Please use this PDF proof to check the layout of your article.

If you would like any changes to be

made to the layout, you can leave instructions in the online proofing interface. First, return to the
online proofing interface by clicking "Edit" at the top page, then insert a Comment in the relevant
location. Making your changes directly in the online proofing interface is the quickest, easiest way to
correct and submit your proof.

Please note that changes made to the article in the online proofing interface will be added to the
article before publication, but are not reflected in this PDF proof.
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

Available online at www.sciencedirect.com

1
ScienceDirect
Advances in Space Research xxx (xxxx) xxx
www.elsevier.com/locate/asr

2 Seafloor depth mapping of central Vietnam’s sea area and its

3 surrounding using gravity anomaly data and gravity geological method

of
4 Nguyen Van Sang a, Khuong Van Long b, Tran Tuan Dung c, Lam Van Nguyen d,
5 Bui Cong Que e, Do Van Mong b, Bui Dang Quang f, Ole BaltazarAndersen g
ReneForsberg g, Dieu Tien Bui h,⇑

o
6
a
7 Department of Geodesy, Faculty of Geomatics and Land Administration, Hanoi University of Mining and Geology, No. 18 Pho Vien, Duc Thang, Bac

Pr
8 Tu Liem, Hanoi 10000, Vietnam
b
9 Vietnam’s People Naval Hydrographic and Oceanographic Department, No. 3, Mac Quyet, Anh Dung, Duong Kinh, Hai Phong 04000, Vietnam
c
10 Institute of Marine Geology and Geophysics, Vietnam Academy of Science and Technology, Building A27, No. 18, Hoang Quoc Viet, Nghia Do, Cau
11 Giay, Hanoi 10000, Vietnam
d
12 Department of Ocean Operations and Civil Engineering, Faculty of Engineering, Norwegian University of Science and Technology (NTNU),
13 Larsgårdsvegen 2, 6025 
Alesund, Norway
e
14 Institute of Geophysics, Vietnam Academy of Science and Technology, Building A8, No. 18, Hoang Quoc Viet, Nghia Do, Cau Giay, Hanoi 10000, Vietnam
f
15
16
17 h
g
ed
Department of Survey, Mapping and Geographic Information Vietnam, No. 2, Dang Thuy Tram Street, Co Nhue 1, Bac Tu Liem, Hanoi 10000, Vietnam
DTU Space, Technical University of Denmark, 2800 Kgs. Lyngby, Denmark
GIS Group, Department of Business and IT, University of South-Eastern Norway, 3800 Bø i Telemark, Norway
18 Received 26 November 2022; received in revised form 8 April 2023; accepted 20 April 2023
19
ct
21
20
22 Abstract
re

23 Seafloor topography mapping has a vital role in territorial management, geological investigation, natural hazard research, and marine
24 construction; therefore, improving the accuracy of seafloor topography mapping is essential. The main objective of this research is to
25 assess the possibility of improving the accuracy of seafloor topographic mapping for the central sea area of Vietnam and its surroundings
26 based on the combination of satellite-derived gravity anomaly data and shipborne data. Firstly, the best satellite-derived gravity anomaly
27 model for the study area is determined by comparing its gravity with the shipborne gravity anomaly. Then, the systematic deviation in
or

28 the satellite-derived gravity anomaly is eliminated and fitted. Then, the substance density contrast (SDC) between the seabed and the
29 seawater is derived. Subsequently, the seafloor depth was computed using the Gravity Geological method. In the next step, the computed
30 depth is fitted with the shipborne depth to improve the accuracy and reduce the systematic deviation. The result shows that the best
31 satellite-derived gravity anomaly model for the study area is the DTU17GRAV, with an accuracy of ± 5.06 mGal. The most suitable
nc

32 SDC in the study area is 1.40 g/cm3. The final seafloor topographic map with 10 10 grid is obtained with the root mean square deviation
33 (RMSD) of ± 83.96 m, which is lower than that of the global terrain model of GEBCO 2022 (RMSD = ±152.88 m). Therefore, it is
34 concluded that the combination of satellite-derived gravity anomaly and shipborne data can significantly improve the accuracy of sea-
35 floor mapping for complex topographic areas. The results of this research are helpful for seabed applications and territorial management
36 in the region.
Ó 2023 COSPAR. Published by Elsevier B.V. All rights reserved. 37
U

38
⇑ Correspondent author.
Keywords: Gravity Geological Method; Satellite-derived Gravity Ano- 39
E-mail addresses: nguyenvansang@humg.edu.vn (N. Van Sang), long-
maly; Gravity-derived Depth; Vietnam 40
doan6@gmail.com (K. Van Long), trantuandung@yahoo.com (T. Tuan
41
Dung), lam.v.nguyen@ntnu.no (L. Van Nguyen), bcque2010@gmail.com
(B. Cong Que), dovanmong@gmail.com (D. Van Mong), dquang1979@g-
mail.com (B. Dang Quang), oa@space.dtu.dk (O. BaltazarAndersen),
rf@space.dtu.dk ( ReneForsberg), Dieu.T.Bui@usn.no (D. Tien Bui).

https://doi.org/10.1016/j.asr.2023.04.033
0273-1177/Ó 2023 COSPAR. Published by Elsevier B.V. All rights reserved.

Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

42 1. Introduction Tatarek, 2019, Thomas, Lee, Coutts et al., 2022). The 79

SDB could provide good accuracy; however, this method 80
43 Seafloor topography plays a vital role in many different is preferable for shallow and clean water body areas 81
44 fields, including tectonic studies (Haxby, Karner, (IHO and IOC, 2018, Pe’eri, Parrish, Azuike et al., 2014). 82
45 LaBrecque et al., 1983) and activities of earthquakes and In the case of LIDAR, this technique, which can be used 83
46 tsunamis (Sepúlveda, Liu and Grigoriu, 2018). In addition, together with Autonomous Surface Vehicles (ASV) and 84
47 maps of the seafloor topography have been considered a Autonomous Underwater Vehicles (UAV), has proven 85
48 necessary document for extending the limits of the conti- it’s efficient for mapping at shallow or under-ice areas 86
49 nental shelf of coastal countries, as stated in the United and high-slope seafloors (Lucieer and Forrest, 2016). On 87
50 Nations Convention on the Law of the Sea (UNCLOS) the other hand, the LIDAR measures the sea depth by 88

of
51 (Jakobsson, Mayer and Armstrong, 2003). Besides, highly determining the different times between surface water and 89
52 accurate seafloor topography is essential for setting up sub- seafloor reflection signals; therefore, it may operate ineffec- 90
53 marine cables, pipes, oil platforms, and other underwater tively in turbidity and high-depth areas (Costa, Battista 91
54 infrastructures (Arnould, Monk, Ierodiaconou et al., and Pittman, 2009). ASV and AUV do not require the 92

o
55 2015, Kaiser, 2020). Also, seafloor topography is essential direct appearance of engineers or ships nearby the survey- 93
56 for high-resolution tidal or ocean modeling (Gonzalez, ing areas; however, the dependence on the power source 94
57 Waldman, Sannino et al., 2023) and ocean environment reduces the effective operation boundary. In addition, a 95

Pr
58 research (Hildebrand and Bellefontaine, 2017). Globally, highly qualified engineer is required when the system has 96
59 the ocean surface area covers about 362 million km2 a problem. Overall, the LIDAR technique is not widely 97
60 (Mathez and Smerdon, 2018), which accounts for approx- applied to determine the seafloor depth due to the limited 98
61 imately 71% of the earth’s surface. However, only about positioning accuracy (Wölfl, Snaith, Amirebrahimi et al., 99
62 18% of the total seafloor topography has been mapped 2019). 100
63 using shipborne depth data (Mayer et al., 2018); therefore, Regarding the satellite altimetry method, McKenzie and 101
64
65
determining the seafloor depth with high accuracy is essen-
tial.(See Table 1).
ed Bowin (1976) pointed out the relationship between gravity
anomaly and bathymetry in the Atlantic Ocean for large
102
103
66 Literature review shows that various techniques have areas. Then, Dixon, McNutt and Smith (1983) concluded 104
67 been introduced for measuring seafloor topography, i.e., that the altimetry satellite SEASAT could be used to derive 105
68 single-beam Echosounder (SBES) (Arseni, Voiculescu, the bathymetry. Smith and Sandwell (1994) showed that a 106
ct
69 Georgescu et al., 2019), multibeam echosounder (MBES) combination of satellite altimetry and shipboard data 107
70 (Calder and Mayer, 2003), satellite-derived bathymetry could improve the accuracy of the bathymetry mapping. 108
71 (SDB) (Sagawa, Yamashita, Okumura et al., 2019, As a result, the global seafloor bathymetric model was 109
72 Traganos, Poursanidis, Aggarwal et al., 2018), Light established early in 1997 using satellite altimetry and ship- 110
re

73 Detection and Ranging (LIDAR) (Liu, Sherman and Gu, board data (Smith and Sandwell, 1997). Several bathymet- 111
74 2007), and satellite altimetry (Brêda, Paiva, Bravo et al., ric models were also successfully derived using altimetric 112
75 2019, Yeu, Yee, Yun et al., 2018). Among these techniques, satellite data, including DNSC08BAT, DTU10BAT, and 113
76 the SBES and MBES can provide high accuracy, but they DTU18BAT (Becker, Sandwell, Smith et al., 2009, 114
or

77 are cost-consuming, especially for large areas. In addition, Knudsen, Andersen, Forsberg et al., 2012). In recent 115
78 it is not easy to apply in ice areas (Kruss, Wiktor and research, the DTU17GRAV gravity anomaly and echo- 116
nc

Table 1
Summary of the data used in this research.
No. Data Coverage (u: latitude, k: Number of
longitude) points
1 DTU10GRAV 7.5°N  u  14.5°N 151,200
U

109.5°E  k  115.5°E
2 DTU13GRAV 7.5°N  u  14.5°N 151,200
109.5°E  k  115.5°E
3 DTU15GRAV 7.5°N  u  14.5°N 151,200
109.5°E  k  115.5°E
4 DTU17GRAV 7.5°N  u  14.5°N 151,200
109.5°E  k  115.5°E
5 SDUST2021GRAV 7.5°N  u  14.5°N 151,200
109.5°E  k  115.5°E
6 Shipborne echo-sounding points measured in the period of 1990–1993 (having both the depth 7.5°N  u  14.5°N 33,313
value and the gravity anomaly value) 109.5°E  k  115.5°E
7 Shipborne echo-sounding points measured in the period of 2009–2016 (having the depth value 7.5°N  u  14.5°N 9,403
only) 109.5°E  k  115.5°E

2
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

117 sounder data can be combined to raise the accuracy of It is noted that although various bathymetry models, 174
118 bathymetry in the Arctic Ocean (Abulaitijiang, Andersen i.e., DTU10BAT (Technical University of Denmark - 175
119 and Sandwell, 2019) by employing the band-pass-filtering DTU), DTU18BAT (DTU), GEBCO 2020 (GEBCO 176
120 function (Smith and Sandwell, 1994). More recently, Bathymetric Compilation Group, 2020), and GEBCO 177
121 Tozer, Sandwell, Smith et al. (2019) pointed out that the 2022 (GEBCO Bathymetric Compilation Group, 2022) 178
122 accuracy of the global bathymetric mapping can be are available in many areas, however, the accuracy of the 179
123 reached ± 150 m in the deep oceans and ± 180 m between seafloor depth derived from these models is still limited. 180
124 coastlines and the continental rise with the use of satellite Thus, how to improve the accuracy of seafloor topography 181
125 altimetry data (i.e., Cryosat-2, SARAL/AltiKa, and mapping in complex areas, i.e., central Vietnam’s Sea area 182
126 Jason-2) and shipborne data. and its surroundings, is still a research question. To partly 183

of
127 To derive bathymetry, a variety of studies considered fill this gap, this research aims to assess the possibility of 184
128 the Gravity-Geologic method (GGM) to inverse bathyme- improving the accuracy of seafloor topographic mapping 185
129 try from marine gravity anomaly. This method was pro- for the central sea area of Vietnam and its surroundings 186
130 posed early to determine the seafloor bathymetry from using combined data, i.e., bathymetry data, gravity data, 187

o
131 gravity anomalies on the ocean’s surface because the differ- and satellite-derived gravity anomalies. This study area is 188
132 ence in density between the ocean water and the ocean a complex seafloor characterized by an extended continen- 189
133 layer is insignificant (Ibrahim and Hinze, 1972, Sun, tal shelf. The average depth is 500 m, and the high-slant 190

Pr
134 Ouyang and Guan, 2018). Crosby, McKenzie and Sclater continental slope is up to 4000 m. 191
135 (2006) studied the relationship between depth, age, and
136 gravity anomalies in the Pacific Ocean. The results show 2. Study area and data 192
137 that the seafloor ripples cause a gravity anomaly of about
138 30 mGal/km for the surrounding areas. The GGM was 2.1. Study area 193
139 also considered for determining the depth in the eastern
140
141 aly and shipborne data (Kwang, Yu-Shen, Jeong et al.,
ed
Japanese ocean area, using satellite-derived gravity anom- The study area is located in the sea of Vietnam and its
vicinity, covering an area of about 363,000 km2, between
194
195
142 2010), with a report that the GGM method is effective with latitudes 8°N and 14°N and longitudes 110°E and 115°E 196
143 a short wavelength under 12 km. The result is different (Fig. 1). The seafloor topography is diverse and complex 197
144 when compared with the method proposed by Smith and because this area has undergone a unique geological devel- 198
ct
145 Sandwell (1994). Therein, this method is suitable for wave- opment process (Nguyen, 2010). The seafloor topography 199
146 lengths longer than 25 km and a density contrast of 10.25 is more than 4000 m deep at the subsidence center, sur- 200
147 g/cm3. rounded by deep-sea plains interspersed with ancient conti- 201
148 Because the GGM only uses point-to-point data to nental remnants. Deep-water troughs are considered traces 202
re

149 inverse depth, not considering the basic characteristics of of ancient subduction zones. Topographical features are 203
150 gravitational effect; therefore, in recent research, An, characterized by continental shelves, continental slope sur- 204
151 Guo, Li et al. (2022) modified the GGM into the IGGM rounding the continental shelf with deep from around 205
152 method and applied it to a study area in the South China 2500 m to 3000 m, and deep depression over 4000 m (Bui 206
or

153 Sea, reporting that the accuracy at checkpoints was and Tran, 2005). 207
154 improved by approximately 17 m. In another research, The continental shelf is located at the western edge of 208
155 Wei, Guo, Zhu et al. (2021) used the HY-2A/GM- the study area, with a mean depth of about 500 m. The gen- 209
156 derived gravity anomalies and shipborne bathymetry to eral slope of the topographic surface is 0.1°  0.2°, but in 210
nc

157 predict the bathymetry over three test areas in the South some places, the terrain slope is up to 5.0°, and the conti- 211
158 China Sea with a notable conclusion that the different geo- nental slope extends from 200 m to 4000 m (Nguyen, 212
159 logical structures affect the accuracy of the GGM-based 2010). The outer boundary of the continental slope is where 213
160 bathymetry significantly. A combination of satellite- the continental crust ends and transitions into the oceanic 214
161 derived gravity anomalies and shipborne data was also crust. The average slope varies from a few degrees to 215
U

162 used to estimate the seafloor depth in the south of Alaska 10.0°. Most continental slopes are hills and low mountains 216
163 and the south of Greenland with good results. Herein, the forming a semi-plain terrain. The continental slope is 217
164 difference in density contrasts was estimated using the formed mainly by stratigraphic structures controlled by 218
165 downward continuation method (Yu-Shen, Jeong, Kwang meridian faults, buckling, and raised protrusions. 219
166 et al., 2010). Xueshuang, Xiaoyun, Running et al. (2017) The low-lying terrain in the study area is mainly dis- 220
167 used the GGM to establish the long-wavelength gravity tributed at depths of over 4000 m, with the deepest reach- 221
168 anomaly model for determining the bathymetry. The result ing 4569 m (VNHOD, 2000). The process of expanding 222
169 demonstrated that the GGM model played a vital role in the area of the seafloor was entirely due to the activity of 223
170 the accuracy of bathymetry estimation. Nevertheless, more the spreading zone in two periods, 37 million years ago 224
171 research on improving the accuracy of the seafloor depth and 17 million years ago. Deep seafloor topography 225
172 for complex areas should be carried out to have reasonable includes two main types: (i) Deep-sea accretion plains, dis- 226
173 conclusions. tributed at a 4000 m  5000 m, developed on the East Sea 227
3
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

o of
Pr
Fig. 1. Location of the study area and the shipborne echo-sounding points used.
ed
228 rift zone, located in the deep basin of the East Sea, expand- GEBCO 2020 model and the GEBCO 2022 model, the res- 256
229 ing to the northeast and narrows to the southwest; (ii) the olution is 1500  15. These models were created by the Gen- 257
230 mountain mass remains eroded on the magmatic rock with eral Bathymetric Chart of the Oceans (GEBCO) using the 258
ct
231 an altitude of over 1000 m. These mountains are dis- SRTM15Plus and other data sources (Tozer, Sandwell, 259
232 tributed alone on the surface of the abyssal plain, rising 2019). The last model (TOPO-V25.1) with a resolution of 260
233 from the bottom of about 1000 m (Bui and Tran, 2005, 10  10 was produced by Scripps Institution of Oceanogra- 261
234 Nguyen, 2010). phy (SIO), University of California San Diego (USA) 262
re

(available at https://topex.ucsd.edu/pub/global_topo_ 263

235 2.2. Data used 1min). 264

236 2.2.1. Satellite-derived gravity anomaly data

or

237 In this research, four gravity anomaly data, DTU10-

238 GRAV, DTU13GRAV, DTU15GRAV, and DTU17- 2.2.3. Shipborne echo-sounding points measured in the period 265
239 GRAV (Andersen, 2013, Andersen and Knudsen, 2014, of 1990–1993 266
240 2016, Andersen and Knudsen, 2020) were collected. These This study used 33,332 shipborne points measured using 267
nc

241 data models were computed from the altimetry measure- the SingleBeam Echosounder in three years, 1990, 1992, 268
242 ments with 10 spacing grids and provided by the Technical and 1993, from two projects (Que et al., 2008). The first 269
243 University of Denmark (DTU) (available at https://ftp. is the scientific cooperation project between the Hanoi 270

244 space.dtu.dk/pub). The SDUST2021GRA data with a grid Institute of Oceanography (Vietnam) and the Institute of 271
Oceanography (Russia). Accordingly, two survey trips 272
245 size of 10  10 were also obtained (freely available at https://
U

were carried out in 1990 and 1992 using the Gagarinsky 273
246 zenodo.org). These data were computed using Ka-band sea ship. 274
247 surface heights (SSHs) from SARAL/AltiKA satellite and The obtained data consist of coordinates, depth, gravity 275
248 Ku-band SSHs from other satellites, including HY-2A anomalies, measured times, and distances of these mea- 276
249 (Zhu, Guo, Yuan et al., 2022). sured points on the route. The second is the cooperation 277
project between Vietnam and France. The Atalante ship 278
250 2.2.2. Global seafloor depth data was used to survey the East Sea in 1993. The data from 279
251 In this research, four global seafloor depth models were the two projects above were then processed and assessed 280
252 used, DTU18BAT, GEBCO 2020, GEBCO 2022, and with accuracy is ± 1.0 mGal (Que, Dung and Tram, 281
253 TOPO-V25.1. The first model with a resolution of 10  10 2008). As a result, for each point, both the depth value 282
254 was provided by the DTU, which was established using and the gravity anomaly value were derived. Among these 283
255 satellite-derived anomaly gravity data. Regarding the 33,332 shipborne points, a total of 19 points were removed 284
4
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

285 due to raw error. As a result, the remaining 33,313 points 3. Methodology 319
286 (Fig. 1) were used in this research.
The methodological flowchart for determining the sea- 320
287 2.2.4. Shipborne echo-sounding points measured in the period floor depth in this research is shown in Fig. 3. The detailed 321
288 of 2009–2016 steps are presented in Sections 3.1 to 3.6. 322
289 This dataset consists of 9,403 points measured using the
290 SingleBeam Echosounder - Simrad EA 500 in 2009 and the
291 Multibeam Echosounder - SeaBeam 3030 and distributed 3.1. Accuracy assessment and selection of satellite-derived 323
292 between latitudes 8°N and 14°N and longitudes 110°E gravity anomaly 324
293 and 115°E (Fig. 1). Herein, the data inside the area between
In this research, 151,200 points were generated for each

of
294 latitudes 12 °N and 14 °N, and longitudes 112°450 E and 325
295 115°000 E were measured using the SingleBeam Echosoun- of the DTU10GRAV, DTU13GRAV, DTU15GRAV, 326
296 der - Simrad EA 500 in 2009. The remaining areas were DTU17GRAV, and SDUST2021GRA. These points were 327
297 measured from 2010 to 2016 using the Multibeam Echo- compared with 151,200 shipborne gravity anomaly points 328

o
298 sounder - SeaBeam 3030. Then, these data were processed to assess the gravity anomaly accuracy using Eq. (1) below: 329
330
299 following the IHO S44 survey standards to derive the final
dgi ¼ Dgalt ship
i  Dg i ; i ¼ 1; 2;    ; n ð1Þ 332
300 depth points. The coordinates of points are WGS84,

Pr
301 whereas the depths were calculated from the mean sea where n is the total of gravity points, Dgship and Dgalt are 333
i i
302 level. It is important to note that the 9,403 points have the shipborne gravity anomaly and the satellite-derived 334
303 the depth value only. Therefore, the gravity anomaly value gravity anomaly, respectively. 335
304 for these points was further determined using the satellite- Subsequently, the mean deviation (dgmean ), standard 336
305 derived gravity anomaly (see Section 4.2 below). deviation (STDÞ, and root mean square deviation (RMSDÞ 337
(Cohen and Sternberg, 1980) were derived using the follow-
306
307
2.2.5. Data preparation
For this research, a total of 42,716 shipborne points
ed ing equations:
338
339
340
308 were used, including 33,313 points measured in the period
309 of 1990–1993 and 9403 points measured in the period of
310 2009–2016. After processing, these points have both depth
ct
311 and gravity anomaly values. For analysis, these 42,716
312 shipborne points were further randomly divided into three
313 parts (Fig. 2): Part 1 consists of 14,244 points and was used
314 to compute the seafloor depth. While Part 2 contained
re

315 14,225 points and was employed for estimating the accu-
316 racy of the depth. Whereas the remaining (Part 3) with
317 14,247 points was adopted for fitting the gravity-derived
318 depth to the shipborne depth (refer to Section 3.6).
or
nc
U

Fig. 3. The methodological flowchart for determining the seafloor depth

Fig. 2. Shipborne data points in the study area. from gravity anomalies.
5
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

2 3 388
1X n C Dgship Dgship C Dgship Dgship    C Dgship Dgship
dgmean ¼ dg ð2Þ 6 1 1 1 2 1 m
7
342 n i¼1 i 6 C Dgship Dgship C Dgship Dgship    C Dgship Dgship 7
C Dgship Dgship 6
¼6 m 7 ð12Þ
7
2 1 2 2 2
343 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4    5
1 X n
2 C Dgship ship C ship    C
STD ¼ ðdgi  dgmean Þ ð3Þ m Dg
ship
Dgm Dg
ship
Dgm Dgm
ship
390
345 n  1 i¼1 1 2

346 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The matrix calculated in Eq.(1) is diagonal if covariance 391

1X n
values between points are not considered. The covariance 392
RMSD ¼ dg2 ð4Þ
348 n i¼1 i between point j and point k is calculated as follows 393
(Forsberg, 1987): 394
349 The minimun value of dgmean , STD, and RMSD is the 395
X

of
3
350 indicators to determine the best gravity anomaly model
351 for the study area. C j;k ðsÞ ¼ f ai logðDi þ ri Þ ð13Þ
i¼0 397
8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 398
352 3.2. Satellite altimetry-derived gravity anomaly improvement >
> r ¼ s2 þ D2i
> i

o
>
>
>
> Di ¼ D þ i  T
>
>
353 To improve the accuracy of satellite-derived gravity >
< a ¼ a ¼ 1
anomalies, the shipborne gravity anomaly points were used

Pr
0 3
354 ð14Þ
in two steps: >
> 1
a ¼ a 2 ¼ 3
355 >
>
>
> f ¼ C0

>
>
>
> D3 D3
356 : log 1
D D3
357 & Adjusting systematic deviation: the bias system is 0 2 400
358 removed by eliminating the mean difference (Eq.(2)) in
where s is the distance between points j and k; whereas 401
the satellite-derived gravity anomaly;
359
360
361 & Fitting gravity anomalies using the Collocation method:
ed C 0 ; D; and T are defined using an approximation of vari-
ance and empirical covariance with theoretical covariance
402
403
in Eq.(13). 404
362 gravity anomaly at point P is calculated as following
363 equations (Neiman, 2010a, Nguyen, 2012):
3.3. Calculation of seafloor depth using the Gravity-Geologic 405
ct
364 method 406
365
 T " #1  
alt
CðDg ; DgP Þ CðDgalt ; Dgalt Þ þ C Dgalt Dgalt CðDgalt ; Dgship Þ Dgalt
Dgfit ¼ : :
P ship
CðDg ; DgP Þ C T ðDgalt ; Dgship Þ CðDgship ; Dgship Þ þ C Dgship Dgship Dgship Based on the Gravity-Geologic (GG) method, the grav- 407
re

367 ð5Þ ity anomaly (Dg) is divided into two small parts: (i) the 408
368 where C is the covariance function. Other elements in Eq. short-wavelength gravity anomaly part (Dgshort ) which 409
369 (5) are as follows (Neiman, 2010a): depended on the variations of local bedrock topography, 410
370       alt   alt  and (ii) the long-wavelength gravity anomaly part (Dglong ) 411
372 C T Dgalt ; DgP ¼ C Dgalt
1 ; Dg P C Dg 2 ; DgP    C Dgk ; Dg P ð6Þ
or

373 generated using the deeper mass variations (Ibrahim and 412
      ship   ship 
375 C T Dgship ; DgP ¼ C Dgship
1 ; Dg P C Dg2 ; Dg P    C Dg m ; Dg P ð7Þ Hinze, 1972, Kwang, Yu-Shen, 2010, Xueshuang, 413
376 2  alt   alt   3
C Dg1 ; Dgalt alt
   C Dgalt alt Xiaoyun, 2017, Yu-Shen, Jeong, 2010). Explanation of 414
1 C Dg 1 ; Dg2 1 ; Dg k
6    alt   
alt 7 short-wavelength gravity and long-wavelength gravity can 415
  6 C Dgalt alt
2 ; Dg1 C Dg 2 ; Dg2
alt
   C Dgalt 2 ; Dg k
7
C Dgalt ; Dgalt ¼ 66
7
7 ð8Þ be found in (Kostoglodov, Kogan and Magnitskaya, 416
nc

4  5
 alt      1981, Phillips and Lambeck, 1980, Xiang, Wan, Zhang 417
378 C Dgk ; Dgalt alt
1 C Dg k ; Dg2
alt
   C Dgalt k ; Dg k
alt

379 2  ship     3 et al., 2017). 418

C Dg1 ; Dgship C Dgship1 ; Dg 2
ship
   C Dgship
1 ; Dgm
ship

6  ship
1
    7 The short-wavelength gravity anomaly for the points 419
  6 C Dg2 ; Dgship C Dgship2 ; Dg 2
ship
   C Dgship
2 ; Dgm
ship 7
C Dgship ; Dgship ¼ 6
6
1 7
7 that have available gravity anomaly and depth (Ibrahim 420
4  5
 ship      and Hinze, 1972, Kwang, Yu-Shen, 2010) is shown below: 421
U

C Dgk ; Dgship
1 C Dg ship
k ; Dg ship
2    C Dg ship
k ; Dg ship
m 422
381
382
ð9Þ Dgshort ¼ 2pGrðDship  Dmax Þ ð15Þ 424
2  alt   alt   3
C Dg1 ; Dgship C Dg1 ; Dgship    C Dgalt ship
1 ; Dgk
where G is the gravitational constant (G = 6.672  10-8
1 2
6  alt     
ship 7 425
  6 C Dg2 ; Dgship C Dgalt
2 ; Dg2
ship
   C Dgalt
2 ; Dgk
7
C Dgalt ; Dgship ¼ 6
6
1 7
7 ð10Þ cm3/gs2), r is the SDC between seawater and seafloor, 426
4  5
 alt ship
  alt ship
  alt ship
 Dship and Dmax are the shipborne depth and maximum depth 427
384 C Dgk ; Dg1 C Dgk ; Dg2    C Dgk ; Dgm
385 2 3 in the study area, respectively (Fig. 4). 428
C Dgalt Dgalt C Dgalt Dgalt    C Dgalt Dgalt The long-wavelength gravity anomaly is calculated 429
6 1 1 1 2 1 k
7
6 C Dgalt Dgalt C Dgalt Dgalt    C Dgalt Dgalt 7 using Eq.(16) (Kwang, Yu-Shen, 2010, Yu-Shen, Jeong, 430
C Dgalt Dgalt ¼6
6
2 1 2 2 2 k 7
7 ð11Þ
4  5
2010) as follows: 431
432
387 C Dgalt Dgalt C Dgalt Dgalt    C Dgalt Dgalt Dglong ¼ Dgship  Dgshort ð16Þ 434
k 1 k 2 k k

6
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

o of
Pr
Fig. 4. Seafloor determination from the gravity anomaly in this research.

435 At these points, which have only the gravity anomalies,

ed the calculated seafloor depth and the shipborne-derived 466
436 the long-wavelength gravity anomaly is interpolated from depth (dD) is derived using Eq. (20) below: 467
468
437 available points ðDglong Þ. The short-wavelength gravity
dDi ¼ Dgrav  Dship
i ; i ¼ 1; 2;    n ð20Þ 470
438 anomaly-derived depth (Dgrav ) can be calculated using i

439 Eqs. (17) and (18) (Kwang, Yu-Shen, 2010, Yu-Shen, The best SDC for the study area is selected based on the 471
440 Jeong, 2010) as: mean difference (dDmean ), standard deviation (STDD ), root 472
441
ct
short fit long mean square deviation (RMSDD ), and the Pearson correla- 473
443 Dg ¼ Dg  Dg ð17Þ
444 tion coefficient (r) between calculated and measured depth 474
Dgshort as formulas (Adler and Parmryd, 2010, Cohen and 475
Dgrav ¼ þ Dmax ð18Þ
446 2pGr Sternberg, 1980). Herein, the best SDC characterizes by 476
re

the dDmean , STDD , RMSDD are smallest, whereas r is largest. 477

447 Eq.(16) shows that the calculation of Dgshort and r is the 478
448 core steps for determining the seafloor depth using the GG 1X n

method. dDmean ¼ dDi ð21Þ

449 n i¼1 480
481
or

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
450 3.4. Interpolation of long-wavelength gravity anomaly 1 X n
STDD ¼ ðdDi  dDmean Þ2 ð22Þ
451 implementing the Collocation method n  1 i¼1 483
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 484
1X n
nc

452 The long-wavelength gravity anomaly at specific point P

is computed using the Collocation method (Neiman, RMSDD ¼ dD2i ð23Þ
453 n i¼1 486
454 2010a) below: 487
455 Pm  ship   grav 
    long  1 i¼1 Di  Dship mean : Di  Dgrav mean
Dglong
P ¼ C T Dglong ; Dglong
P : C Dg ; Dglong þ C D :Dglong r ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð24Þ
Pm 2 P m
ðDship  Dship ðDgrav  Dgrav
2
ð19Þ mean Þ : mean Þ
U

457 i i
i¼1 i¼1 489
458 where C and Dglong are covariance function and available
459 long-wavelength gravity anomaly matrix, respectively. where Dship grav
mean and Dmean are the mean measured depth and 490
460 The variance and covariance values are calculated using the mean depth. 491
461 Eqs. (13) and (14) in Section 3.2.
3.6. Unification of calculated seafloor depth and direct echo 492
462 3.5. Determination of suitable substance density contrast soundings 493

463 Determining the suitable SDC is essential for imple- The differences between the measured depth and the 494
464 menting the GG method (Kim, Hsiao, Kim et al., 2010), gravity anomaly-derived depth include the systematic bias 495
465 and this is an iterative process. The difference between and the stochastic element. The systematic bias was elimi- 496
7
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

531
2 3
497 nated by removing the mean values, whereas the stochastic C Dship Dship C Dship Dship    C Dship Dship
m
498 elements were fitted by using the Collocation method. 6 1 1 1 2 1
7
6 C Dship Dship C Dship Dship    C Dship Dship 7
499 Assume that k and m were numbers of gravity anomaly- C Dship Dship 6
¼6 m 7
ð33Þ
7
2 1 2 2 2

4    5
500 calculated depths Dgrav grav grav
1 ; D2 ; :::; Dk and measured depths
501
ship ship ship
D1 ; D2 ; :::; Dm , respectively. The short formula for the C Dship
m D
ship C ship ship    C ship ship
Dm D Dm Dm 533
1 2

502 depth determination at point P is shown in Eq.(25)

The variance and covariance values can be calculated in 534
503 (Neiman, 2010a, Nguyen, 2013):
Eqs. (13) and (14) in section 3.2 for the depth. 535
504
1
506 Dfit
P ¼ C TD ðC D þ CDÞ D ð25Þ
4. Results and analysis 536

of
507 The full description of Eq.(25) can be rewritten as
508 follows: 4.1. Accuracy assessment of the satellite-derived gravity 537
509 anomaly data 538
 T " #1
CðDgrav ; DP Þ CðDgrav ; Dgrav Þ þ C Dgrav Dgrav CðDgrav ; Dship Þ
Dfit
P ¼ : :
CðDship ; DP Þ C T ðDgrav ; Dship Þ CðDship ; Dship Þ þ C Dship Dship

o
To assess the accuracy of the gravity anomaly from the 539
 grav 
D five satellite-derived gravity anomaly models, DTU10- 540
Dship GRAV, DTU13GRAV, DTU15GRAV, DTU17GRAV, 541

Pr
511 ð26Þ
and SDUST2021GRAV, a total of 13,713 shipborne grav- 542
ity anomaly points were used. These points were randomly 543
512 where:
513 selected from the 33,313 shipborne gravity anomaly points 544
C T ðDgrav ; DP Þ ¼ ½C ðDgrav grav grav
1 ; DP ÞC ðD2 ; DP Þ    C ðDk ; DP Þ
mentioned in Section 2.2.3. The result is shown in Table 2. 545
The results show that dgmean is from 34.42 mGal to 546
515 ð27Þ
516
34.10 mGal, indicating that there is a systematic devia- 547
      ship   ship 

518
C D T ship
; DP ¼ C Dship
1 ; DP C D2 ; DP    C Dm ; DP
ed ð28Þ
tion between the satellite-derived gravity anomalies and
the shipborne-derived gravity anomalies. Among them,
the DTU17GRAV model provides the highest accuracy
548
549
550
519
2 grav 3 results. The STD and RMSD are ± 6.22 mGal and 551
C ðDgrav grav grav grav
1 ; D1 ÞC ðD1 ; D2 Þ    C ðD1 ; Dk Þ
grav

6 C ðDgrav ; Dgrav ÞC ðDgrav ; Dgrav Þ    C ðDgrav ; Dgrav Þ 7 ± 5.60 mGal, respectively. Therefore, this DTU17GRA 552
6 k 7
C ðDgrav ; Dgrav Þ ¼ 6 2 1 2 2 2
7 ð29Þ
ct
4  5 model is selected for the following computational steps. 553
521 C ðDgrav grav grav grav grav grav
k ; D1 ÞC ðDk ; D2 Þ    C ðDk ; Dk Þ
522 2  ship ship   ship ship   3
4.2. Improvement of the satellite-derived gravity anomaly by 554
C D1 ; D1 C D1 ; D2    C Dship ship
1 ; Dm fitting with the shipborne gravity anomaly 555
6  ship ship   ship ship   ship ship  7
re

  6 C D2 ; D1 C D2 ; D2    C D2 ; Dm 7
C Dship ; Dship ¼ 6
6
7
7
4  5 In this step, 13,713 shipborne gravity anomaly points 556
 ship ship   ship ship   ship ship 
C Dm ; D1 C Dm ; D2    C Dm ; Dm are randomly divided into two parts (Fig. 5): the first one 557
524 ð30Þ containing 12,648 points (approximately 90.0 %) was used 558
or

525 2  grav ship   grav ship   3

for fitting with the DTU17GRAV model, and the second 559
ship
C D1 ; D1 C D1 ; D2    C Dgrav
1 ; Dk one consisting of the remaining (1,065 points, about 10.0 560
6  grav ship   grav ship   grav ship  7
  6 C D2 ; D1 C D2 ; D2    C D2 ; Dk 7 %) was used for validating the result. 561
C Dgrav ; Dship ¼ 6
6
7
7 ð31Þ
4  5 The fitting process was implemented using the GPFit 562
 grav ship   grav ship   grav ship 
nc

527 C Dm ; D1 C Dm ; D2    C Dm ; Dk and GPCol1 functions in the Gravsoft program 563

528 2 3 (Forsberg and Tscherning, 2008). The result is shown in 564
C Dgrav Dgrav C Dgrav Dgrav    C Dgrav Dgrav Fig. 6 and Table 3. 565
6 1grav 1grav 1grav 2grav 1 k
7
6 C D2 D2 C D2 D2    C D2 Dgrav grav 7 The result shows that the RMSD of the DTU17GRAV 566
C Dgrav Dgrav 6
¼6 k 7 ð32Þ
 7 reduces from ± 7.25 mGal before the fitting gravity anom- 567
4 5
U

aly to ± 5.06 mGal after the fitting gravity anomaly, 568

530 C Dgrav grav C grav grav    C grav grav
D1 Dk D2 Dk Dk
k whereas the Mean of the DTU17GRAV decreases from 569

Table 2
Comparison of the satellite-derived gravity anomaly and the shipborne gravity anomaly.
No. Model dgmax (mGal) dgmin (mGal) dgmean (mGal) STD (mGal) RMSD (mGal)
1 DTU10GRAV 54.27 34.42 2.67 ± 5.90 ± 6.48
2 DTU13GRAV 54.92 34.10 2.63 ± 5.82 ± 6.38
3 DTU15GRAV 57.43 –32.26 2.72 ± 5.66 ± 6.28
4 DTU17GRAV 56.07 –33.38 2.70 ± 5.60 ± 6.22
5 SDUST2021GRAV 59.16 31.17 2.66 ± 5.68 ± 6.27

8
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

Table 3
Statistics of the DTU17GRAV gravity anomaly before the fitting gravity
anomaly and after the fitting gravity anomaly (in mGal).
No DTU17GRAV Max Min Mean STD RMSD
1 Before the fitting 23.98 9.08 4.98 ± 5.28 ± 7.25
2 After the fitting 19.66 14.34 0.28 ± 5.05 ± 5.06

4.3. Result of the suitable density contrast 574

Theoretically, it is assumed that the SDC is roughly 575

of
1.64 g/cm3, which is the mean density difference between 576
the bedrock (2.67 g/cm3) and the seawater (1.03 g/cm3). 577
However, the SDC is varied across the local geology areas. 578
Therefore, to find a suitable SDC value for the study area, 579

o
we tested SDC from 1.10 g/cm3 and 3.00 g/cm3. The selec- 580
tion of the value for each step mainly depends on the result 581

Pr
of the previous step, with the target being to minimize the 582
RMSD value and maximize the r values. The result is 583
shown in Table 4 and Fig. 7. Based on the computed 584
results, the SDC value of 1.40 g/cm3 is the best value for 585
Fig. 5. Distribution of the fitting and the validating points in this research. the study area, which minimized the error (RMSD = 586
ed ±85.54 m) and maximized the Pearson correlation coeffi- 587
cient (r = 0.997). 588

4.4. Determination of the seafloor depth from the gravity 589

anomaly corresponding to the most appropriate SDC 590
ct
4.4.1. Computation of long-wavelength gravity anomaly 591
Using the SDC value of 1.40 g/cm3 in the previous step, 592
the long-wavelength gravity anomalies at the shipborne 593
points were obtained (Fig. 8). Their maximum, minimum, 594
re

and average values for the study area were + 7.50 mGal, 595
269.30 mGal, and 166.40 mGal, respectively. 596

4.4.2. Calculation of the seafloor depth using the gravity 597

or

anomaly 598
Fig. 9 shows the seafloor depth map computed from the 599
gravity anomaly in the study area. This map was obtained 600
from 107,341 points. The maximum, minimum, and aver- 601
age values of this map were 211.68 m, 4613.78 m, and
nc

602
2708.85 m, respectively. It could be seen that these points 603
with positive depth values belong to Paracel Islands and 604
Spratly Islands. 605
U

4.4.3. Comparison of the gravity-derived seafloor depth with 606

the shipborne depth 607
It could be seen that, after fitting gravity, RMSD of the 608
gravity-derived seafloor depth reduced from ± 119.74 m to 609
± 85.54 m, and Mean decreased from 33.12 m to 1.18 m, 610
Fig. 6. The fitted gravity anomaly map for the study area.
indicating that the accuracy of the gravity-derived seafloor 611
depth has improved significantly with the use of the ship- 612
borne gravity points. 613
570 + 4.98 mGal before the fitting to + 0.28 mGal after the fit-
In order to check if the deviation between the gravity- 614
571 ting. These indicate that a significant improvement is
derived seafloor depths and the shipborne depths is statis- 615
572 obtained regarding the accuracy of the satellite-derived
tically random, an analysis is carried out on the correlation 616
573 gravity anomaly from DTU17GRAV.
between the number of the gravity-derived seafloor depth 617
9
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

Table 4
Comparison of the calculated depths based on the SDC difference.
No. SDC (g/cm3) Max (m) Min (m) Mean (m) RMSD (m) STD (m) r
1 1.10 849.28 913.69 1.46 ± 87.94 ± 87.93 0.997
2 1.20 887.22 931.70 1.35 ± 86.36 ± 86.35 0.997
3 1.30 918.81 948.45 1.26 ± 85.67 ± 85.66 0.997
4 1.40 944.29 964.04 1.18 ± 85.54 ± 85.54 0.997
5 1.50 964.90 972.93 1.13 ± 85.91 ± 85.91 0.997
6 1.67 1012.90 991.93 1.05 ± 86.54 ± 86.54 0.997
7 2.00 1075.14 1023.70 0.96 ± 87.72 ± 87.71 0.997
8 3.00 1148.49 1136.98 0.87 ± 90.75 ± 90.75 0.997

o of
Pr
ed
Fig. 7. The dependence of SDC on RMSD and r.
ct
points where the DV is less than RMSD. It is noted that 618
re

the DV is computed using Eq. (20). The result is shown 619

in Table 6. 620
It could be seen that 83.78 % of the points have DV less 621
than 1-time RMSD, whereas the percentage of the points 622
increases to 98.00 % when 3-time RMSD is considered. 623
or

In contrast, only 47 points (0.33%) have DV larger than 624

500 m (Table 6). These indicate that the DV is statistical 625
randomness. 626
Fig. 10 shows the distribution of DV in the study area. It 627
nc

could be seen that the distribution of DV follows the nor- 628

mal distribution rule (Fig. 10a) and a high correlation 629
between the computed depth and the shipborne depth 630
(Fig. 10b). 631
In order to show the location of these points regarding 632
U

the DV magnitude, a 3D map was further compiled 633

(Fig. 11). It can be observed that the points with high 634
DV are mainly in areas of complex topography and around 635
the Spratly Islands. Besides, the DV decreases significantly 636
after the fitting (Fig. 11b). 637

Fig. 8. The map of long-wavelength gravity anomaly in the study area.

10
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

645
& Step 1 - Systematic deviation adjusting. The gravity- 646
derived seafloor depth of these points was adjusted via 647
the mean value of 1.18 m (Table 5); 648
649
& Step 2 - Random deviation fitting. After removing the 650
systematic deviation, the gravity-derived seafloor depths 651
were fitted with the shipborne depths using the Colloca- 652
tion method. Because the Collocation method requires 653
the mean value of the depth to be zero, we removed 654

of
the mean depth before the fitting, and then, after the fit- 655
ting, the mean depth was restored (Fig. 12). And then, 656
the final seafloor depth map was obtained. 657
658

o
4.5.2. Comparison of the final seafloor depth map to the 659
shipborne depth 660

Pr
In order to assess the accuracy, the final seafloor depth 661
map was compared to the 14,225 shipborne depth points 662
(in Part 2 in Section 2.2.5). The difference between the ship- 663
borne depth and the seafloor depth before and after the fit- 664
ting depth is shown in Table 7. 665
The results in Table 7 indicate that the depth’s relative 666
ed accuracy (the accuracy divided by the depth) increased by
1.84%, and the RMSD reduced from ± 85.54 m to
667
668
± 83.94 m. Besides, all absolute maximum, minimum, 669
and mean values are reduced, and the Pearson correlation 670
Fig. 9. The gravity-derived seafloor depth map calculated from the gravity coefficient increases. 671
ct
anomaly.

4.5.3. Comparison of the global depth models with the 672

shipborne depth 673
638 4.5. Unification of the gravity-derived seafloor depth and the In this research, the depth from the four global models 674
re

639 shipborne depth (DTU18BAT, GEBCO 2020, GEBCO 2022, and TOPO- 675
V25.1) was further compared with the shipborne depth 676
640 4.5.1. Fitting the gravity-derived seafloor depth and the points. The result is shown in Table 7 and Fig. 13. The 677
641 shipborne depth result shows that the seafloor depth’s Max, Min, Mean, 678
or

642 The gravity-derived seafloor depth points were fitted and RMSD before and after the fitting depth are lower 679
643 with the 14,247 shipborne depth points (Part 3 in Sec- than that of the four global models, whereas r is higher 680
644 tion 2.2.5) as follows: (Table 7), indicating that the accuracy of the seafloor depth 681
maps (before the fitting depth and after the fitting depth) is 682
nc
U

Fig. 10. The map of DVs’ distribution (a) and DVs’ correlation (b).
11
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

o of
Pr
ed
ct
re
or
nc

Fig. 11. Distribution of the gravity-derived seafloor depth point with their DV: (a) before and (b) after the fitting depth.
U

Table 5
The difference between the gravity-derived seafloor depth (m) and the shipborne depth (m).
No. Gravity-derived seafloor depth Max Min Mean RMSD STD
1 Before the fitting gravity anomaly 1197.72 1061.19 33.13 ± 119.74 ± 115.06
2 After the fitting gravity anomaly 944.29 964.04 1.18 ± 85.54 ± 85.55

683 higher than those of the four global models in the study be seen that the largest differences are in the Spratly Islands 687
684 area. area, which is indicated by the red color in the map. 688
685 The distribution of the difference between the shipborne The correlation between the shipborne depth data and 689
686 depth data and six seafloor maps is shown in Fig. 13. It can the six models is further illustrated in Fig. 14. It can be seen 690

12
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

Table 6
DV of the gravity-derived seafloor depth point versus RMSD.
No. Number of points Percentage (%) DV (m)
1 11,917 83.78 RMSD < DV < RMSD
2 13,544 95.21  2  RMSD < DV < 2  RMSD
3 13,940 98.00  3  RMSD < DV < 3  RMSD
4 47 0.33 DV < 500 or DV > 500

o of
Pr
ed
ct
re

Fig. 12. The 3D map of the final seafloor depth of the study area.
or

Table 7
The difference between the shipborne depth and the seafloor depth before and after the fitting depth (the final seafloor depth).
Model Max (m) Min (m) Mean (m) RMSD (m) STD (m) r
Before the fitting depth 944.29 964.04 1.18 ± 85.54 ± 85.54 0.997
After the fitting depth (the final seafloor depth) 877.81 958.93 0.73 ± 83.96 ± 83.95 0.997
nc

DTU18BAT 2009.85 3344.26 10.92 ± 210.22 ± 209.94 0.982

GEBCO 2020 2208.10 3496.83 9.02 ± 187.99 ± 187.78 0.986
GEBCO 2022 2166.09 1815.86 4.46 ± 152.88 ± 152.82 0.991
TOPO-V25.1 1911.45 1772.79 2.61 ± 157.46 ± 157.44 0.990
U

691 that the R2 of the first two models (Fig. 14a and 14b) is 5. Discussion 699
692 0.994, which is higher than that of the four global models,
693 the DTU18BAT model (R2 = 0.964), the GEBCO 2020 Because of the mean deviation (dgmean ) between the five 700
694 model (R2 = 0.971), the GEBCO 2022 model models (DTU10GRAV, DTU13GRAV, DTU15GRAV, 701
695 (R2 = 0.981), the TOPO-V25.1 model (R2 = 0.980). These DTU17GRAV, and SDUST21GRAV) and shipborne 702
696 confirm that the accuracy of the seafloor depth maps (be- gravity data (Table 2) larger than 2.63 mGal; it could be 703
697 fore the fitting and after the fitting) is better than those said that there are systematic errors existed in these models. 704
698 of the four global models. Besides, the standard deviation ranges from ± 5.60 mGal 705
to ± 5.90 mGal. Therefore, fitting the values of these mod- 706

13
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

o of
Pr
ed
ct
re
or

Fig. 13. Maps showing the difference between the shipborne depth data and six models: (a) the seafloor depth before the fitting depth; (b) the seafloor
depth after the fitting depth (the final seafloor depth); (c) the DTU18BAT model; and (d) the GEBCO 2020 model, (e) the GEBCO 2022 model; and (f) the
TOPO-V25.1 model.
nc

707 els with shipborne gravity data is necessary, which helps to depth measurement, with the SDC is 1.40 g/cm3.Thus, to 723
708 eliminate that systematic errors and improve accuracy. increase the accuracy of the gravity depth determination, 724
709 The result of this research demonstrates that the Collo- it is necessary to improve the satellite-derived gravity 725
710 cation method (Neiman, 2010b) is a good solution for anomaly. Thus, the use of combined data from recent 726
711 reducing the systematic errors above and improving the new satellite generations (i.e., SARAL/AltiKa and 727
U

712 accuracy. Thus, the MD and RMSD of the satellite- Sentinel-6) and the old satellite generations (i.e., Jason 1, 728
713 derived gravity anomaly before the fitting are 4.98 mGal 2, and Cyosat-2) is the suitable solution. 729
714 and ± 7.25 mGal, respectively, corresponding to 33.12 m The accuracy of the gravity depth result is also 730
715 (MD) and ± 119.74 m (RMSD) of our computed gravity- strongly influenced by the SDC; therefore, it is essential 731
716 derived depth. However, after the fitting gravity anomaly, to determine the SDC properly. The finding of this work 732
717 the MD and RMSD values reduced to 0.28 mGal and denotes that the iteration process is the suitable method 733
718 ± 5.05 mGal, respectively. These correspond to 1.18 m to determine the SDC. Besides, the shipborne measure- 734
719 (MD) and ± 85,536 m (RMSD) of the gravity-derived sea- ment is expensive and time-consuming; therefore, com- 735
720 floor depth. bining the satellite-derived gravity data and shipborne 736
721 In other words, in this research, an error of ± 1.0 mGal measured data can reduce the workload of the direct 737
722 in the gravity anomaly leads to an error of ± 17.0 m in the measurement. 738

14
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

o of
Pr
ed
ct
re
or

Fig. 14. The correlation between the shipborne depth data and the six models: (a) the seafloor depth before the fitting depth; (b) the seafloor depth after
the fitting depth (the final seafloor depth); (c) the DTU18BAT model; and (d) the GEBCO 2020 model, (e) the GEBCO 2022 model; and (f) the TOPO-
V25.1 model.
nc

739 Finally, although the GEBCO 2022 provides the sea- using a combination of satellite-derived gravity anomaly 753
740 floor depth with the highest accuracy (RMSD = and shipborne data. A seafloor topographic map with 754
741 ±152.88 m) for the study area compared to those of 10 10 grid obtained by Gravity-Geology method. Based 755
742 DTU18BAT, GEBCO 2020, and TOPO-V25.1; however, on the obtained results, some conclusions can be drawn 756
U

743 the accuracy is still lower than that from our result (RMSD as follows: 757
744 = ±83.96 m). This is because the seafloor topography of
745 the study area is complex, with different coral reefs and 758
746 sinking islands that are difficult to estimate the depth from & The selection of the gravity model for calculating the sea 759
747 gravity. Thus, the inclusion of shipborne data plays a crit- flood depth mapping plays an important role. Among 760
748 ical role in the improvement of depth accuracy. the available gravity models, the DTU17GRAV model 761
is capable of providing the best result. 762
749 6. Conclusion 763
& The accuracy of the satellite-derived gravity anomaly is 764
750 This study explores the possibility of improving the improved significantly by fitting the shipborne gravity 765
751 accuracy of seafloor topographic mapping for a complex anomaly data. The RMSD value of the DTU17GRAV 766
752 region of the central coast of Vietnam and its vicinity by
15
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

767 reduces from RMSD ± 7.25 mGal to ± 5.06 mGal, and (continued)
768 the MD decreases from + 4.98 mGal to + 0.28 mGal.
Nomenclature Description
769
770 & The iteration process is a suitable method to determine dD Depth deviation
771 the SDC, and the appropriate SDC is 1.40 g/cm3 for dDmean Mean deviation of depth
772 the study area. STD Standard deviation
773 RMSD Root mean square deviation
774 & Overall, the accuracy (RMSD) of the seafloor topo- 844
775 graphic map in this research is ± 83.96 m compared to
776 the shipborne depth data. This RMSD is very much
References 845

of
777 lower than that of GEBCO 2022, RMSD =
778 ±152.88 m. Therefore, it is concluded that the combina- Abulaitijiang, A., Andersen, O.B., Sandwell, D.T., 2019. Improved Arctic 846
779 tion of satellite-derived gravity anomaly and shipborne Ocean bathymetry derived from DTU17 gravity model. Earth Space 847
780 data can improve the accuracy of seafloor topographic Sci. 6, 1336–1347. https://doi.org/10.1029/2018EA000502. 848

o
781 mapping significantly. Adler, J., Parmryd, I., 2010. Quantifying colocalization by correlation: the 849
Pearson correlation coefficient is superior to the Mander’s overlap 850
782
coefficient. Cytometry 77A, 733–742. https://doi.org/10.1002/cyto. 851

Pr
a.20896. 852
783 Declaration of Competing Interest An, D., Guo, J., Li, Z., Ji, B., Liu, X., Chang, X., 2022. Improved gravity- 853
geologic method reliably removing the long-wavelength gravity effect 854
784 The authors declare that they have no known competing of regional seafloor topography: a case of bathymetric prediction in the 855
South China Sea. IEEE Trans. Geosci. Remote Sens. 60, 1–12. https:// 856
785 financial interests or personal relationships that could have 857
doi.org/10.1109/TGRS.2022.3223047.
786 appeared to influence the work reported in this paper. Andersen, O.B., Knudsen, P., 2014. Global and Arctic Marine Gravity 858
Field From Recent Satellite Altimetry (DTU13). In: 76th EAGE 859
787 Acknowledgment
ed Conference and Exhibition. European Association of Geoscientists &
Engineers, pp. 1–5. https://doi.org/10.3997/2214-4609.20140897.
Andersen, O.B., Knudsen, P., 2016. Deriving the DTU15 Global high
860
861
862
788 This research has been supported by project code: 863
resolution marine gravity field from satellite altimetry. ESA Living
789 B2021-MDA-06. The authors would like to thank the Viet- Planet Symposium 2016. Prague, Czech Republic. 864
790 namese Ministry of Education & Training, the Hanoi Andersen, O.B., Knudsen, P., 2020. The DTU17 Global Marine Gravity 865
ct
791 University of Mining and Geology. Field: First Validation Results. In: Mertikas, S.P., Pail, R. (Eds.), 866
792 We would like to thank: (1) DTU Space, the Technical Fiducial Reference Measurements for Altimetry. Springer, Interna- 867
tional Association of Geodesy Symposia, pp. 83–87. https://doi.org/ 868
793 University of Denmark, for Gravsoft software and the data 869
10.1007/1345_2019_65.
794 on the DTU10GRAV, DTU13GRAV, DTU15GRAV,
re

Andersen, O.B. Marine Gravity and Geoid from Satellite Altimetry. 870
795 DTU17GRAV, and DTU18BAT model; (2) the GEBCO Geodetic Department, DTU - Space, Juliane Maries Vej 30, DK - 871
796 Compilation Group for providing the GEBCO 2020 and 2100, Denmark., pp. 401-51, 2013. https://doi.org/10.1007/978-3-540- 872
797 the GEBCO2022 data; (3) The ZENODO for providing 74700-0_9. 873
Arnould, J.P., Monk, J., Ierodiaconou, D., et al., 2015. Use of 874
798 SDUST2021 mode land the Scripps Institution of 875
or

anthropogenic sea floor structures by Australian fur seals: potential

799 Oceanography (SIO), University of California San Diego positive ecological impacts of marine industrial development? PLoS 876
800 (USA) for the TOPO-V25.1 model, and (4) Vietnam’s Peo- One 10, e0130581. 877
801 ple Naval Hydrographic and Oceanographic Department Arseni, M., Voiculescu, M., Georgescu, L.P., Iticescu, C., Rosu, A., 2019. 878
802 for providing the shipborne data. Testing different interpolation methods based on single beam 879
echosounder river surveying. case study: Siret River. Int. J. Geo- 880
nc

Informat. 8. https://doi.org/10.3390/ijgi8110507. 881

803 Appendix. Becker, J.J., Sandwell, D.T., Smith, W.H.F., Braud, J., Binder, B., 882
Depner, J., 2009. Global bathymetry and elevation data at 30 arc 883
Nomenclature Description seconds resolution: SRTM30_PLUS. Mar. Geod. 32, 355–371. https:// 884
doi.org/10.1080/01490410903297766. 885
Dgalt Satellite-derived gravity anomaly 886
Brêda, J.P.L.F., Paiva, R.C.D., Bravo, J.M., Passaia, O.A., Moreira, D.
U

Dgship Ship-derived gravity anomaly M., 2019. Assimilation of satellite altimetry data for effective river 887
Dgfit Fitted gravity anomaly bathymetry. Water Resour. Res. 55, 7441–7463. https://doi.org/ 888
Dglong Long-wavelength gravity anomaly 10.1029/2018WR024010. 889
Dgshort Short-wavelength gravity anomaly Bui, C.Q., Tran, T.D. Building an atlas of basic characteristics of natural 890
and environmental conditions in Vietnam and adjacent waters. State- 891
dg Gravity anomaly deviation 892
level project KC-09-02, Marine Research Program. Hanoi, Vietnam,
dgmean Mean deviation of gravity anomaly 2005 893
Dship Ship-derived depth Calder, B.R., Mayer, L.A., 2003. Automatic processing of high-rate, high- 894
Dgrav Gravity-derived depth density multibeam echosounder data. Geochem. Geophys. Geosyst. 4. 895
Dfit Fitting the gravity-derived seafloor depth https://doi.org/10.1029/2002GC000486. 896
Cohen, F.E., Sternberg, M.J.E., 1980. On the prediction of protein 897
and the shipborne depth 898
structure: the significance of the root-mean-square deviation. J. Mol.
Biol. 138, 321–333. https://doi.org/10.1016/0022-2836(80)90289-2. 899

16
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

900 Costa, B., Battista, T., Pittman, S., 2009. Comparative evaluation of Liu, H., Sherman, D., Gu, S., 2007. Automated extraction of shorelines 967
901 airborne LiDAR and ship-based multibeam SoNAR bathymetry and from airborne light detection and ranging data and accuracy assess- 968
902 intensity for mapping coral reef ecosystems. Remote Sens. Environ. ment based on Monte Carlo simulation. J. Coast. Res. 23, 1359–1369. 969
903 113, 1082–1100. https://doi.org/10.1016/j.rse.2009.01.015. https://doi.org/10.2112/05-0580.1. 970
904 Crosby, A.G., McKenzie, D., Sclater, J.G., 2006. The relationship Lucieer, V.L., Forrest, A.L., 2016. Emerging Mapping Techniques for 971
905 between depth, age and gravity in the oceans. Geophys. J. Int. 553– Autonomous Underwater Vehicles (AUVs). In: Finkl, W.C., 972
906 73, 2006. https://doi.org/10.1111/j.1365-246X.2006.03015.x. Makowski, C. (Eds.), Seafloor Mapping Along Continental Shelves: 973
907 Dixon, T.H., McNutt, M.K., Smith, S.M., 1983. Bathymetric prediction Research and Techniques for Visualizing Benthic Environments. 974
908 from SEASAT altimeter. J. Geophys. Res. Solid Earth 88, 1563–1571. Springer International Publishing, p. 14. https://doi.org/10.1007/978- 975
909 https://doi.org/10.1029/JC088iC03p01563. 3-319-25121-9_2. 976
910 Forsberg, R., 1987. A new covariance model for inertial gravimetry and Mathez, E., Smerdon, J. Greenland, Antarctica, and Sea-Level Rise. 977
911 gradiometry. J. Geophys. Res. Solid Earth 92, 1305–1310. https://doi. Climate Change. Columbia University Press, 305-30, 2018. https://doi. 978

of
912 org/10.1029/JB092iB02p01305. org/10.7312/math17282-012 979
913 Forsberg, R., Tscherning, C.C., 2008. An overview manual for the Neiman, Y.M., 2010. Covariance method in physical geodesy and 980
914 GRAVSOFT geodetic gravity field modelling programs. DTU Space. collocation. Moscow Scientific World, Moscow, Russia. 981
915 GEBCO Bathymetric Compilation Group. The GEBCO_2020 Grid - a Nguyen, V.S., 2013. Fitting the satellite-derived gravity anomaly with the 982
916 continuous terrain model of the global oceans and land. . British ship-derived gravity anomaly. Scient. Tech. J. Mining – Geol. 44, 44– 983

o
917 Oceanographic Data Centre, National Oceanography Centre, NERC, 48. 984
918 UK, 2020. https://doi.org/10.5285/a29c5465-b138-234d-e053- Nguyen, T.T. Study of geological structure in deep sea (over 200m water) 985
919 6c86abc040b9. South Vietnam as a scientific basis for searching related mineral 986

Pr
920 GEBCO Bathymetric Compilation Group. The GEBCO_2022 Grid - a resources. Final report on topic KC09.18/06-10., 2010. 987
921 continuous terrain model of the global oceans and land. NERC EDS Nguyen, V.S. Determination of gravity anomalies for Vietnamese waters 988
922 British Oceanographic Data Centre NOC, 2022. https://doi.org/ by satellite altimeter results. Moscow State University of Geodesy and 989
923 10.5285/e0f0bb80-ab44-2739-e053-6c86abc0289c Cartography, Russian Federation, pp. 160, 2012 990
924 Gonzalez, N.M., Waldman, R., Sannino, G., Giordani, H., Somot, S., Pe’eri, S., Parrish, C., Azuike, C., Alexander, L., Armstrong, A. Satellite 991
925 2023. Understanding tidal mixing at the Strait of Gibraltar: a high- remote sensing as a reconnaissance tool for assessing nautical chart 992
926 resolution model approach. Prog. Oceanogr. 212. https://doi.org/
ed adequacy and completeness. Marine Geodesy 37, 293-314, 2014. 993
927 10.1016/j.pocean.2023.102980 102980. https://doi.org/10.1080/01490419.2014.902880 994
928 Haxby, W.F., Karner, G.D., LaBrecque, J.L., Weissel, J.K., 1983. Digital Phillips, R.J., Lambeck, K., 1980. Gravity fields of the terrestrial planets: 995
929 images of combined oceanic and continental data sets and their use in long-wavelength anomalies and tectonics. Rev. Geophys. 18, 27–76. 996
930 tectonic studies. Trans. Am. Geophys. Union 64, 995–1004. https:// https://doi.org/10.1029/RG018i001p00027. 997
931 doi.org/10.1029/EO064i052p00995. Que, B.C., Dung, T.T., Tram, L., 2008. Construction of bouguer gravity 998
932 Hildebrand, L.P., Bellefontaine, N.A., 2017. Ocean Governance and anomaly map in the Vietnam Sea and adjacent area. Vietnam J. Mar. 999
933 sustainability. Ship. Oper. Manag. 231–48. https://doi.org/10.1007/ Sci. Technol. 8. https://doi.org/10.15625/1859-3097/6298. 1000
ct
934 978-3-319-62365-8_11. Sagawa, T., Yamashita, Y., Okumura, T., Yamanokuchi, T., 2019. 1001
935 Ibrahim, A., Hinze, W.J., 1972. Mapping buried bedrock topography with Satellite derived bathymetry using machine learning and multi- 1002
936 gravity. Groundwater 10, 18–23. https://doi.org/10.1111/j.1745- temporal satellite images. Remote Sens. (Basel) 11. https://doi.org/ 1003
937 6584.1972.tb02921.x. 10.3390/rs11101155. 1004
re

938 IHO, IOC. The IHO-IOC GEBCO Cook Book. International Hydro- Sepúlveda, I., Liu, P.-L.-F., Grigoriu, M.D., 2018. Probabilistic tsunami 1005
939 graphic Organization, Intergovernmental Oceanographic Commission hazard assessment in South China Sea with consideration of uncertain 1006
940 (IHO-IOC), Monaco, 2018. http://dx.doi.org/10.25607/OBP-709. earthquake characteristics. J. Geophys. Res. Solid Earth 124, 658–688. 1007
941 Jakobsson, M., Mayer, L., Armstrong, A., 2003. Analysis of data relevant https://doi.org/10.1029/2018JB016620. 1008
942 to establishing outer limits of a continental shelf under law of the sea. Smith, W.H.F., Sandwell, D.T., 1994. Bathymetric prediction from dense 1009
or

943 Int. Hydrogr. Rev. 4, 1–18 https://journals.lib.unb.ca/index.php/ihr/ satellite altimetry and sparse shipboard bathymetry. J. Geophys. Res. 1010
944 article/view/20602/23764. 99, 21803–21824. https://doi.org/10.1029/94JB00988. 1011
945 Kaiser, M.J., 2020. The Offshore Pipeline Construction Industry: Activity Smith, W.H.F., Sandwell, D.T., 1997. Global sea floor topography from 1012
946 Modeling and Cost Estimation in the US Gulf of Mexico. Gulf satellite altimetry and ship depth soundings. Science 277, 1956–1962. 1013
947 Professional Publishing. https://doi.org/10.1126/science.277.5334.1956. 1014
nc

948 Kim, K.B., Hsiao, Y.-S., Kim, J.W., Lee, B.Y., Kwon, Y.K., Kim, C.H., Sun, Z., Ouyang, M., Guan, B., 2018. Bathymetry predicting using the 1015
949 2010. Bathymetry enhancement by altimetry-derived gravity anomalies altimetry gravity anomalies in South China Sea. Geod. Geodyn. 9, 1016
950 in the East Sea (Sea of Japan). Mar. Geophys. Res. 31, 285–298. 156–161. https://doi.org/10.1016/j.geog.2017.07.003. 1017
951 https://doi.org/10.1007/s11001-010-9110-0. Thomas, N., Lee, B., Coutts, O., Bunting, P., Lagomasino, D., Fatoyinbo, 1018
952 Knudsen, P., Andersen, O.B., Forsberg, R., Roberto, S., Henning, S. L., 2022. A purely spaceborne open source approach for regional 1019
953 Satellite bathymetry and other satellite derived data. National Space bathymetry mapping. IEEE Trans. Geosci. Remote Sens. 60, 1–9. 1020
U

954 Institute, DTU-SPACE, 2012 https://doi.org/10.1109/TGRS.2022.3192825. 1021

955 Kostoglodov, V., Kogan, M., Magnitskaya, E., 1981. Isostasy of the Tozer, B., Sandwell, D.T., Smith, W.H.F., Olson, C., Beale, J.R., Wessel, 1022
956 southern Mid-Atlantic Ridge: long-wavelength and short-wavelength P., 2019. Global bathymetry and topography at 15 arc sec: SRTM15+. 1023
957 effects. J. Geophys. Res. Solid Earth 86, 7825–7841. https://doi.org/ Earth Space Sci. 6, 1847–1864. https://doi.org/10.1029/ 1024
958 10.1029/JB086iB09p07825. 2019EA000658. 1025
959 Kruss, A., Wiktor, J., Tatarek, A. Acoustic detection of macroalgae in a Traganos, D., Poursanidis, D., Aggarwal, B., Chrysoulakis, N., Reinartz, 1026
960 dynamic Arctic environment (Isfjorden, West Spitsbergen) using P., 2018. Estimating Satellite-Derived Bathymetry (SDB) with the 1027
961 multibeam echosounder. 2019 IEEE Underwater Technology (UT). google earth engine and Sentinel-2. Remote Sens. (Basel) 10, 859. 1028
962 IEEE, pp. 1-7, 2019. https://doi.org/10.1109/UT.2019.8734323. https://doi.org/10.3390/rs10060859. 1029
963 Kwang, B.K., Yu-Shen, H., Jeong, W.K., Bang, Y.L., Yi, K.K., Chang, VNHOD. Report on the results of the project: Measuring and mapping 1030
964 H.K., 2010. Bathymetry enhancement by altimetry-derived gravity the sea in the Spratly Islands area, DK1 scale 1:200000; 1:50000. 1031
965 anomalies in the East Sea (Sea of Japan). Mar. Geophys. Res. 285–98, Vietnam’s People Naval Hydrographic and Oceanographic Depart- 1032
966 2010. https://doi.org/10.1007/s11001-010-9110-0. ment, 2000 1033

17
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033
 JASR 16678 No. of Pages 18, Model 5+
2 May 2023

N. Van Sang et al. Advances in Space Research xxx (xxxx) xxx

1034 Wei, Z., Guo, J., Zhu, C., Yuan, J., Chang, X., Ji, B., 2021. Evaluating adaptive mesh. Mar. Geod. https://doi.org/10.1080/ 1049
1035 accuracy of HY-2A/GM-derived gravity data with the gravity-geologic 01490419.2017.1335257. 1050
1036 method to predict bathymetry. Front. Earth Sci. 9. https://doi.org/ Yeu, Y., Yee, J.-J., Yun, H.S., Kim, K.B., 2018. Evaluation of the 1051
1037 10.3389/feart.2021.636246. accuracy of bathymetry on the nearshore coastlines of Western Korea 1052
1038 Wölfl, A.-C., Snaith, H., Amirebrahimi, S., et al., 2019. Seafloor mapping– from satellite altimetry, multi-beam, and airborne bathymetric 1053
1039 the challenge of a truly global ocean bathymetry. Front. Mar. Sci. 283. LiDAR. Sensors 18. https://doi.org/10.3390/s18092926. 1054
1040 https://doi.org/10.3389/fmars.2019.00283. Yu-Shen, H., Jeong, W.K., Kwang, B.K., Bang, Y.L., Cheinway, H., 1055
1041 Xiang, X., Wan, X., Zhang, R., Li, Y., Sui, X., Wang, W., 2017. 2010. Bathymetry estimation using the gravity-geologic method: an 1056
1042 Bathymetry inversion with the gravity-geologic method: a study investigation of density contrast predicted by the downward contin- 1057
1043 of long-wavelength gravity modeling based on adaptive mesh. uation method. Terr. Atmos. Ocean. Sci. 22, 347–358. https://doi.org/ 1058
1044 Mar. Geod. 40, 329–340. https://doi.org/10.1080/ 10.3319/TAO.2010.10.13.01(OC). 1059
1045 01490419.2017.1335257. Zhu, C., Guo, J., Yuan, J., Li, Z., Liu, X., Gao, J., 2022. SDUST2021- 1060

of
1046 Xueshuang, X., Xiaoyun, W., Running, Z., Yang, L., Xiaohong, S., GRA: global marine gravity anomaly model recovered from Ka-band 1061
1047 Wenbin, W., 2017. Bathymetry inversion with gravity-geologic and Ku-band satellite altimeter data. Earth Syst. Sci. Data 14, 4589– 1062
1048 method: a study of long-wavelength gravity modeling based on 4606. https://doi.org/10.5194/essd-14-4589-2022. 1063
1064

o
Pr
ed
ct
re
or
nc
U

18
Please cite this article as: N. Van Sang, K. Van Long, T. Tuan Dung et al., Seafloor depth mapping of central Vietnam’s sea area and its sur-
rounding using gravity anomaly data and gravity geological method, Advances in Space Research, https://doi.org/10.1016/j.asr.2023.04.033