based on the speed of the upstream current.
Orographically forced upward motion vs wind speed in the low-level troposphere for the three cases of heavy orographic rainfall shown in Fig. 1. Black, blue, and red colors correspond to the case of Typhoon Morakot, Indian MCS, and Typhoon Maemi, respectively. The pixels where surface rain rate > 10 mm h−1 observed by the TRMM PR are plotted.
Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0332.1
This section presents the points of improvement for orographic rainfall detection: 1) inclusion of upstream low-level atmospheric information and 2) a variable threshold of w depending on the mean horizontal wind speed.
a. Inclusion of upstream low-level atmospheric information
As presented in the conceptual diagram of orographic rainfall (Fig. 3), the upstream current in the low-level troposphere should also be taken into consideration because the current scheme calculates w and Q using surface wind at the target point. In this study, GSMaP2 is referred to as w and Q derived from upstream current in the low-level troposphere in Eq. (4).
Figure 5 shows the horizontal distributions of w with wind field derived from surface data (GSMaP1) and those derived from the low-level atmosphere (GSMaP2) for the three studied cases. For the case of Typhoon Morakot, although the wind directions over south Taiwan have similar distributions between GSMaP1 (Fig. 5a) and GSMaP2 (Fig. 5b), the wind speed for GSMaP2 is 20%–25% stronger. Thus, the amplitude of w for GSMaP2 is generally larger than for GSMaP1 except for an absence of upward motion over the eastern coast. For the case of the Indian MCS, the amplitude of w for GSMaP2 (Fig. 5d) is stronger than GSMaP1 (Fig. 5c), because the speed and the westerly component of the mean low-level wind are stronger. For the case of Typhoon Maemi, an area of weak positive w extends from the southwest coast to inland regions of the Korean peninsula for GSMaP1 (Fig. 5e). Conversely, negative w appears at 127°–128°E for GSMaP2, because the position of the typhoon center in the upstream low-level current is shifted southward by about 0.5° from the surface (Fig. 5f), and the positive amplitude of w is larger along the east coast because of the stronger easterly wind component. Both the distributions and amplitudes of Q for GSMaP1 and GSMaP2 are similar (not shown).
Horizontal distributions of orographically forced upward motion (colors) and wind vectors (barbs, long arrowhead: 10 m s−1; short arrowhead: 5 m s−1) derived from (a) surface data and (b) the upstream low-level current at 1800 UTC 8 Aug 2009. (c),(d) As in (a) and (b), but for 0000 UTC 30 Jun 2003. (e),(f) As in (a) and (b), but for 0000 UTC 12 September 2003.
Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0332.1
The horizontal distributions of rainfall amount and the flags of orographic rainfall conditions for GSMaP2 for the cases of Typhoon Morakot and the Indian MCS are shown in Figs. 6a and 6b and Figs. 6c and 6d, respectively. As compared with those for GSMaP1 (Figs. 2a,b and 2c,d), there is no distinct change in either case except for the coastal areas and the generally smoother and broader distributions. This might be attributable to the increase in wind speed between the surface and the upstream low-level current. However, the application of the low-level upstream current to the case of Typhoon Maemi for GSMaP2 (Figs. 6e,f) caused the area of heavy rainfall over western inland parts to diminish, becoming closer in extent to the areas of heavy rainfall observed by the TRMM PR.
As in Fig. 2, but for horizontal distributions of near-surface rainfall rate and flags of the orographic rainfall condition for GSMaP2.
Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0332.1
b. Variable threshold of w depending on mean horizontal wind
Figures 7a and 7b show the horizontal distributions of rainfall amount and the flags of orographic rainfall conditions for GSMaP3 for the case of Typhoon Morakot. GSMaP3 produces the distribution most similar to TRMM PR (Fig. 1a). The area of overdetection of orographic rainfall, seen in GSMaP1 (Fig. 2b) and GSMaP2 (Fig. 6b) in the western foothill area, is diminished (Fig. 7b). In the case of the Indian MCS, the area of heavy rainfall for GSMaP3 (Fig. 7c) is almost the same as that for GSMaP2 (Fig. 6d) owing to the moderate threshold of w for the weak value of U (~5 m s−1). If the conservative threshold of w (0.2 m s−1) for the strong value of U (~20 m s−1) is applied, the heavy orographic rainfall over the coastal land as seen in TRMM PR (Fig. 1d) does not appear even with the change to the upstream low-level current. For the case of Typhoon Maemi, the heavy rainfall over the inland regions for GSMaP3 (Fig. 7e) is similar to that for TRMM PR (Fig. 1k).
As in Fig. 2, but for GSMaP3.
Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0332.1
4. Improvement of orographic rainfall estimation
The bias of rainfall amounts for GSMaP1 is compared with that for the TRMM PR for the three case studies (Figs. 8a–c). In the case of Typhoon Morakot (Fig. 8a), the severe overestimations of >50 mm h−1, which can be seen extending over the western foothills, and even those in the Central Mountain Range, are false alarms. For the Indian MCS case (Fig. 8b), the rainfall rates for GSMaP1 are more than 50 mm h−1 larger than for the TRMM PR, not only over the coastal areas around 18°N but also in the surrounding pixels, which is due to the wider detected area under the orographic rain conditions. In the case of Typhoon Maemi (Fig. 8c), the false alarms over the central inland region cause severe overestimation. Figure 9a shows a scatter diagram of rainfall amounts under orographic rain conditions for the TRMM PR versus GSMaP1 over land in the Asian region (0°–35°N, 65°–135°E) in June–August 2007–08 because the scheme for GSMaP1 mainly operates in this region (Fig. 12 in Y15). All of the products were averaged over 0.1° × 0.1° grids for coherence with the resolution. Some of the pixels at rates of 1–10 mm h−1 for the TRMM PR, which are possibly false alarms of orographic rainfall, caused severe overestimation of over 100 mm h−1 in some cases. This is because the values of rainpct85 estimated from the vertical profiles of orographic rainfall are more sensitive than those values estimated from the profiles of nonorographic rainfall to the scattering signals. Therefore, we performed the same adjustment method for nonorographic rainfall in Eq. (6) except the maximum weight was reduced from 1.00 to 0.75 since this adjustment with the former maximum weight results in underestimation in S15. This adjustment was applied to GSMaP3 (referred to as adjusted GSMaP3).
Differences in rainfall rate between the PR data and those obtained with GSMaP1 from the TMI on (a) 8 Aug 2009 (66 832), (b) 30 Jun 2007 (54 821), and (c) 12 Sep 2003 (33 207). (d)–(f) As in (a)–(c), but for adjusted GSMaP3. The positive (negative) values represent over (under) estimation.
Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0332.1
Scatter diagrams of the surface rain rate retrieved from (a) the PR and GSMaP1 and (b) adjusted GSMaP3 in the Asian region (0°–35°N, 65°–135°E) in June–August 2007–08. The thin and thick lines show the 1:1 lines and linear fitting lines, respectively.
Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0332.1
The bias of rainfall amounts for adjusted GSMaP3 for the three case studies are illustrated in Figs. 8d–f to see an improvement from GSMaP1. For the Typhoon Morakot case (Fig. 8d), the area of severe overestimation over the foothills is narrower, and the bias of the false alarms is mitigated. The bias of rain amount is also retained (~20 mm h−1) for the adjusted GSMaP3 in the case of the Indian MCS (Fig. 8e). The overestimation due to false alarms is modified over inland parts of the Korean peninsula (Fig. 9f). Verification scores for GSMaP1 and adjusted GSMaP3 for the three cases in the area illustrated in Fig. 8 over land are presented in Table 1. The absolute bias and root-mean-square error (RMSE) for adjusted GSMaP3 are mitigated, and the correlation coefficient is improved for all of the cases, particularly for Typhoon Morakot and Typhoon Maemi. The rain pixels for the adjusted GSMaP3 are compared with those of the TRMM PR in Fig. 9b. The severe overestimation at rates of 1–10 mm h−1 for the TRMM PR is mitigated, and the bias from the TRMM PR is smaller than that for GSMaP1. Verification scores for the Asian region over land shown in Fig. 9 are presented in Table 2. The severe overestimation (i.e., large positive absolute bias) as seen in GSMaP1 is distinctly reduced in adjusted GSMaP3, and the root-mean-square error and correlation coefficient for the adjusted GSMaP3 showed an improvement over GSMaP1. The same improvement can be found in the other regions: for example, RMSE is 1.53 (1.52) and 1.77 (1.73) for GSMaP1 (adjusted GSMaP3) for Africa and America, respectively (the same domain in Figs. 12 and 13 in Y15). These improvements in the other regions are small in comparison with the Asian region because the target areas applying the orographic/nonorographic rainfall classification scheme are limited (Fig. 10 in Y15), and the rainfall amount is relatively small (Fig. 12 in Y15).
The verification scores of GSMaP1 (top numbers) and adjusted GSMaP3 (bottom numbers) with PR for the samples illustrated in Fig. 8 over land.
The verification scores of GSMaP1 and adjusted GSMaP3 with PR for the samples in the Asian region (0°–35°N, 65°–135°E) in June–August 2007–08.
5. Summary
An orographic/nonorographic rainfall classification scheme has been introduced since the GSMaP MWR V6 algorithm was provided. However, problems remain regarding overestimation and false alarms of heavy orographic rainfall. This study demonstrated an improvement in the detection of orographic rainfall by including information on the upstream flow in the low-level troposphere, which determines the shift in the position of orographic enhancement on windward slopes of mountains. The GSMaP estimates were compared with surface rainfall data from the TRMM PR for three cases of orographic heavy rainfall and for Asian regions over land. It is difficult to capture such cases causing severe disasters even throughout the TRMM PR and GPM DPR observation period because of their narrow swaths. Although contributions from such a rare case of heavy orographic rainfall are small, it is valuable to quantitatively estimate such heavy orographic rainfall for prediction and mitigation of disasters.
The first point of improvement in the detection of orographic rainfall was the introduction of the upstream flow in the low-level atmosphere. The indices of orographic rainfall conditions (w and Q) in previous versions of this scheme are calculated using surface atmospheric field data. Low-level winds, including low-level jets, are stronger than surface winds, and ground data are influenced considerably by surface conditions such as friction. To consider the effect of the low-level upstream environment, the mean horizontal wind speed between the surface and 2.5 km above the surface was used for calculating w and Q. For w, the upstream low-level wind at a distance of 0.5° from the target point was also considered.
The second point of improvement in the detection of orographic rainfall was the implementation of a variable threshold for w that depends on the mean horizontal wind. Both S15 and Y15 selected lower values of w and Q for the Indian subcontinent and global implementation than T13 for the case of Typhoon Morakot because many orographic rainfall pixels are missed. Conversely, this selection causes overdetection of orographic rainfall conditions for cases with strong winds.
This study also considered the area of orographic enhancement as a function of the upstream wind speed. Under a weak upstream current, orographic rainfall enhancement might extend even over the foothills of mountains, because there could be sufficient time for the precipitation to be enhanced by the orography. Conversely, the precipitation enhancement might occur primarily near the mountain peak under a strong upstream current. This is because in rapid upstream flows, there is insufficient time for the enhancement of precipitation over the windward side of the mountain.
For the three cases of heavy orographic rainfall, this study demonstrated improvement in the detection of orographic rainfall. Furthermore, the estimation of orographic rainfall when applying all the revisions had better values than for the current algorithm for absolute bias, root-mean-square error, and correlation coefficient. The same improvement can be found not only over the Asian land region but also in other regions such as Africa and America.
This study is able to give a solution to overestimates of orographic rainfall for region and quantity. However, some orographic rainfall cases over inland regions [e.g., the summer 2010 floods in Pakistan introduced by Houze et al. (2011)] and other types of orographic enhancement as reviewed by Houze (2012) have not been detected, and the separation from tall orographic rainfall (currently switched off in the scheme) remains. Future work will include the implementation of static stability information from the lower troposphere (dTυ/dzlow, where Tυ is the vertical temperature and dz is from 1.5 to 4.5 km), as suggested by Shige and Kummerow (2016). The adjusted GSMaP3 classification scheme has been applied to GSMaP [version 7 (V7)] MWR algorithms for global precipitation measurements.
Acknowledgments
This work was supported by the Japanese Aerospace Exploration Agency (JAXA) Precipitation Measuring Mission (PMM) project. This study was also supported partly by the Ministry of Science and Technology of Taiwan under Research Grant MOST103-2111-M-002-011-MY3. The TRMM data were provided by JAXA and NASA.
REFERENCES
Aonashi, K., and Coauthors, 2009: GSMaP passive microwave precipitation retrieval algorithm: Algorithm description and validation. J. Meteor. Soc. Japan, 89A, 110–136, doi:10.2151/jmsj.87A.119.
Hou, A. Y., and Coauthors, 2014: The Global Precipitation Measurement mission. Bull. Amer. Meteor. Soc., 95, 701–722, doi:10.1175/BAMS-D-13-00164.1.
Houze, R. A., Jr., 2012: Orographic effects on precipitating clouds. Rev. Geophys., 50, RG1001, doi:10.1029/2011RG000365.
Houze, R. A., Jr., K. L. Rasmussen, S. Medina, S. R. Brodzik, and U. Romatschke, 2011: Anomalous atmospheric events leading to the summer 2010 floods in Pakistan. Bull. Amer. Meteor. Soc., 92, 291–298, doi:10.1175/2010BAMS3173.1.
Kida, S., T. Kubota, S. Shige, and T. Mega, 2017: Development of a rain/no-rain classification methods over land for the microwave sounder algorithm. Remote Sensing of Aerosols, Clouds, and Precipitation, T. Islam et al., Eds., Elsevier, 362 pp., https://www.elsevier.com/books/remote-sensing-of-aerosols-clouds-and-precipitation/islam/978-0-12-810437-8.
Kubota, T., and Coauthors, 2007: Global precipitation map using satellite-borne microwave radiometers by the GSMaP project: Production and validation. IEEE Trans. Geosci. Remote Sens., 45, 2259–2275, doi:10.1109/TGRS.2007.895337.
Kubota, T., T. Ushio, S. Shige, S. Kida, M. Kachi, and K. Okamoto, 2009: Verification of high-resolution satellite-based rainfall estimates around Japan using a gauge-calibrated ground-radar dataset. J. Meteor. Soc. Japan, 87A, 203–222, doi:10.2151/jmsj.87A.203.
Kummerow, C. D., D. L. Randel, M. Kulie, N.-Y. Wang, R. Ferraro, S. J. Munchak, and V. Petkovic, 2015: The evolution of the Goddard profiling algorithm to a fully parametric scheme. J. Atmos. Oceanic Technol., 32, 2265–2280, doi:10.1175/JTECH-D-15-0039.1.
Kwon, E.-H., B.-J. Sohn, D.-E. Chang, M.-H. Ahn, and S. Yang, 2008: Use of numerical forecasts for improving TMI rain retrievals over the mountainous area in Korea. J. Appl. Meteor. Climatol., 47, 1995–2007, doi:10.1175/2007JAMC1857.1.
Lin, Y.-L., 2007: Mesoscale Dynamics. Cambridge University Press, 630 pp.
Liu, G., 1998: A fast and accurate model for microwave radiance calculations. J. Meteor. Soc. Japan, 76, 335–343, doi:10.2151/jmsj1965.76.2_335.
Mega, T., and S. Shige, 2016: Improvements of rain/no-rain classification methods for microwave radiometer over coasts by dynamic surface-type classification. J. Atmos. Oceanic Technol., 33, 1257–1270, doi:10.1175/JTECH-D-15-0127.1.
Onogi, K., and Coauthors, 2007: The JRA-25 Reanalysis. J. Meteor. Soc. Japan, 85, 369–432, doi:10.2151/jmsj.85.369.
Seto, S., N. Takashi, and T. Iguchi, 2005: Rain/no-rain classification methods for microwave radiometer observations over land using statistical information for brightness temperatures under no-rain conditions. J. Appl. Meteor., 44, 1243–1259, doi:10.1175/JAM2263.1.
Seto, S., T. Kubota, N. Takahashi, T. Iguchi, and T. Oki, 2008: Advanced rain/no-rain classification methods for microwave radiometer observations over land. J. Appl. Meteor. Climatol., 47, 3016–3029, doi:10.1175/2008JAMC1895.1.
Shige, S., and C. D. Kummerow, 2016: Precipitation-top heights of heavy orographic rainfall in the Asian monsoon region. J. Atmos. Sci., 73, 3009–3024, doi:10.1175/JAS-D-15-0271.1.
Shige, S., and Coauthors, 2009: The GSMaP precipitation retrieval algorithm for microwave sounders. Part I: Over-ocean algorithm. IEEE Trans. Geosci. Remote Sens., 47, 3084–3097, doi:10.1109/TGRS.2009.2019954.
Shige, S., S. Kida, H. Ashiwake, T. Kubota, and K. Aonashi, 2013: Improvement of TMI rain retrievals in mountainous areas. J. Appl. Meteor. Climatol., 52, 242–254, doi:10.1175/JAMC-D-12-074.1.
Shige, S., M. K. Yamamoto, and A. Taniguchi, 2015: Improvement of TMI rain retrieval over the Indian subcontinent. Remote Sensing of the Terrestrial Water Cycle, Geophys. Monogr., Vol. 206, Amer. Geophys. Union, 27–42, doi:10.1002/9781118872086.ch2.
Skofronick-Jackson, G., and Coauthors, 2017: The Global Precipitation Measurement (GPM) mission for science and society. Bull. Amer. Meteor. Soc., 98, 1679–1695, doi:10.1175/BAMS-D-15-00306.1.
Sohn, B. J., H.-J. Han, and E.-K. Seo, 2010: Validation of satellite-based high-resolution rainfall products over the Korean peninsula using data from a dense rain gauge network. J. Appl. Meteor. Climatol., 49, 701–714, doi:10.1175/2009JAMC2266.1.
Sohn, B. J., G.-H. Ryu, H.-J. Song, and M.-L. Ou, 2013: Characteristic features of warm-type rain producing heavy rainfall over the Korean peninsula inferred from TRMM measurements. Mon. Wea. Rev., 141, 3873–3888, doi:10.1175/MWR-D-13-00075.1.
Spencer, R. W., H. M. Goodman, and R. E. Hood, 1989: Precipitation retrieval over land and ocean with the SSM/I: Identification and characteristics of the scattering signal. J. Atmos. Oceanic Technol., 6, 254–273, doi:10.1175/1520-0426(1989)006<0254:PROLAO>2.0.CO;2.
Stensrud, D. J., 1996: Importance of low-level jets to climate: A review. J. Climate, 9, 1698–1711, doi:10.1175/1520-0442(1996)009<1698:IOLLJT>2.0.CO;2.
Takayabu, Y. N., 2008: Observing rainfall regimes using TRMM PR and LIS data. GEWEX News, No. 18, International GEWEX Project Office, Silver Spring, MD, 9–10.
Taniguchi, A., and Coauthors, 2013: Improvement of high-resolution satellite rainfall product for Typhoon Morakot (2009) over Taiwan. J. Hydrometeor., 14, 1859–1871, doi:10.1175/JHM-D-13-047.1.
Yamamoto, M. K., and S. Shige, 2015: Implementation of an orographic/nonorographic rainfall classification scheme in the GSMaP algorithm for microwave radiometers. Atmos. Res., 163, 36–47, doi:10.1016/j.atmosres.2014.07.024.
Yu, C.-K., and L.-W. Cheng, 2008: Radar observations of intense orographic precipitation associated with Typhoon Xangsane (2000). Mon. Wea. Rev., 136, 497–521, doi:10.1175/2007MWR2129.1.
Yu, C.-K., and L.-W. Cheng, 2014: Dual-Doppler-derived profiles of the southwesterly flow associated with southwest and ordinary typhoons off the southwestern coast of Taiwan. J. Atmos. Sci., 71, 3202–3222, doi:10.1175/JAS-D-13-0379.1.
Yu, C.-K., D. P. Jorgensen, and F. Roux, 2007: Multiple precipitation mechanisms over mountains observed by airborne Doppler radar during MAP IOP 5. Mon. Wea. Rev., 135, 955–984, doi:10.1175/MWR3318.1.