Toward a Unified Near-Field Intensity Map of the 2015 Mw 7.8 Gorkha, Nepal, Earthquake

Abstract

We develop a unified near-field shaking intensity map for the 25 April 2015
Mw 7.8 Gorkha, Nepal, earthquake by synthesizing intensities derived from macroseismic
effects that were determined by independent groups using a variety of
approaches. Independent assessments by different groups are generally consistent,
with minor differences that are likely due in large part to differences in spatial
sampling. Throughout most of the near-field region, European Macroseismic
Scale (EMS-98) intensities were generally close to 7 EMS. In the Kathmandu
Valley, intensities were somewhat higher (6.5–7.5) along the periphery of the valley
and in the adjacent foothills than in the central valley, where they were ≈6. The
results are consistent with instrumental intensity values estimated from available
data using a published relationship between peak ground acceleration (PGA) and
intensity. Using this relationship to convert intensities to PGA, we estimate
strong-motion PGA de-amplification factors of ≈0.7 in the central Kathmandu
Valley, with amplification of ≈1.6 in adjacent foothills. The results support the
conclusion that the Kathmandu Valley experienced a pervasively nonlinear
response during the Gorkha main shock. [DOI: 10.1193/120716EQS226M]

Full text

Toward a Unified Near-Field Intensity
Map of the 2015 Mw7.8Gorkha, Nepal,
Earthquake
Sujan Raj Adhikari,a) Gopi Baysal,a) Amod Dixit,a) Stacey S. Martin,b)
Mattieu Landes,c) Remy Bossu,c) and Susan E. Houghd)
We develop a unified near-field shaking intensity map for the 25 April 2015
Mw7.8 Gorkha, Nepal, earthquake by synthesizing intensities derived from macro-
seismic effects that were determined by independent groups using a variety of
approaches. Independent assessments by different groups are generally consistent,
with minor differences that are likely due in large part to differences in spatial
sampling. Throughout most of the near-field region, European Macroseismic
Scale (EMS-98) intensities were generally close to 7 EMS. In the Kathmandu
Valley, intensities were somewhat higher (6.5–7.5) along the periphery of the val-
ley and in the adjacent foothills than in the central valley, where they were ≈6. The
results are consistent with instrumental intensity values estimated from available
data using a published relationship between peak ground acceleration (PGA) and
intensity. Using this relationship to convert intensities to PGA, we estimate
strong-motion PGA de-amplification factors of ≈0.7 in the central Kathmandu
Valley, with amplification of ≈1.6 in adjacent foothills. The results support the
conclusion that the Kathmandu Valley experienced a pervasively nonlinear
response during the Gorkha main shock. [DOI: 10.1193/120716EQS226M]
INTRODUCTION
The 25 April 2015 Mw7.8 Gorkha, Nepal, earthquake ruptured a 150-km-long segment
of the Main Himalayan Thrust (MHT) Fault, propagating directly below the Kathmandu
Valley at a depth of approximately 12 km (Avouac et al. 2015,McNamara et al. 2016). Whereas
the earthquake caused a serious societal impact in Nepal in terms of loss of life and property,
damage and fatalities were lower than experts had feared for an event of this magnitude in this
location. Although there were instances of catastrophic collapse, overall damage was especially
and surprisingly low within the Kathmandu Valley, where only a small percentage of buildings
sustained substantial to heavy damage. Studies have proposed that the relatively limited damage
resulted from long-period source radiation (Galetzka et al. 2015), pervasive nonlinear response
within the valley (Bhattarai et al. 2015,Dixit et al. 2015,Rajaure et al. 2016), and the distribu-
tion of main shock high-frequency radiation (Avouac et al. 2015,Hough et al. 2016). There is
thus an urgent need to better understand ground motions from the main shock and their
a)
National Society for Earthquake Technology –Nepal, Kathmandu, Nepal
b)
Earth Observatory of Singapore, Singapore
c)
European-Mediterranean Seismological Centre, Essonne, France
d)
U.S. Geological Survey, Pasadena, CA
Earthquake Spectra, Volume 33, No. S1, pages S21–S34, December 2017; © 2017, Earthquake Engineering Research Institute
S21

implications for seismic hazard. Given that instrumental data are limited, it is further necessary
to consider intensity data derived from macroseismic observations, which are expected to reflect
relatively high frequency (≈15Hz) ground motions (Sokolov and Chernov 1998). In recog-
nition of the importance of gathering macroseismic data, intensity and/or damage surveys were
undertaken after the earthquake by a number of independent groups, using a number of different
approaches including direct surveys of intensities estimated from standard questionnaires
(Adhikari et al. 2015; this study), direct surveys of building damage and other macroseismic
effects (Mencin et al. 2016,Ohsumi et al. 2016), Internet-based questionnaires (Bossu et al.
2017), and assessment of media accounts (Martin et al. 2015,McGowan et al. 2016). In this
study, we focus on the intensity distribution in the near-field region, outside of Nepal. The
intensity distribution is constrained primarily by the results of Martin et al. (2015) as well
as intensities determined from Web-based questionnaires (e.g., Bossu et al. 2017).
Of particular note, an extensive Building Damage Assessment (BDA) was conducted by
the National Society for Earthquake Technology (NSET) with several international partner
agencies under three projects that were supported by various donor agencies. A total of
291,891 individual buildings were assessed in the Kathmandu Valley and throughout the
affected municipalities, to determine the observed damage grade according to the 1998
European Macroseismic Scale guidelines (EMS-98; Grünthal et al. 1998) and the building
type. Using this database it is possible to determine the statistical incidence of damage grades
for mud-brick masonry (assumed vulnerability Class A), cement masonry (Class B), and
reinforced concrete frame (Class C) structures at 847 individual locations within all munici-
palities impacted by the earthquake. We use these results to assign intensities using the
EMS-98 scale, hereinafter referred to simply as EMS intensities.
To assess intensities reliably from media accounts, it is widely recognized that it is crucial not
to give undue weight to isolated instances of damage, and to apply EMS-98 guidelines with an
appreciation for issues associated with macroseismic effects of earthquakes in the Indian sub-
continent (see Martin and Hough 2016 for summary). Even when the same general guidelines are
followed, intensity assessments from different approaches can differ because of either inherent
subjectivity in interpretation of accounts (e.g., Hough and Page 2011) or reliance on different
types of information. Hough and Pande (2007) showed that media-based intensities for the 2001
Bhuj, India, earthquake were generally higher than intensities assigned from direct surveys
(Pande and Kayal 2003) of macroseismic effects. This discrepancy is consistent with a reporting
bias whereby media reports are not necessarily overdramatized per se, but are preferentially avail-
able from locations with relatively dramatic effects (e.g., Bakun and Scotti 2006,Hough 2013).
As a first step toward synthesizing the results of independent studies, we compare the results from
independent surveys to determine whether they are generally consistent.
We further use a relationship between peak ground acceleration (PGA) and EMS inten-
sity (Hough et al. 2016,Worden et al. 2012) to generate a map of PGAEMS for the main
shock. We use this map to explore the amplification (and de-amplification) of PGA within
the Kathmandu Valley.
INTENSITY RESULTS
We review briefly the independent intensity assessments considered in this study (Table 1
and Figure 1): (1) The NSET study was undertaken by a team of researchers beginning in
S22 ADHIKARI ET AL.

May 2015, using an intensity questionnaire that was formulated considering building types
and local conditions in Nepal (Adhikari et al. 2015,Adhikari et al. 2016,Murakami et al.
2015). Intensities were assigned for a total of 265 locations. These locations are concentrated
in the Kathmandu Valley, but sites outside the valley were surveyed as well. (2) The Nanyang
Table 1. List of intensity surveys considered
Study Agencies Approach Locations
1 NSET Direct questionnaire survey 265
2 NTU Media-based 3,831
3 EMSC Internet-based questionnaire 106
4 CU Ground survey 72
5 DYFI-1 Internet-based questionnaire 68
6 DYFI-2 Media-based 16
7 NSET-BDA BDA 291,891
85˚ 86˚
28˚
CU
6.67+/-0.71
012345678910
85˚ 86˚
NSET
6.13+/-0.53
85˚ 86˚
DYFI
8.59+/-0.41
28˚
NTU
6.57+/-1.03 BDA
6.74/-0.74 EMSC
6.39+/-2.23
(a) (b) (c)
(d) (e) (f)
Figure 1. Intensity values throughout the near-field region from independent studies: (a) NTU,
(b) BDA, (c) EMSC, (d) CU, (e) NSET, and (f) DYFI-1 and DYFI-2. In each panel, intensities are
shown for individual locations (circles; color scale shown in lower left panel), which in some cases
are averages from more than one questionnaire. The DYFI data include intensity values determined
from online questionnaires and values determined from either instrumental data (one location) or
from assessment of media accounts. Each panel also shows main shock rupture (Lindsey et al.
2015) (blue line), main shock epicenter (black star), epicenter of the 10 May 2015 Dolakha after-
shock (white star), and average near-field intensity 1σshown for each data set.
TOWARD A UNIFIED NEAR-FIELD INTENSITY MAP OF THE 2015 MW7.8 GORKHA, NEPAL, EARTHQUAKE S23

Technological University (NTU) study was undertaken immediately following the main
shock, relying on detailed media accounts made available over the Internet, including photo-
graphs and CCTV footage, to determine EMS-98 values for 3,831 locations within and out-
side of Nepal (Martin et al. 2015). (3) The European Mediterranean Seismic Centre (EMSC)
data set includes intensities determined automatically based on user-specified thumbnail
illustrations that correspond to observed effects at different shaking levels (Bossu et al.
2017). (4) The University of Colorado, Boulder (CU) study was undertaken immediately fol-
lowing the earthquake, relying on assessment of EMS intensities based on direct inspection of
damage and other effects, supplemented with assignments based on media accounts, at a total of
72 locations (Mencin et al. 2016). (5) The DYFI-1 study includes 68 modified Mercalli intensity
(MMI) values determined automatically from online questionnaires (see Wald et al. 1999).
(6) The DYFI-2 study includes 16 near-field MMI values determined in one case from instru-
mental data and for 15 locations from assessment of media accounts (McGowan et al. 2016).
Lastly, (7) the NSET-BDA study includes a total of over 800 intensity values determined from
the exhaustive BDA project, which involved direct surveys of 291,891 individual buildings.
The NSET-BDA study far surpasses all of the other studies in both the number of indi-
vidual locations surveyed and the rigor of damage assessments, providing the opportunity to
assign EMS intensities based on the observed statistical incidence of damage grades for
different vulnerability classes. In effect, the availability of this extensive data set, which pro-
vides far more detailed information than is usually available for investigations of macroseis-
mic effects, provides a unique opportunity to assess the veracity of the other studies, which
represent more commonly used approaches to assess macroseismic intensity.
Table 1lists the approach used to assess intensities in each study. In some cases, more
than one approach was used within the same study. The NTU study was overwhelmingly
based on interpretation of media accounts from outside and within Nepal, with a small
number of EMS-98 values based on both media accounts and available information from
ground surveys. The CU study was primarily based on ground surveys, supplemented with
a small number of intensities assessed from media accounts. Maps of intensity values from
studies 1–6 are shown for the near-field region (Figure 1) and for the Kathmandu Valley
(Figure 2), including the average and sample standard deviation for each data set.
SYNTHESIS AND ANALYSIS OF DATA
Our initial comparison of the independent data sets (Figures 1and 2) reveals that the
NSET, BDA, NTU, EMSC, and CU studies are generally consistent to within 1σ. The
DYFI-1 data set includes few near-field values. DYFI-2 values, estimated subjectively
from a limited number of media accounts (McGowan et al. 2016), are higher than average
intensities from other studies. Qualitatively, the difference between the DYFI-2 results and
the other results are generally consistent with a reporting bias (e.g., Bakun and Scotti 2006,
Hough 2013,Hough and Pande 2007). Whereas the NTU study was also based overwhel-
mingly on media accounts, the reporting bias appears to have been obviated by the exhaustive
nature of the Martin et al. (2015) study, which involved a painstaking search of all available
data sources including CCTV footage made available over the Internet following the earth-
quake. Martin et al. (2015) further interpreted EMS values with an appreciation for issues asso-
ciated with assessment of intensities in the Indian subcontinent (see Martin and Hough 2016),
S24 ADHIKARI ET AL.

including the fact that masonry buildings in the Indian subcontinent are pervasively more
vulnerable than masonry buildings in Europe, where the EMS scale was developed.
Individual EMS values from the EMSC study (Bossu et al. 2017) reveal significantly
more variability than any of the other studies. This variability likely reflects the nature
of the data, which are derived from individual thumbnail questionnaire results with no spatial
averaging, and without information about building type or vulnerability. Reports from indi-
vidual locations with instances of catastrophic collapse would have been interpreted as high
(9–10 EMS) intensity. Of note, however, the average raw intensity value across the
Kathmandu Valley is consistent with the average value from all of the other studies
apart from DYFI-2. While DYFI data have generally been shown to provide reliable indi-
cators of ground motion and to be highly useful to investigate ground motions (e.g., Atkinson
and Wald 2007), the DYFI-2 values for the Gorkha earthquake were assigned based on media
accounts rather than the usual algorithm. Because the DYFI-2 data set is both small and an
outlier among the studies, we will not include it in our synthesis.
In light of the above considerations, our synthesis map for the Kathmandu Valley
includes the NTU, EMSC, CU, DYFI-1, NSET, and BDA results. The small differences
85˚15' 85˚20' 85˚25'
27˚40'
27˚45'
6 km
CU
6.9+/-0.68
012345678910
85˚15' 85˚20' 85˚25'
6 km
NSET
6.0+/-0.44
85˚15' 85˚20' 85˚25'
6 km
DYFI
8.2+/-0.39
27˚40'
27˚45'
6 km
NTU
6.5+/-0.63
6 km
BDA
6.5+/-0.66
6 km
EMSC
6.4+/-2.23
(a) (b) (c)
(d) (e) (f)
Figure 2. Intensity values in the Kathmandu Valley from independent studies: (a) NTU,
(b) BDA, (c) EMSC, (d) CU, (e) NSET, and (f) DYFI-1 and DYFI-2. In each panel, intensities
are shown for individual locations (circles; color scale shown in lower left panel), which in
some cases are averages from more than one questionnaire. Available DYFI data from within
the Kathmandu Valley (squares) are all determined from either instrumental data (one location)
or from assessment of media accounts. Average near-field intensity 1σshown for each data
set in each panel. Southern edge of main shock rupture shown in top right panel from Lindsey
et al. (2015) (dark line).
TOWARD A UNIFIED NEAR-FIELD INTENSITY MAP OF THE 2015 MW7.8 GORKHA, NEPAL, EARTHQUAKE S25

among the average intensities for these studies could reflect (1) systematic differences in
subjective interpretation, (2) random scatter, and/or (3) sampling bias. Among these five
studies, average intensity values across both the near-field and Kathmandu Valley are con-
sistent within 1σ. It is reasonable to conclude that subjective interpretations were generally
consistent among these studies, and consistent with the algorithms used for the EMSC and
DYFI studies, and differences result primarily from differences in sampling and/or spatial
averaging. Furthermore, to the extent that there might have been differences in subjective
interpretation, it is reasonable to average the results (e.g., Hough and Page 2011). We, there-
fore, develop synthesis maps by combining the six intensity data sets, without attempting to
calibrate individual intensity data sets. For the combined data set, we omit several isolated
EMS values of 9–10 from the EMSC data set: because these intensities are point values, we
expect that determination of a representative intensity for a given location will require aver-
aging, which requires sufficient spatial sampling. In Figure 3, we present the combined inten-
sity data set for the Kathmandu Valley. For the region defined by the map limits, the average
EMS intensity is 6.4 0.9.
Figure 4presents contoured EMS intensities throughout the near-field region. To gen-
erate the interpolated map, we use a Laplacian smoothing operator with a tension factor
of 1.0, which ensures that no maxima or minima are generated except at control points
(Wessel and Smith 1999). To improve the visualization of the variability of EMS values,
Figure 3. EMS intensity values throughout the Kathmandu Valley, omitting EMSC values.
S26 ADHIKARI ET AL.

in Figure 5, we show EMS residuals throughout the near-field region, calculated by sub-
tracting the near-field average of 6.6 from all values. (This average value is calculated using
all intensities within 84.6–86.4W, 27.4–28.3N. Within the estimated rupture perimeter the
average is ≈7.)
Figures 3and 4reveal that EMS-98 values were overwhelmingly within a narrow
range, 6–7.5, throughout the Kathmandu Valley as well as throughout the near-field region,
reaching values as high as 8–8.5 in only a few areas. This is consistent with the conclusions of
earlier studies based on more limited data sets (Adhikari et al. 2016,Hough et al. 2016,
Martin et al. 2015,Mencin et al. 2016). The spatial distribution of EMS intensities in
the Kathmandu Valley is qualitatively similar to that observed in earlier studies, with the
lowest values in the central, deepest part of the valley (Figure 3), and somewhat higher values
around the periphery of the valley and in the adjacent foothills. Figure 3does not reveal any
85˚15' 85˚20' 85˚25'
27˚40'
27˚45'
6 km
PGA-based EMS
27˚40'
27˚45'
6 km
PGV-based EMS
84˚30' 85˚00' 85˚30' 86˚00'
27˚30'
28˚00'
28˚30'
2601 345 78910
(a)
(b)
Figure 4. (a) (left) Contoured EMS intensities (color scale indicated) across the near-field region;
the main shock rupture (Lindsey et al. 2015) and the Kathmandu Valley are shown with solid and
dashed lines, respectively. Small gray dots indicate locations where intensities are estimated. The
intensity distribution in the northwest corner of the map, which is effectively unconstrained, is
masked. (b) (right) Contoured intensities in the Kathmandu Valley (same color scale shown in
Figure 4a). Filled circles in two panels indicate instrumental intensities estimated from PGA
(bottom) and PGV (top).
TOWARD A UNIFIED NEAR-FIELD INTENSITY MAP OF THE 2015 MW7.8 GORKHA, NEPAL, EARTHQUAKE S27

significant north-to-south trend across the Kathmandu Valley; high residuals are observed in
foothill locations in all directions with the possible exception of the southwestern corner,
where intensities are poorly constrained.
The spatial distribution of intensities throughout the near-field region reveals some differ-
ences with respect to the results of earlier studies (Hough et al. 2016,Martin et al. 2015).
Consistent with the earlier studies, we find intensities to be generally higher toward the northern
edge of the rupture, but unlike the earlier studies the swath of high intensities is less narrowly
concentrated along the northern edge; instead, there is a swath of relatively high intensities
close to the along-strike midline of the rupture. The average EMS intensity across the
near-field region is ≈7, slightly higher than that estimated by Mencin et al. (2016).
One can further consider the consistency of directly observed macroseismic data with
limited instrumental recordings of the main shock. Within the Kathmandu Valley, available
instrumental data include conventional strong-motion recordings from a total of six sites
(Bhattarai et al. 2015,Dixit et al. 2015,Rajaure et al. 2016) as well as high-rate (5 samples
per s) GPS from two sites (Galetzka et al. 2015, also see Hashash et al. 2016). Stations KKN4
and KTP are hard-rock reference sites, and TVU is a presumed thin-sediment site near the edge
of the valley. The rest of the sites are within the valley. The GPS data provide no constraint on
ground motions at frequencies above 2.5 Hz, and therefore no constraints on PGA, but the GPS
data do constrain peak ground velocity (PGV; Table 2). Instrumental intensities, IPGA and IPGV ,
can be estimated from recorded PGA and PGV values using an intensity-prediction relation-
ship; we use the relationships determined by Worden et al. (2012):
EQ-TARGET;temp:intralink-;e1;41;121IPGA ¼1.78 þ1.55logðPGAÞ,logðPGAÞ≤1.57
IPGA ¼1.60 þ3.70logðPGAÞ,logðPGAÞ>1.57 (1)
84˚30' 85˚00' 85˚30' 86˚00'
27˚30'
28˚00'
28˚30'
–1 0 1 2
Figure 5. Contoured EMS intensity residuals (color scale indicated) across the near-field region;
the main shock rupture (Lindsey et al. 2015) and the Kathmandu Valley are shown with solid blue
and dashed black lines, respectively. The intensity distribution in the northwest corner of the map,
which is effectively unconstrained, is masked.
S28 ADHIKARI ET AL.

and
EQ-TARGET;temp:intralink-;e2;62;412IPGV ¼3.78 þ1.47logðPGVÞ,logðPGVÞ≤0.53
IPGV ¼2.89 þ3.16logðPGVÞ,logðPGVÞ>0.53 (2)
where PGA and PGV are in cm/s2and cm/s, respectively. The values are given in Table 2and
compared to observed intensities in Figure 4b.
As shown in Figure 4b, we find good consistency between IPGA values and directly esti-
mated intensities. Not only is the overall shaking level consistent; both estimated and directly
estimated intensities reveal the same subtle difference between intensities in the valley versus
the adjacent foothills. In contrast, IPGV values are considerably higher than directly estimated
intensities (Figure 4b, top). Intensities as high as 9 are grossly inconsistent with the low
overall level of damage to highly vulnerable structures, revealing that the intensity-PGV
relationship cannot explain ground motions within the Kathmandu Valley during this
earthquake.
To further explore the distribution of ground motions, we now use the Worden et al.
(2012) relationships to estimate PGAEMS (in cm/s2) from estimated EMS intensities.
Although the Worden et al. (2012) relationship was developed using data from
California, and using MMI rather than EMS, Hough et al. (2016) show that it provides a
good fit to available data from the Gorkha main shock for PGA values above ≈1%g.
Figure 6shows estimated PGAEMS amplification factors across the Kathmandu Valley,
estimated relative to the observed average, PGAAV E (e.g., PGAEMS∕PGAAV E ). Estimated
amplification factors are all in the range of 0.4–4.5, with 93.6%between 0.5–2.9 (2.5%
are below 0.5; 4.2%are above 2.9). The pattern of de-amplification within the central
Kathmandu Valley, with modest amplification around the periphery of the valley and in
the adjacent foothills, is clearly illuminated. Draping this result over topography (Figure 7)
Table 2. Instrumental data from the Kathmandu Valley. List of strong-motion
and high-rate GPS stations that recorded the Gorkha main shock, including pub-
lished PGA and PGV values (%gand cm/s, respectively), as well as IPGA and
IPGV estimates from Equations 1and 2
Latitude Longitude PGA PGV
Station (°N) (°E) (g)IPGA (cm/s) IPGV
KAT 27.713 85.316 16.5 6.6 107 9.3
DMG 27.719 85.317 17.8 6.7 ––
KTP 27.682 85.273 24.6 7.2 52 8.3
TVU 27.681 85.288 24.2 7.2 99 9.2
PTN 27.681 85.319 15.4 6.5 74 8.8
THM 27.713 85.377 15.0 6.4 90 9.1
NAST 27.657 85.328 ––90 9.1
KKN4 27.801 85.279 ––70 8.7
TOWARD A UNIFIED NEAR-FIELD INTENSITY MAP OF THE 2015 MW7.8 GORKHA, NEPAL, EARTHQUAKE S29

85˚15' 85˚20' 85˚25'
27˚40'
27˚45'
0.5 1.0 1.5 2.0 2.5
Figure 6. PGA residuals (amplification/de-amplification factors) relative to the average, calcu-
lated by dividing estimated PGAEMS values by the average across the Kathmandu Valley.
Figure 7. PGA residuals (amplification/de-amplification factors; same color scale as shown
in Figure 6) from Figure 6draped over topography (factor of 2 vertical exaggeration) within
the Kathmandu Valley. Dashed line indicates southern limit of main shock rupture from
Lindsey et al. (2015).
S30 ADHIKARI ET AL.

further illustrates this result. We note that residuals in the southwest corner of the region are
effectively unconstrained.
The amplification factors shown in Figure 7are calculated relative to the average esti-
mated PGAEMS across the region rather than relative to directly estimated hard-rock reference
values. The choice of baseline is subjective; alternatively, one could divide all values by an
average calculated for foothill sites only; however, the results could be biased if, as suggested
by Martin et al. (2015), topographic amplification was common. We note that this ambiguity
is not unique to this study, and has likely been an unrecognized source of uncertainty in many
past site response studies that relied on reference sites. We further note that Rajaure et al.
(2016) estimated amplification factors of 0.60–0.92 directly from strong-motion recordings
of the main shock from three sediment sites and one hard-rock site. While the baseline used
in this study might be open to question, results are thus consistent with instrumentally
constrained amplification factors.
DISCUSSION AND CONCLUSIONS
Given the limited instrumental recordings of the 2015 Gorkha main shock, macroseismic
data offers the best available information to constrain the distribution of near-field ground
motions. In this study, we compare and combine macroseismic data sets, including the exten-
sive BDA, that were collected and interpreted by independent teams over the months follow-
ing the main shock. We find that assessment of EMS intensities done by independent groups
was generally consistent. The consistency of the media-based NTU results (Martin et al.
2015) with results from direct surveys indicates that media-based intensity distributions
can reliably capture representative shaking effects, if spatially rich macroseismic information
is collected and carefully interpreted.
Directly estimated intensities are consistent with instrumental intensities estimated from
available instrumental data in the Kathmandu Valley using the intensity-PGA relationship
developed by Worden et al. (2012), but are inconsistent with instrumental intensities using
the Worden et al. (2012) intensity-PGV relationship. IPGV values are not plausible in light of
the pervasively low level of damage to highly vulnerable structures in the Kathmandu Valley,
suggesting that, for this earthquake and this location, damage was controlled by PGA rather
than PGV. Two factors plausibly account for this result: (1) vernacular structures in the val-
ley, almost all of which are 1 to 4 stories, are not vulnerable to the longer period shaking that
controlled PGV, such that high PGV values were not generally damaging, and (2) high-
frequency energy levels and PGA were unusually low within the Kathmandu Valley for rea-
sons summarized in the following paragraph.
The combined EMS data set provides improved spatial resolution of the distribution of
ground motions during the Gorkha main shock compared to early studies based on indivi-
dual surveys. The distribution of near-field ground motions illuminated by this study con-
firms the earlier result that ground motions, and damage, within the central Kathmandu
Valley were lower than expected because of a pervasive nonlinear response of valley sedi-
ments (Bhattarai et al. 2015,Dixit et al. 2015,Hough et al. 2016,Rajaure et al. 2016), with
an estimated PGA de-amplification factor of ≈0.7 in the central valley. Whereas other fac-
tors, including source radiation and the distribution of high-frequency shaking might
have also contributed to the relatively low level of shaking in the Kathmandu Valley
TOWARD A UNIFIED NEAR-FIELD INTENSITY MAP OF THE 2015 MW7.8 GORKHA, NEPAL, EARTHQUAKE S31

(e.g., Avouac et al. 2015,Galetzka et al. 2015), it is clear from Figure 5that shaking in the
valley was significantly lower than in other regions along the southern edge of the main
shock rupture. The de-amplification factor within the central valley was sufficient to reduce
EMS-98 intensities from 7+ to close to 6, a significant difference given the vulnerability of
vernacular structures.
The overall distribution of near-field intensities reveals some similarities but also some
differences from the distribution revealed by earlier studies (Hough et al. 2016,Martin et al.
2015). As in earlier studies (Hough et al. 2016), we observe lower ground motions along the
southern edge of the rupture, and higher shaking in the forward directivity direction. The
swath of relatively high intensities, however, is less strongly concentrated along the far
northern edge of the rupture, instead extending from the approximate north-south centerline
of the rupture to its northern edge. We note, however, that the intensity distribution along the
far northern edge of the rupture, along the southern edge of the high Himalaya, is poorly
constrained (see Figure 4a). The improved intensity distribution will provide a useful con-
straint for future main shock rupture modeling in the Kathmandu Valley, as well as for devel-
opment of shaking scenarios for future large earthquakes in the region.
REFERENCES
Adhikari, S. R., Basyal, G. K., Pradhan, S., and Shrestha, S., 2015. Post-earthquake assessment of
buildings using mobile technology and Google Imaginary Plus GIS visualization, in Proceed-
ings, 14th International Symposium on New Technology for Urban Safety of Mega Cities in
Asia,29–31 October 2015, Kathmandu, Nepal.
Adhikari, S. R., Guragain, R., and Murakami, H., 2016. Study on shaking intensity distribution of
the 2015 Gorkha earthquake in Nepal, in Proceedings, International Symposium on the 1st
Anniversary of the Gorkha earthquake,25–26 April 2016, Kathmandu, Nepal. doi:10.13140/
RG.2.2.10485.78562.
Atkinson, G. M., and Wald, D. J., 2007. “Did you feel it?”Intensity data: A surprisingly good
measure of earthquake ground motion, Seismological Research Letters 78, 362–368.
Avouac, J.-P., Meng, L., Wei, S., Wang, T., and Ampuero, J.-P., 2015. Unzipping lower edge of
locked Main Himalayan Thrust during the 2015, Mw7.8 Gorkha earthquake, Nature
Geoscience 8, 708–711. doi:10.1038/ngeo2518.
Bakun, W. H., and Scotti, O., 2006. Regional intensity attenuation models for France and the
estimation of magnitude and location of historical earthquakes, Geophysical Journal Interna-
tional 164, 596–610. doi:10.1111/j.1365-246X.2005.02808.x.
Bhattarai, M., Adhikari, L. B., Gautam, U. P., Laurendeau, A., Labonne, C., Hoste-Colomer, R.,
Sèbe, O., and Hernandez, B., 2015. Overview of the large 25 April 2015 Gorkha, Nepal, earth-
quake from accelerometric perspectives, Seismological Research Letters 86, 6. doi:10.1785/
0220150140.
Bossu, R., Landès, R., Roussel, F., Steed, R., Mazet-Roux, R., Martin, S. S., and Hough, S. E.,
2017. Thumbnail-based questionnaires for the rapid and efficient collection of macroseismic
data from global earthquakes, Seismological Research Letters 88, 5. doi:10.1785/0220160120.
Dixit, A. M., Ringler, A., Sumy, D., Cochran, E., Hough, S. E., Martin, S. S., Gibbons, S.,
Luetgert, J., Galetzka, J., Shrestha, S. N., Rajaure, S., and McNamara, D., 2015. Strong motion
recordings of the M7.8 Gorkha, Nepal, earthquake sequence from low-cost, Quake Catcher
Network accelerometers, Seismological Research Letters 86, 1533–1539. doi:10.1785/
0220150146.
S32 ADHIKARI ET AL.

Galetzka, J., Melgar, D., Genrich, J. F., Geng, J., Owen, S., Lindsey, E. O., Xu, X., Bock, Y.,
Avouac, J.-P., Adhikari, L.-B., Upreti, B. N., Pratt-Sitaula, B., Bhattarai, T. N., Sitaula, B.,
Moore, A., Hudnut, K. W., Szeliga, W., Normandeau, J., Fend, M., Flouzat, M., Bollinger, L.,
Shreshta, P., Koirala, B., Gautam, U., Bhatterai, M., Gupta, R., Kandel, T., Timsina, C.,
Sapkota, S. N., Rajaure, S., and Maharjan, N., 2015. Slip pulse and resonance of the
Kathmandu basin during the 2015 Gorkha earthquake, Nepal, Science 349, 1091–1095.
doi:10.1126/science.aac6383.
Grünthal, G., Musson, R. M. W., Schwartz, J., and Stucchi, M., (eds.), 1998. The European
Macroseismic Scale EMS-98, Conseil de l’Europe, Cahiers du Centre Européen de Géody-
namique et de Séismoligie, Luxembourg, p. 101.
Hashash, Y. M. A., Tiwari, B., Moss, R. E. S., Asimaki, D., Clahan, K. B., Kieffer, D. S., Dreger,
D. S., Macdonald, A., Madugo, C. M., Mason, H. B., Pehlivan, M., Rayamajhi, D.,
Acharya, I., and Adhikari, B., 2015. Geotechnical Field Reconnaissance: Gorkha (Nepal)
Earthquake of April 25 2015 and Related Shaking Sequence, Geotechnical Extreme Event
Reconnaissance GEER Association Report No. GEER-040.
Hough, S. E., 2013. Spatial variability of “Did you feel it?”Intensity data: Insights into sampling
biases in historical earthquake intensity distributions, Bulletin of the Seismological Society of
America 103, 2767–2781. doi:10.1785/0120120285.
Hough, S. E., Martin, S. S., Gahalaut, V. K., Joshi, A., Landès, M., and Bossu, R., 2016. A
comparison of observed and predicted ground motions from the 2015 Mw7.8 Gorkha,
Nepal, earthquake, Natural Hazards 84, 1661–1684. doi:10.1007/s11069-016-2505-8.
Hough, S. E., and Page, M., 2011. Toward a consistent model for strain accrual and release for the
New Madrid Seismic Zone, central United States, Journal of Geophysical Research: Solid
Earth 116, 2156–2202. doi:10.1029/2010JB007783.
Hough, S. E., and Pande, P., 2007. The 1868 Hayward Fault, California, earthquake: Implications
for earthquake scaling relations on partially creeping faults, Bulletin of the Seismological
Society of America 97, 638–645. doi:10.1785/0120060072.
Lindsey, E. O., Natsuaki, R., Xu, X., Shimada, M., Hashimoto, M., Melgar, D., and Sandwell,
D. T., 2015. Line-of-sight displacement from ALOS-2 interferometry: Mw7.8 Gorkha earth-
quake and Mw7.3 aftershock, Geophysical Research Letters,42, 6655–6661. doi:10.1002/
2015GL065385.
Martin, S. S., and Hough, S. E., 2016. Reply to “Comment on ‘Ground motions from the 2015
Mw7.8 Gorkha, Nepal, earthquake constrained by a detailed assessment of macroseismic data’
by Stacey Martin, Susan E. Hough, and Charleen Hung”by Andrea Tertulliani, Laura
Graziani, Corrado Castellano, Alessandra Maramai, and Antonio Rossi, Seismological
Research Letters 87, 957–962. doi:10.1785/0220160061.
Martin, S. S., Hough, S. E., and Hung, C., 2015. Ground motions from the 2015 Mw7.8 Gorkha,
Nepal, earthquake constrained by a detailed assessment of macroseismic data, Seismological
Research Letters 86, 1524–1532. doi:10.1785/0220150138.
McGowan, S. M., Jaiswal, K. S., and Wald, D. J., 2016. Using structural damage statistics to
derive macroseismic intensity within the Kathmandu Valley for the 2015 M7.8 Gorkha, Nepal
earthquake, Tectonophysics 714–715, 158–172. doi:10.1016/j.tecto.2016.08.002.
McNamara, D.E., Yeck, W. L., Barnhart, W. D., Schulte-Pelkum, V., Bergman, E., Adhikari, L.B.,
Dixit, A., Hough, S. E., Benz, H. M., and Earle, P. S., 2016. Source modelling of the 2015
Mw7.8 Nepal (Gorkha) earthquake sequence: implications for geodynamics and earthquake
hazards, Tectonophysics 714–715,21–20. doi:10.1016/j.tecto.2016.08.004.
TOWARD A UNIFIED NEAR-FIELD INTENSITY MAP OF THE 2015 MW7.8 GORKHA, NEPAL, EARTHQUAKE S33

Mencin, D., Bendick, R., Upreti, B. N., Adhikari, D. P., Gajurel, A. P., Bhattarai, R. R., Shrestha,
H. R., Bhattarai, T. N., Manandhar, N., Galetzka, J., and Knappe, E., 2016. Himalayan strain
reservoir inferred from limited afterslip following the Gorkha earthquake, Nature Geoscience
9,533–537. doi:10.1038/ngeo2734.
Murakami, H., Gurgain, R., Adhikari, S. R., Basyal, G., and Mori, S., 2015. Seismic intensity
questionnaire survey for the 2015 Gorkha, Nepal earthquake –preliminary results and damage
observations, Annual Conference and International Symposium on Earthquake Engineering
2015, Japan Association for Earthquake Engineering, 19–20 November 2015, Tokyo, Japan.
Ohsumi, T., Hiroshi, I., Inoue, H., Aoi, S., and Fujiwara, H., 2016. Investigation of Damage in
and Around Kathmandu Valley Related to the 2015 Gorkha, Nepal Earthquake,Technical
Note of the National Research Institute for Earth Science and Disaster Prevention,404,
53 pp.
Pande, P., and Kayal, J. R. (eds.), 2003. Kutch (Bhuj) earthquake 26 January 2001, Geological
Survey of India, Special Publication 76, Lucknow.
Rajaure, S., Asimaki, D., Thompson, E., Hough, S. E., Ampuero, P., Martin, S. S., Inbal, A., and
Dhital, M., 2016. Strong motion observations of the Kathmandu valley response during the
M7.8 Gorkha earthquake sequence, Tectonophysics 714–715, 146–157. doi:10.1016/j.tecto.
2016.09.030.
Sokolov, Yu. Y., and Chernov, Yu. K, 1998. On the correlation of seismic intensity with Fourier
Amplitude Spectra, Earthquake Spectra 14, 679–694.
Wald, D. J., Quitoriano, V., Dengler, L., and Dewey, J. W., 1999. Utilization of the Internet for
rapid community intensity maps, Seismological Research Letters 70, 680–697. doi:10.1785/
gssrl.70.6.680.
Wessel, P., and Smith, W. H. F., 1999. Free software helps map and display data, Eos Transac-
tions 72, 441–445. doi:10.1029/90EO00319.
Worden, C. B., Gerstenberger, M. C., Rhoades, D. A., and Wald, D. J., 2012. Probabilistic rela-
tionships between ground-motion parameters and modified Mercalli intensity in California,
Bulletin of the Seismological Society of America 102, 204–221. doi:10.1785/0120110156.
(Received 7 December 2016; accepted 8 August 2017)
S34 ADHIKARI ET AL.