This forum is shutting down! Please post new discussions at community.naturalcapitalproject.org

Offshore Wind Wind Data Points

Hi,

I'm working with the offshore wind model for Hawaii and I'm getting results that are unexpected.  Specifically high values in areas of low wind resource.  I would like to determine if this is due to the use of global wind data referenced here: 

Elliott, D. L., C. G. Holladay, W. R. Barchet, H. P. Foote, and W. F. Sandusky. 1986. Wind energy resource atlas of the United States. DOE/CH 10093-4. Solar Technical Information Program, Richland, Washington..  

I'm wondering if there is a way to view the wind data points BIN file to compare it to the most current offshore wind analysis.  


I would also be interested to learn how the BIN files were created.

Thank you,
Matt

Comments

  • DougDoug Administrator, NatCap Staff
    edited December 2015

    Hey Matt,


    In response to being able to view / manipulate the wind data binary files, I can definitely help out with that.


    The data that we manipulate from comes in the following format:


    N,Latitude,Longitude,I,J,DataCoverage,Ramda-10m,Ramda-20m,Ramda-30m,...,Ramda-90m,Ramda-100m,...,Ramda-150m,K,WindSpd_Avg
    1,31.79444,123.7613,1,1,100,6.355385,6.858911,7.171751,...,8.092981,8.187322,...,1.905783,5.667084
    2,31.79568,123.8145,2,1,100,6.430588,6.940056,7.256615,...,8.188746,8.284203,...,0.999436,6.433274


    Where, 

    - I,J : an index

    - Ramda10m - 150m: Weibull scale parameter-Ramda for 10m to 150m winds

    - K: Weibull shape parameter-K


    We do a few formatting steps to get the data ready for the model and conversion to binary:


    1) Save / Export as CSV type
    2) Remove all columns except: 'Latitude', 'Longitude', 'Ramda-XXm' (keep all Ramda columns), 'K'
    3) Rename Ramda-10m to Ram-10m for all similar columns

    So columns should look like: 

    Latitude,Longitude,Ram-10m,Ram-20m,Ram-30m,Ram-40m,Ram-50m,Ram-60m,Ram-70m,Ram-80m,Ram-90m,Ram-100m,Ram-110m,Ram-120m,Ram-130m,Ram-140m,Ram-150m,K

    With the formatting done, we use the attached script called "wind_data_to_binary.py" to convert the CSV file to a binary file. (attached is the script saved as a text file, since python files are not allowed to be attached)

    Usually this is done on the command line like so:

        

    >> python wind_data_to_binary.py formatted_csv_data.csv output_binary.bin


    where 'formatted_csv_data.csv' in an input of the data and 'output_binary.bin' is the file you'd like to save the binary to.

    I attached a script as well that will allow for conversion back to a CSV file from the binary format. 


    >> python binary_to_csv.py binary_wind_data.bin csv_out_file.csv


    where 'binary_wind_data.bin' is the binary wind data and 'csv_out_file.csv' is the file to write the unpacked binary to (again, attached is the script as a txt file since python file can not be uploaded)


    - Doug
    Post edited by Doug on
  • DougDoug Administrator, NatCap Staff
    Do you think you could attach your file of global wind parameters? If everything checks out there, it would be great to get your AOI, and other input files to debug the high values on this end.

    Cheers,

    Doug
  • DougDoug Administrator, NatCap Staff
    I also wanted to note, that I think using the attached txt files above you can do:

    > python binary_wind_data_txt.txt in_file.bin out_file.csv

    Without needing to convert the text file back into a python extension.
  • Hi Doug,

    Thank you for your response.  This is intriguing, it looks like the model will account for the nacelle height is that correct?

    I have attached the Global wind parameters, AOI, and the output record for one of the trial runs, which should have all the parameters.

    I'll take a look at the wind files.

    Thank you,
    Matt


  • Hi Doug,

    It looks like the .bin file associated with the Offshore wind model doesn't include the last field WindSpd_Avg.  Below are the fields included after running the BIN to CSV python script you provided.

    LONG LATI Ram-010m Ram-020m Ram-030m Ram-040m Ram-050m Ram-060m Ram-070m Ram-080m Ram-090m Ram-100m Ram-110m Ram-120m Ram-130m Ram-140m Ram-150m K-010m


    The patterns in these parameters are intriguing but I have not yet found a clear association with the results.

    Thank you,
    Matt 
    1056 x 816 - 656K
  • DougDoug Administrator, NatCap Staff
    Hi Matt,

    Sorry, could you mention what you mean by "the nacelle height"? I'm not up to speed on my wind energy terminology. 

    That's right, we do not use the WindSpd_Avg column. Instead, we use the Weibull distribution to get our wind speed from the curve. The Ramda and K value are used to fit the curve to get this value.

    Thanks for sharing your data and parameters. I'll take a look at these and ask one of the model developers who has experience with the model to do the same. Hopefully we can provide helpful feedback. Thanks for using our models and to help in making them better!

    Cheers,

    Doug
  • DougDoug Administrator, NatCap Staff
    Hi Matt,

    I ran your data on my end and took a look. I recorded a maximum NPV of under 500. As far as the high positive values on this side of the islands, it seems to correspond to higher wind values. If you look at the output point shapefile, you'll see that the Ram-90m values are much higher in these ares then in the lower npv areas. Does this not correspond with your experience in the area?

    I did notice that changing the 'exponent_power_curve' parameter in the global wind data parameters CSV file does have quite the effect. When I ran the model with it set at 3, I got a high NPV of under 500. When I ran it set to 2 I got a NPV of just under 900. Quite the difference.

    I'm going to ask our main science developer of the model his thoughts and see if he can provide any insight.

    Cheers,

    Doug
  • RGriffRGriff Member
    edited December 2015
    Hi Matt,
    I just conducted a run of the offshore wind energy model using your parameters in this location. In addition to the approach Doug outlined for creating a csv from the wind data bin file, you can also view wind data for your AOI after running the model by looking in the intermediate folder and opening "wind_energy_points_from_data.shp." Inspecting this data in comparison to the outputs, it appears that there is generally high wind availability in locations with high net present value, and vice versa. Is there a location that you are particularly concerned with?

    It is important to remember that NPV is not only a function of wind energy production, but also costs, which vary spatially based on the cost of transmitting energy to the grid. However, that does not seem to be a very large issue here as all potential sites are relatively close to land. 

    The model does account for nacelle/hub height and adjusts the 10m height of wind values in wavewatch 3 data to the appropriate wind speed at hub height using the wind profile power law. The hub height can be adjusted in the wind turbine specifications csv table.

    As a general side note, the reason why the wind energy data is given as a bin file is to reduce the size of downloading a global dataset of wind speeds in the sample data that comes with InVEST. Doug and I have been talking about ways to have the wind data be in csv format to increase the transparency and flexibility of the model for users to enter their own data. We are not there yet, but hope to be in the not too distant future. 

    Rob
    Post edited by RGriff on
  • Hi Guys,

    Thank you very much for the feedback.  I've been pretty amazed by the complexity and detail associated with the Offshore wind model.  Your responses have answered the main questions and concerns I have had and led me to the conclusion that the Wind speed model which is publicly available (see attached) is not necessarily the best data for available wind resource. If you overlay the location data for turbines http://eerscmap.usgs.gov/windfarm/ , Maui and Hawaii islands match this data closely for on shore wind.  It makes sense though that in the offshore environment, distance from the islands would provide a higher proportion of time with available wind resource.

    Thank you,
    Matt
    1056 x 816 - 450K
  • Hi Matt,
    In comparing your attached map to the wavewatch 3 data, there do appear to be a few locations where wind speeds vary between the two, namely between the big island and Maui. It appears that the data you have may be of higher resolution and captures localized variation that is more difficult to extract from a global dataset. Do you happen to have a link to the raw data for that map?
    Regards,
    Rob
     
  • Just as a followup to this, the latest version of the offshore wind energy model (3.3.0) now allows the user to enter their own wind speed data. The model still comes packaged with global and Northwest Atlantic wind speed data. Please see the user's guide for guidance on preparing your own data. 
Sign In or Register to comment.