GEOG 566 Spring 2022
Submitted by: Sanjaya Paudel
Questions asked:
- Where are the hotspots for wave power in Oregon Coast?
- How does the spatial pattern of Wave Power (Dependent variable) vary with locations?
- What maybe the explanatory factors (Wave height and Peak Period) affecting the Wave Power in Oregon coast?
- How does Bathymetry of the location affect wave power?
- How does the natural and man-made features in coast affect the wave power potential?
A description of the dataset you examined, with spatial and temporal resolution and extent.
The datasets to be used in this research was downloaded from Department of Energy’s Water Power Technology Office’s (WPTO), and Marine and Hydrokinetic Toolkit (MHKit). The downloaded csv has Wave Characteristics: Significant Wave Height & Peak Period. Using these datasets, a novel Net Wave Power Assessment (WNPA) is performed which will give us the metrices of extractable wave power from each station downloaded. The dataset for bathymetry of the Oregon Coast was obtained from NOAA which was in netcdf(.nc) format. It was converted to raster using NETCDF tool in ArcGIS.
Hypotheses: predictions of patterns and processes you looked for.
The wave power varies by location because the wave characteristics responsible for wave power such as wave height and peak period varies by location.
More the distance away from coast more will be the wave power, however hotspots may arise near the coastal region too.
The man made and natural features may have affected on wave power potential.
The bathymetry of the location indirectly affects the wave characteristics such as wave height and peak period, however, the bathymetry may also affect wave power production directly.
Approaches: analysis approaches you used
I used ArcGIS Pro to visualize and quantify the spatial patterns in which I was interested. For Interpolation of my point data features I used three different approaches: Kriging, IDW and Thin plate spline. And for determining the relationship between my dependent variable and explanatory variables, I used Geographically Weighted Regression (GWR), a geoprocessing tool available in ArcGIS Pro toolbox.
I had a csv file containing 4,575 stations with coordinates, wave height, significant wave height, peak period, bathymetry, and wave power. First, I added the csv file in ArcGIS then displayed the stations using “Display X, Y Data” tool. The stations are shown in figure 1 below.
There were gaps in my datasets, I decided to use interpolation to properly visualize the wave power over this area and find the hot spots. There were different options of interpolation available in Arc tool box. I tried 3 of them and compared their results with each other to find out which interpolation technique is best suitable for my datasets. The resulted maps are shown in Figure 2 of result section. To quantify the differences between 3 different interpolation technique, I divided the total dataset into two parts: one for interpolation modeling and next one for validating. I choose 100 random stations from whole datasets and used it for validation and used rest of 4,475 stations for modeling all three interpolations. After interpolation of the surfaces, I used “Extract values by Point” tool in ArcGIS to extract the values obtained at that 100-point stations. I subtracted the true(original) value at that location with interpolated values, then calculated Root Mean Square (RMS) to see which has higher RMS. The method with less RMS is believed to be better model. The calculated RMS for 3 interpolation methods is shown in Table 1 below of Result Analysis section.
Later after viewing the interpolation result and visualizing the hotspots, I was eager to find out why there are such hotspots and what factors are driving it at that location. I then used GWR tool in ArcGIS to view the relationships with my dependent variable (wave power) and explanatory variables (wave height, peak period, bathymetry). We have to define the number of neighbors and some other parameters such as cell size for the GWR calculation. The GWR tool in ArcGIS provide us the summary report with the goodness of fit (R-squared). R square varies from 0.0 to 1.0, with higher values being preferable to higher influence. It can be thought of as the percentage of dependent variable variance that the regression model accounts for. The GWR tool also gives us the graphs showing the relationship between dependent variable and explanatory variables along with the relationship between the explanatory variables as well. The result is presented in the Figure 3 below.
Result Analysis:
Kriging had less RMS value compared to IDW and Spline method.
From the figure above, I found out that the hots-spots are around Newport and Yaquina head. The wave power potential at those locations were 5 kilowatts per meters. Close examination of Newport area showed that those region with high wave power has bathymetry around 15 to 20 meters. The close analysis of Yaquina head also showed the bathymetry to be around 15-20 meters. My next case study region was Waldport which had similar geographic feature as Newport. i.e with the mouth of river flowing to ocean). However, the wave power potential was only around 3 kilowatts per meter which is comparatively low with compared to Newport or Yaquina Head. To find out why, I compared the bathymetry of that location. Mean elevation of Waldport area was -24 m in case of Waldport but in case of Newport it was near -15 meters. However, we cannot conclude that the bathymetry of 15 meters is the only factor because in other region with same bathymetry they had wave power lower than Newport. Also, Newport region had a artificial manmade structure as Jetty controlling the flow of the river which may also be the factor for high wave power potential at Newport.
Figure 3: The graph obtained using GWR tool showing relationship between the Net Power (P_netA), Bathymetry (RASTERVALU), Peak Period (mean_peak_period) and Wave height ( mean_significant_wave_height ). The diagonal of the chart shows the histograms of each variable.
What did you learn from each of the analyses you conducted (i.e., from each exercise)?
I performed 3 different approaches in 3 different exercise. In first exercise, I compared different interpolation methods. Stochastic and deterministic, it turned out that the stochastic interpolation Kriging outperform deterministic method as IDW. I learned that the spatial autocorrelation is an important phenomenon to consider while performing analysis of the geographic datasets.
In second exercise, when performing GWR, I learned that we can calculate the relationship between the dependent variable and explanatory variables and more importantly, relationship between each variable as well to see which variables are closely related and which are not.
At last, I performed manual analysis to find out why there are hotspots in some areas and how the natural and man-made features affect the wave power. I learned that the natural and man-made feature do affect the wave power because the region in Newport where there was high potential for wave power had a man-made feature (jetty structure).
Significance. How are these results important to science? To resource managers?
The significance of this study is that it promotes the renewable and clean energy source. The wave energy generated is due to the natural phenomena which will continue until the sun and wind are prevalence in Ocean. The study is also important for the stake holders and investors who are looking for commercialization of the Wave power along the US coast.
Software learning. Your learning: what did you learn about software (a) Arc-Info, (b) GIS programming in Python, (c) programming in R, (d) Modelbuilder in Arc,or (e) other?
In this study, I used ArcGIS Pro for most of the analysis. This time new thing I learn was that we could obtain a graph showing distribution of each variables and their R-square value.
Statistics learning. What did you learn about statistics, including (a) hotspot, (b) spatial autocorrelation (including correlogram, wavelet, Fourier transform/spectral analysis), (c) cross-correlation/regression (cross-correlation, geographically weighted regression [GWR], regression trees, boosted regression trees), (d) multivariate methods (e.g., PCA, multiple component analysis), (e) other techniques (change detection/confusion matrices, other)?
Kriging was not only the best interpolation techniques among them, but Kriging was also helpful to examine the spatial autocorrelation of my dependent variables and explanatory variables using Geostatistical Wizard. It was informative to view the Semivariagram of each variable and see the resemblance of their relationships similar to obtained from GWR.
Evolving question. How did the results of each analysis lead you to change/refine your question? Write out the original question you stated at the beginning of the class, and restate the question(s) you now plan to address.
At first, I was trying to see where the hotspots of wave power are in Oregon Coast. Later after the interpolation and analysis of the results, new question evolved, why is there hotspots in some areas and what may be the factors causing it?
Future techniques. What techniques would you like to explore to answer your research questions in the future?
In future, I would like to see how the topography of the land near and away from the coast affect the wave power. The wave power is the function of wind and ocean interaction, the topography of the land may affect the wind flow strength and its direction.
