2019 Volume 28 Pages 54-60
This fine resolution basin database provides basin model over 10ha sub-basins of the Sanriku coastal area located in the north west Pacific side of Japan. The data contain watershed boundary and flow path data about the target area created by use of a 10m digital elevation model (DEM). Using this data, the coverage area was increased over 721.66km2 compared to the low resolution public data set provided by the government. It increased the coastline coverage from 17% to 84% compared with the low resolution data set in 1km grid. For use in the coastal area or river discharge model, this data also contains the overall drainage basin boundary at the river mouth. The area of this survey was highly damaged by the tsunami on The 2011 Great East Japan Earthquake and post-disaster anthropogenic impact. Thus, this database will be of interest in the context of changes in land use and river discharge to assess both post-disaster anthropogenic impacts on the ecosystems.
Complete dataset is available via site: http://www.godac.jamstec.go.jp/catalog/data_catalog/metadataDisp/JAMSTEC-R_28DP01?lang=en&view=simple
The 2011 Great East Japan Earthquake occurred on 11 March 2011. It caused huge tsunamis, long shaking and ground subsidence in the East zone of Japan and resulted in destruction of coastal areas especially in the area along the pacific side of the Tohoku (North East) area of Japan. As well as to human society, the natural environment was severely damaged in many places (Yamakita et al., 2017a, 2018; Yamakita, 2018a). However, the consequence of tsunami is a natural disaster that has occurred repeatedly over a long time scale. Therefore, it is considered one of the factors characterizing the ecosystem in this area. On the other hand, construction, following the earthquake, to repair the sea walls and towns modifies the coastal environment and land use by artificially altering the environment. These changes are potentially new types of factors in the upheaval of the ecosystem that have not been seen in previous natural disasters. For this reason, consideration of broad scale human disturbance due to the earthquake and tsunami, is an important issue in the research on coastal ecosystems after the earthquake. However, such studies do not exist in broad spatial scales except for the flow of marine debris (Lebreton and Borrero, 2013).
In order to understand the impact of land-use changes on the watershed and coastal natural environment, it is common to assess river discharge effluents from changes in land use. A spatial model of the river flow is indispensable for this analysis. There are river basin model data sets available from the global level such as GDBD world basin database (Masutomi et al., 2009), Hydro SHEDS (Farr et al., 2007), and the national level, in the case of Japan. Use of such broad scale data has been successfully applied such as for the national goal of nature conservation, evaluation of ecosystem services or consideration of global water shortage (Oki and Kanae, 2006; Yamakita et al., 2017b; Yamakita, 2018a, 2018b).
However, there is a gap between the broad spatial scale management of basins using modelling with GIS databases and the local scale practical management using data of the individual rivers (Borowski and Hare, 2007; Yamazaki et al., 2017). This is caused by the resolution of data, differences in the types of data including data acquisition, and the purpose of the management. In the case of GDBD and Hydro SHED those are based on 1km resolution data which is the derivative of GTOPO30 (USGS https://lta.cr.usgs.gov/GTOPO30) and 3 arc-second data based on Shuttle Radar Topography Mission (SRTM) . Using high-resolution and accurate basin data allows us to consider the potential application of broad-scale management (such as GIS modelling approaches) in local management.
Additionally, the Sanriku area, the region affected by the 2011 tsunami, shows complex coastline geomorphology so called Ria coast area. Thus, not many rivers can be covered by broad scale (i.e lower resolution) data. Because of this situation, it is not possible to evaluate the impacts of terrestrial input in this area even if locally measured field data were available.
One of the ways to solve this problem is to have higher resolution basin data. In recent years, higher resolution digital elevation models (DEM) are more available worldwide. For example, the ASTER GDEM (ASTER-Global DEM) and AW3D-DEM (ALOS World 3D-DEM) has approximately 30m resolution and coverage all over the world (Tachikawa et al., 2011; Tadono et al., 2015). However, relying on the single satellite product contain various observational errors (Yamazaki et al., 2017). In the case of Japan, 10m resolution as public official base map which manually produced using aerial photographs and field survey are available from Geospatial Information Authority of Japan (GSI). In some locations 5m resolution DEMs are also available, however; data coverage was not thorough at the time we downloaded data. Using these data it is possible to create accurate higher resolution basin models in the coastal area. Such basin data will be applied to the analysis of the effect of river input inside a bay and the effect of land modification on those relationships.
We focused on the pacific side of the Tohoku area, located in the temperate western North Pacific ecosystems which fall within one of the most productive area in the earth (Fig.1). This area was heavily damaged by the tsunami on March 11, 2011 reached over 10m and resulted in significant ground subsidence. The impact of the post-disaster anthropogenic activities, specifically in terms of rebuilding broad areas also resulted in major disturbances.
Location of the study area, Sanriku, north east pacific side of Japan. Background image is CS stereograph of bathometry data using JTOPO 30.
We obtained a 10m digital elevation model (DEM) of this area from the digital elevation model of Base map information download service, provided by Geospatial Information Authority of Japan (GSI) (GSI, 2016). The final update was 10 May 2009. The data are the highest resolution covering the entire the study equally. Original data contains 0.1m count elevation which guarantee 1m accuracy extracted from standard topography map of 25,000 as 0.4 second grid point data. We used it under UTM N54 in 10m resolution after applying open source DEM data transformation tool for Base map information elevation Ver1.6.6 by Ecolis Inc. which internally using GeoTransform function in GDAL (Ecolis, 2016). To model the stream flow, the sinks which is a cells without no lower neighbor has to be removed (general ideas of flow analysis was explained in the instruction of GDBD; NIES, 2009). To remove sinks caused by ponds or artifacts in the DEM, we used breaching sink removal methods (Jones, 2002; Lindsay, 2016b). This method requires less modification of DEM and more accurate than the simple filling method, which was commonly used in previous studies. Compare to the simple filling methods that increase elevation values of the cells lower than surrounding area, the breaching methods connect cells which have lower elevation values by degrading higher elevation cells in between.
We also used existing stream line data from National Land Numerical Information download service provided by Ministry of Land, Infrastructure and Transport of Japan (MLIT, 2009) to condition DEM to enforce channel flows to match existing channels (burn in). To burn in the existing stream line decrement value, the distance decay value which defined by Lindsay (2016b) are specified as 100 and 0.5 successively. Thus the present stream lines have more likelihood to be stream line in the result, but when the calculation from new DEM exceed the likelihood with given weight new stream line will be selected. This framework will be the effective especially in the areas contain near flat area. We used an open-source desktop GIS and remote sensing software named Whitebox GAT for these preprocessing and latter calculations of the basins (Lindsay, 2016a, 2016c). Due to the limitation of memory we first obtained tentative watershed calculated in 40m resolution resampled DEM and 10km buffer from that watersheds are used for the calculation.
2.3 Creation of the stream line and boundary of basinsWe created boundary of basins and flow lines based on the flow direction and flow accumulation analysis, calculated by the DEM. To calculate the flow direction (flow pointer), D8 algorithm, which assign channel flow directions based on the simple 8 azimuth angles, was used (Esri, 2014; Jenson and Domingue, 1988).
Using above DEM data flow accumulation raster and stream line raster were calculated. To define stream initiation points which is the most upstream point of streamline, we adapted 10ha river sub-basin size as minimum size of the sub-basins based on visual comparison with topographic charts from the GSI.
As the river mouth points are used to build the basin boundary, we used all sink nodes (potential river mouth) calculated by the river link node algorithm in the Whitebox GAT. After the creation of river basin polygon, we also extracted area did not have river mouth in 10ha smallest basin size limitation. It was extracted as Coastal Drainage Area (CDA), which did not have an exact river mouth.
2.4 Comparison with the dataset of broader scaleTo evaluate the improved resolution of this basin model compared to the broader scale basin data provided by MLIT, we counted the overlap area of the basins. We also counted the coastline grids which did not have the river basin in 1km resolution.
8282 coastal basins which cover 721.66km2 were add as new basins (Fig.2). 3600 polygons which cover 146.45km2 were considered as the rest of the mixture of small basins. Average distances and standard deviation of these new basins from coast lines are 731.66m±946.99 and 32.71m±58.63 respectively. It was significantly close to the coast line compared to the distribution of sub-basins using the low resolution data. We eliminated one collapsed polygon in low resolution data by MLIT for this comparison.
Distribution of produced basins. Green area and right blue broad lines shows basins and stream lines in lower resolution using data obtained from the public data base. Red colored area, blue lines and black lines shows basins, stream lines and boundary of the basing, newly produced by our analysis. Black colored areas along the coast line show the remaining area which were not included in the basins. Mismatch with low resolution scale data showed by the yellow colored area and it was prevailed along the boundary of the basins from the public data base which showed by gray broad lines.
Consistency with broader scale data provided by MLIT was evaluated with our finer scale data by calculating the overlapped area and mismatch of the boundary. (Out of our 24156km2 basin area in our finer scale basins, 21225km2 overlapped with the broader scale basin data.) Among the overlapped area, 32295 area which showed under 0.01km2 area for each considered as mismatch of boundary which caused by the difference of resolution of the data or difference of mountain ridge in topography data. Although those were 41.09km2 area in total, it was only located in the boundary of the basins as showed by the red colored area in the Fig.2.
When we extract the basins within 1km rasterized grids of the coastline, the number of grids which contain basins was increased from 165 to 626 grids. Although there are still rests of the mixture of small basins over 153 grids, our data of the river basins increased their coverage from 17.48% to 83.79% of the coast line.
The data files are saved in ESRI Shapefile as vector data format and GeoTiff raster image format which are commonly used for geographic information system (GIS) software. The shapefile consists of a main file, an index file, and a dBASE table with the extension of .shp, .shx and .dbf successively. We additionally provided map projection file and metadata file of those shapefiles in .prj and .xml format. The files are projected using Universal Transverse Mercator (UTM) projection and WGS 1984 UTM Zone 54 of northern hemisphere.
3.2 List of the filesThe data provided as Shape geospatial vector data format for geographic information system (GIS) software (provided as zip compressed file). All data ware coordinated in GCS WSG 1984 and projected as UTM Zone N54.
Polygon data of overall river basin and rest area (CDA).
Polygon data of overall river basin over 10km2.
Polygon data of Sub-basin over 10km2. The data contains outside of our focused area until the 10km buffer at the mountain side.
Line data of the stream calculated by the model.
Point data of calculated river mouth. The data contains outside of our focused area until the 10km buffer at the mountain side.
The data provided as GeoTiff raster image format.
Flow accumulation raster in 10m resolution. The value means the accumulated number of upstream cells.
Flow direction in 10m resolution. The value means the unique value of the compass direction (East = 1, South East = 2, South = 4, South West = 8, West = 16, North West = 32, North = 64, North East = 128).
Extent of the data are as follows.
The data provided as Shape geospatial vector data format.
| ・tohoku002_100_05_framework.shp: | |||
| W | 140.526264 | E | 142.082284 |
| N | 40.187951 | S | 38.240858 |
| ・tohoku002_100_05_watershed.shp: | |||
| W | 140.526264 | E | 142.079877 |
| N | 40.187951 | S | 38.246872 |
| ・tohoku002_100_05_subbasin.shp: | |||
| W | 140.409148 | E | 142.079880 |
| N | 40.188132 | S | 38.237680 |
| ・tohoku002_100_05_stream_1000_selected_utm.shp: | |||
| W | 140.531252 | E | 142.079698 |
| N | 40.187366 | S | 38.247186 |
| ・tohoku002_100_05_stream_1000_sink.shp: | |||
| W | 140.422167 | E | 142.069933 |
| N | 40.18536 | S | 38.245626 |
The data provided as GeoTiff raster image format.
| ・flow_accumulation.tif: | |||
| W | 140.409113 | E | 142.180832 |
| N | 40.192186 | S | 38.202250. |
| Column 15082 Row 22021 Cell size 10×10m | |||
*Data under 0m and area apart from 10km buffer of our focus does not contain values.
| ・flow_direction.tif: | |||
| W | 140.409113 | E | 142.180832 |
| N | 40.192186 | S | 38.202250 |
| Column 15082 Row 22021 Cell size 10×10m | |||
*Data under 0m and area apart from 10km buffer of our focus does not contain values.
3.4 Details of the data tableData table of each shape files are as follows.
The database file of the shape file named "tohoku002_ 100_05_framework.dbf" has latter parameter names:
The database file of the shape file named "tohoku002_ 100_05_watershed.dbf" has latter parameter names:
The database file of the shape file named "tohoku002_ 100_05_subbasin.dbf" has latter parameter names:
The database file of the shape file named "tohoku002_ 100_05_stream_ 1000_selected_utm.dbf" has latter parameter names:
The database file of the shape file named "tohoku002_ 100_05_stream_ 1000_sink.dbf" has latter parameter names:
The database is expected to be used for research on the analysis of the effect of river input in coastal area, the effect of land modification on bays and so on.
This data is provided from Research Information and data Access Site of TEAMS (RIAS) which is the database of the JAMSTEC Tohoku Ecosystem-Associated Marine Sciences (TEAMS). <http://www.i-teams.jp/catalog/rias/; (ID:0202_R01)>
Refer to this article for using our data. This is compatible with the Creative Commons Attribution 4.0 International license (CC-BY 4.0) (https://creativecommons.org/licenses/by/4.0/).
These data belong to the Tohoku Ecosystem-Associated Marine Sciences (TEAMS) and Project Team for Analyses of Changes in East Japan Marine Ecosystems, JAMSTEC
We are grateful to Ms. Emily Egginton Skeehan for the English editing. We thank the editor and anonymous reviewers for providing valuable comments on the manuscript. This work was financially supported by the research program on Tohoku Ecosystem-Associated Marine Sciences (TEAMS) subsidized by the Ministry of Education, Culture, Sports, Science and Technology (MEXT).
