Offshore digital water depth model construction method based on multi-model space weighted fusion

Abstract

The invention discloses an offshore digital water depth model (DBM) construction method based on multi-model space weighted fusion, and relates to the technical field of geographic information. The method comprises the following steps of: firstly, respectively performing abnormal point elimination and reference unification on sea chart data and multi-source DBM data of a construction area; performing terrain segmentation on the region through a multi-scale segmentation method; dividing the area into a plurality of subareas by taking a mean value of the depth, the gradient and the surface cutting depth as a threshold value, and evaluating the terrain complexity of each subarea; and finally, each DBM is endowed with a weight to carry out linear weighted fusion, and the optimal weight of fusion among the DBMs in each partition is calculated by taking the minimum value of a root mean square error (RMSE) of the DBM and a measured value as a constraint, so that the optimal fusion DBM is obtained. The invention provides a method for carrying out spatial domain weighted fusion on a plurality of DBM data, which does not exist in a traditional multi-source data fusion method, and more accurate basic data can be provided for offshore marine geological disaster assessment, offshore engineering construction, marine environment protection and the like.

Description

Technical Field

[0001] The present invention relates to the field of geographic information technology applications, and in particular to a method for constructing an offshore digital water depth model based on multi-model spatial weighted fusion. Background Art

[0002] Digital Bathymetric Model (DBM) is an ordered data set that digitally describes the seafloor topography within a certain range. It is an important source of information for studying marine science, building marine infrastructure, and ensuring the safety of ship navigation. Obtaining high-quality DBM has always been a key issue in marine surveying and mapping and the application of geographic information products. Accurate DBM is of great significance in marine geological disaster assessment, offshore engineering construction, marine environmental protection, etc. First, by analyzing DBM data, the location and scale of potential disasters such as landslides and mudslides can be identified; secondly, in the construction of ports, the depth of the water area is often assessed and the port site is determined based on DBM; finally, accurate DBM is also needed to determine marine ecological sensitive areas, important ecological functional areas, and delineate marine ecological protection red lines.

[0003] There are many studies on establishing DBM with multi-source and multi-temporal bathymetric data at home and abroad. At present, most of the methods used to establish DBM with multi-source data in the world are inverse distance weighting, Kriging interpolation, spline interpolation, bilinear interpolation, etc. For example, Jakobsson et al. (2012) integrated historical bathymetric data, single beam, and multi-beam data to draw the international bathymetric map of the Arctic Ocean. Jha et al. (2013) used a multi-point geostatistical method of direct sampling to determine the value of each water depth pixel by sampling the training image and adjusting the result to available data, and then predicted and fused the high-resolution bathymetric data. At this stage, most existing studies tend to supplement regional DBM with newly acquired bathymetric data, and there are few studies on reconstructing the seabed topography using existing data sets. The accuracy of DBM data sets may change with changes in terrain conditions.

[0004] Therefore, a method based on weighted fusion of multiple DBMs in spatial domain is proposed to construct offshore DBM. Summary of the invention

[0005] Purpose of the invention: The purpose of the present invention is to solve the deficiencies in the prior art and to propose a method for constructing an offshore digital water depth model based on multi-model spatial weighted fusion, which is used to construct a high-precision offshore DBM.

[0006] Technical solution: A method for constructing an offshore digital water depth model based on multi-model spatial weighted fusion of the present invention is characterized by comprising the following steps in sequence:

[0007] (1) Obtain the nautical chart vectorized data and multi-source DBM data of the study area, remove the outliers of the nautical chart vectorized data, unify all DBMs into TIFF format, unify the spatial reference to the WGS-84 spherical coordinate system, and unify the depth datum to the theoretical depth datum;

[0008] (2) The study area was segmented using a multi-scale segmentation method, and the local variance (LV) value of the homogeneity of DBM objects under different scale parameters was calculated to determine the optimal segmentation; when the rate of change (ROC) value of LV reached its peak, the corresponding segmentation scale was optimal;

[0009] (3) Calculate four parameters: depth, terrain slope (slope), terrain roughness (R), and surface cutting depth (D);

[0010] (4) Taking the average of the four factors in (3) as the threshold, divide the seafloor topography area;

[0011] (5) Calculate the coefficient of variation (CV) of water depth in each sub-area and evaluate the topographic complexity of each sub-area;

[0012] (6) Use the spatial domain weighted average method to fuse multi-source DBM data. In each partition, input n DBM data and assign a weight w to each DBM. i , perform linear weighted average fusion on n data;

[0013] (7) Taking the chart vectorized data as the measured value, the RMSE between the fused DBM and the measured value is calculated;

[0014] (8) Traverse the input data fusion weights, determine the w with the minimum RMSE as the optimal weight, fuse and concatenate, and obtain the optimal fusion DBM;

[0015] Furthermore, the LV calculation formula in step (2) is:

[0016]

[0017] In the formula, x is the gray value of the image object; is the average gray value of the image object.

[0018] Furthermore, the slope calculation formula in step (3) is:

[0019]

[0020] Where, d x and d yare the rates of change of elevation in the x and y directions respectively.

[0021] Furthermore, the terrain roughness (R) calculation formula in step (3) is:

[0022]

[0023] In the formula, S c is the surface area of ​​the surface unit, S h is the projection on the horizontal plane.

[0024] Furthermore, the calculation formula for the surface cutting depth (D) in step (3) is:

[0025]

[0026] In the formula, is the average value of the domain class, z max It is the maximum water depth value of the domain class.

[0027] Furthermore, the CV calculation formula in step (5) is:

[0028]

[0029] In the formula, z i is the water depth in the area; is the average water depth in the area; n is the total number of water depth points in the area.

[0030] Furthermore, the original input data in step (6) is (x i ,y i ), the linear weighted average fusion formula is:

[0031]

[0032] Beneficial effects: The present invention first removes outliers and unifies benchmarks for the chart data and multi-source DBM data of the construction area; then performs terrain segmentation on the area through a multi-scale segmentation method; again, the area is divided into multiple partitions using the mean of depth, slope, and surface cutting depth as a threshold, and the terrain complexity of each partition is evaluated; finally, each DBM is given a weight for linear weighted fusion, and the RMSE minimum value of the DBM and the measured value is used as a constraint to calculate the optimal weight for the fusion between each DBM in each partition, and obtain the optimal fused DBM. The present invention provides a method for weighted fusion of multiple DBM data in the spatial domain that is not available in traditional multi-source data fusion methods, and can provide more accurate basic data for offshore marine geological disaster assessment, offshore engineering construction, marine environmental protection, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1It is the overall flow chart of the present invention;

[0034] Figure 2 This is a location map of the sea area in the example area of ​​the embodiment;

[0035] Figure 3 It is a distribution diagram of water depth points in the sample area of ​​the nautical chart of the embodiment;

[0036] Figure 4 It is the example area DBM of the embodiment;

[0037] Figure 5 is the optimal terrain segmentation scale map of the sample area of ​​the embodiment;

[0038] Figure 6 It is a topographic zoning plan map of the sample area of ​​the embodiment;

[0039] Figure 7 Statistical diagram of coefficient of variation of water depth in sample area of ​​embodiment;

[0040] Figure 8 It is a DBM weighted fusion flow chart provided by the present invention;

[0041] Fig. 9 This is the final multi-source DBM fusion result diagram of the sample area in this embodiment. DETAILED DESCRIPTION

[0042] The technical solution of the present invention is described in detail below, but the protection scope of the present invention is not limited to the embodiments.

[0043] like Figure 1 As shown, a method for constructing an offshore digital water depth model based on multi-model spatial weighted fusion of the present invention is characterized by comprising the following steps in sequence:

[0044] (1) Obtain the nautical chart vectorized data and multi-source DBM data of the study area, remove the outliers of the nautical chart vectorized data, unify all DBMs into TIFF format, unify the spatial reference to the WGS-84 spherical coordinate system, and unify the depth datum to the theoretical depth datum;

[0045] (2) The study area was segmented using a multi-scale segmentation method, and the local variance (LV) value of the homogeneity of DBM objects under different scale parameters was calculated to determine the optimal segmentation; when the rate of change (ROC) value of LV reached a peak, the corresponding segmentation scale was optimal;

[0046] (3) Calculate four parameters: depth, terrain slope (slope), terrain roughness (R), and surface cutting depth (D);

[0047] (4) Taking the average of the four factors in (3) as the threshold, divide the seafloor topography area;

[0048] (5) Calculate the coefficient of variation (CV) of water depth in each sub-area and evaluate the topographic complexity of each sub-area;

[0049] (6) Use the spatial domain weighted average method to fuse multi-source DBM data. In each partition, input n DBM data and assign a weight w to each DBM. i , perform linear weighted average fusion on n data;

[0050] (7) Taking the chart vectorized data as the measured value, the RMSE between the fused DBM and the measured value is calculated;

[0051] (8) Traverse the input data fusion weights, determine the w with the smallest RMSE as the optimal weight, fuse and concatenate them, and obtain the optimal fusion DBM.

[0052] Example:

[0053] For example Figure 2 For the sample area shown, the following steps were taken:

[0054] Step 1: Obtain the chart vectorized data and multi-source DBM data, remove the outliers of the chart vectorized data, unify all DBMs into TIFF format, unify the spatial reference to the WGS-84 spherical coordinate system, and unify the depth reference to the theoretical depth reference plane; here we take the sea area sample area on the east side of the Taiwan Strait as an example, the location of the sample area is as follows: Figure 2 As shown in the figure, the distribution data of water depth points in the sample area chart are as follows Figure 3 As shown in Figure 2, the DBM water depth data of the sample area is as follows: Figure 4 As shown, there are six DBM data sources here, from (a) to (f), they are SRTM30_PLUS V11.0, ETOPO 2022 15″, TOPO V25.1, ETOPO 2022 30″, GEBCO_2023, and SRTM15_PLUS V2.5.5;

[0055] Step 2: Use the multi-scale segmentation method to segment the study area, calculate the local variance (LV) value of the homogeneity of DBM objects under different scale parameters, and determine the optimal segmentation; when the rate of change (ROC) value of LV reaches the peak, the corresponding segmentation scale is optimal; the optimal terrain segmentation scale of the sample area is as follows: Figure 5 As shown, the shape factor and compactness are set to 0.2 and 0.5, respectively;

[0056] Step 3, calculate four parameters: depth, terrain slope (slope), terrain roughness (R) and surface cutting depth (D);

[0057] Step 4: Use the average value of the four factors in (3) as the threshold to divide the seabed terrain area; the terrain zoning scheme is as follows: Figure 6As shown, here the depth and slope mean are used as thresholds to divide the area into four regions;

[0058] Step 5: Calculate the coefficient of variation (CV) of water depth in each sub-area to evaluate the topographic complexity of each sub-area. The coefficient of variation of water depth in each sub-area is statistically shown as follows: Figure 7 As shown, the coefficient of variation of water depth in each area under different zoning schemes is different, reflecting the different topographic complexity of each area;

[0059] Step 6: Use the spatial domain weighted average method to fuse multi-source DBM data. In each partition, input n DBM data and assign weight w to each DBM. i , linear weighted average fusion is performed on n data; the process is as follows Figure 8 As shown in the figure, the n DBM data input are all recorded as DBMi, and the weight assigned to DBMi is w i , and w i The initial values ​​are

[0060] Step 7: Taking the chart vectorized data as the measured value, calculate the RMSE between the fused DBM and the measured value;

[0061] Step 8: traverse the input data fusion weights, determine the w with the minimum RMSE as the optimal weight, fuse and splice, and obtain the optimal fusion DBM; the optimal fusion weights of the six DBMs in each sub-area of ​​the sample area are shown in Table 1.

[0062] Table 1 Best fusion weights in examples

[0063]

[0064] The DBMs of each partition are fused according to the optimal fusion weight, and then the DBMs of the four partitions are spliced ​​into the optimal fusion DBM of the complete sample area. The final result is as follows: Fig. 9 As shown, by comparing with the original multi-source DBM data, it can be found that its accuracy is higher, which significantly improves the quality of DBM;

[0065] The final DBM fusion result is quantitatively evaluated using accuracy (RMSE) and Moran's index. The Moran's index calculation formula is as follows:

[0066]

[0067] Where n is the total number; w i,j is the spatial weight between elements i and j; z i is the deviation of the error value of factor i from the mean value, S o is the aggregation of all spatial weights;

[0068] The accuracy evaluation results of the fused DBM are shown in Table 2, and the Moran's index evaluation results are shown in Table 3. The fused DBM obtained by the method provided by the present invention has an RMSE of 4.572m, which is 28% higher than GEBCO 2023, the most accurate of the six DBMs. Its Moran's index is lower than that of most DBMs in Table 3, and the error space distribution is very stable. It can be seen that this method is successful and has high accuracy and practicality.

[0069] Table 2 Accuracy evaluation of DBMs in examples

[0070]

[0071] Table 3 Moran index of DBMs in examples

[0072]

[0073]

Claims

1. A method for constructing offshore digital water depth model based on multi-model spatial weighted fusion, characterized by: The following steps are included in sequence: (1) Obtain the nautical chart vectorized data and multi-source DBM data of the study area, remove the outliers of the nautical chart vectorized data, and unify the spatial reference coordinate system, depth datum and data format of all DBMs; (2) The study area was segmented using a multi-scale segmentation method, and the local variance (LV) value of the homogeneity of DBM objects under different scale parameters was calculated to determine the optimal segmentation; when the rate of change (ROC) value of LV reached its peak, the corresponding segmentation scale was optimal; (3) Calculate four parameters: depth, terrain slope (slope), terrain roughness (R), and surface cutting depth (D); (4) Taking the average of the four factors in (3) as the threshold, divide the seafloor topography area; (5) Calculate the coefficient of variation (CV) of water depth in each sub-area and evaluate the topographic complexity of each sub-area; (6) Use the spatial domain weighted average method to fuse multi-source DBM data. Input n DBM data and assign weight w to each DBM. i , perform linear weighted average fusion on n data; (7) Taking the chart vectorized data as the measured value, the RMSE between the fused DBM and the measured value is calculated; (8) Traverse the input data fusion weights, determine the w with the minimum RMSE as the optimal weight, and obtain the optimal fusion DBM.

2. The offshore digital water depth model construction method based on multi-model spatial weighted fusion according to claim 1 is characterized in that: In the step (1), the spatial reference coordinate system of all DBMs is the WGS-84 spherical coordinate system, the depth reference plane is unified as the theoretical depth reference plane, and the data format is unified as the TIFF format.

3. The offshore digital water depth model construction method based on multi-model spatial weighted fusion according to claim 1 is characterized in that: The LV calculation formula in step (2) is: In the formula, x is the gray value of the image object; is the average gray value of the image object.

4. The offshore digital water depth model construction method based on multi-model spatial weighted fusion according to claim 1 is characterized in that: The slope calculation formula in step (3) is: Where, d x and d y are the rates of change of elevation in the x and y directions respectively. The calculation formula of terrain roughness (R) is: In the formula, S c is the surface area of ​​the surface unit, S h It is the projection on the horizontal plane. The calculation formula for surface cutting depth (D) is: In the formula, is the average value of the domain class, z max It is the maximum water depth value of the domain class.

5. The offshore digital water depth model construction method based on multi-model spatial weighted fusion according to claim 1 is characterized in that: The CV calculation formula in step (5) is: In the formula, z i is the water depth in the area; is the average water depth in the area; n is the total number of water depth points in the area.

6. The offshore digital water depth model construction method based on multi-model spatial weighted fusion according to claim 1 is characterized in that: The original input data in step (6) is (x i ,y i ), the linear weighted average fusion formula is:

Images