| Suisun Bay and Delta Bathymetry |
|
| United States Geological Survey |
| Foxgrover, A., Smith, R. E., and Jaffe, B. E. |
|
|
In order to represent many of the fine channel features of the Delta we have developed 10-meter horizontal resolution grid. While the underlying data is very dense in some areas (over 130 soundings per hectare near Roe Island) in most of the Delta a 10 m resolution grid requires interpolation and extrapolation over tens of cells. Simple interpolation techniques produce grossly inaccurate bathymetry models. The web of interlocking channels and varying sounding densities produce unusable results when simple interpolation or distance-weighted gridding techniques are used. See how poorly the bathymetry grids in figure 10(a) and 10(b) agree with the shore outlined channel.
Figure 9 shows example input data (soundings, shoreline, and contours) for a complex channel. In figure 10, we compare results for a simple linear interpolation with a distance-weighted interpolation. The resulting grids are fragmented. Although they correlate well with individual soundings, they do not maintain channel integrity.
|
 |
| Figure 9 - Shoreline, contours and soundings for a complex channel. |
|
| By adding contours as described in the methods section and using a routine optimized for natural drainage systems (ArcInfo - TOPOGRID), we are able to produce a much more realistic model of bathymetry. |
 |
| Figure 10 - (a) IDW - Distance weighted interpolation (b) TIN - Linear interpolation between sounding (c) Topogrid - best fit to contours and soundings. |
|
|
We can evaluate the accuracy of the depth model by comparing the grid cell value with the underlying soundings. Differences are computed only for those cells where soundings exist. The table below summarizes this comparison.
|
 |
| Figure 11 - Sub regions of the Delta. |
| Area |
Number of Soundings |
Mean Error (ft) |
Standard Deviation (ft) |
Mean Error (m) |
Standard Deviation (m) |
| North |
266265 |
0.02 |
2.3 |
0.006 |
0.70 |
| Central |
181727 |
0.21 |
2.4 |
0.064 |
0.74 |
| South |
123736 |
0.30 |
1.8 |
0.092 |
0.54 |
| Suisun |
355214 |
0.10 |
1.9 |
0.030 |
0.58 |
| All |
973468 |
0.13 |
2.2 |
0.040 |
0.67 |
|
| Another method for assessing the accuracy of the grid fit, especially in the Delta channels, is to compare cross sections produced by the point soundings with the grid estimates. Here are some examples taken from a range of channel geometries. Although cross-sections were selected to represent a range of channel size and geometry, they were constrained to areas where point data exists. As shown in figure 10 comparing the grid with a strict interpolation in areas without soundings will only demonstrates the inadequacies of the TIN. See figure 12 for the location of the cross sections. |
 |
| Figure 12 - Location of cross sections used to compare soundings and the grid. |
| Channel Type |
Width (m) |
Click on image |
Grid C/S Area (sq m) |
Soundings C/S Area (sq m) |
Difference (%) |
| (1) Suisun Bay |
1800 |
|
18098 |
18092 |
0.03 |
| (2) Broad channel |
850 |
|
6114 |
6221 |
1.7 |
| (3) Medium channel complex soundings |
400 |
|
3912 |
3961 |
1.2 |
| (4) Complex channel |
200 |
|
618 |
530 |
15 |
| (5) Small channel |
100 |
|
402 |
431 |
7.0 |
| (6) Small channel pattern soundings |
100 |
|
524 |
449 |
15 |
| (7) Only cross section soundings |
120 |
|
517 |
558 |
7.6 |
| (8) Narrow channel |
35 |
|
71 |
54 |
27 |
|
| To generate a cross section at a specific location, go to the Cross Section Tool. |
 |
| Top |
|