You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have a query regarding the genesis frequency calculated during the DataProcess module and that used to determine the number of tracks in the TrackGenerator module.
When doing the Data Process step, I have set restrictToWindfieldDomain=True, so that my historical data stats all only relate to the domain I am processing. I have also tried to set my domain to cover the majority of the SP basin:
gridLimit={'xMin':130.0,'xMax':205.0,'yMin':-35.0,'yMax':0.0}
Domain and historic genesis density shown here:
I would not expect there to be very many TCs tracking through this domain that do not form there, so I would not expect the number of TC events to be very different for the genesis frequency.
During the DataProcess, in the _frequency function, the following is output:
17:50:11: Extracting annual frequency of events
17:50:11: Mean annual frequency: 21.6984126984127
17:50:11: Standard deviation: 6.7132537013173765
17:50:11: Calculating annual distribution of observations
From this function, the "frequency" file is generated.
At the start of TrackGenerator, to determine how many cyclones will be generated, the mean genesis frequency is calculated, using the CalcFrequency function from DataProcess/CalcFrequency.py. The following is output:
18:10:31: No genesis frequency specified: auto-calculating
18:10:31: Calculating annual frequency of TC events
18:10:31: Estimated annual genesis frequency for domain: 13.75
From this function, the "region_frequency" file is generated.
I am confused by the difference between the DataProcess and TrackGenerator frequency results. I have plotted up the data in the frequency and region_frequency files:
The CalcFrequency function uses the origin_year and origin_lon_lat data output during the DataProcess module. Given I have set restrictToWindfieldDomain=True, these data should only relate to TCs in my region of interest. Looking at the data in the origin_year file, I have plotted up a histogram of the frequency of years:
Allowing for the fact that region_frequency relates to season counts, while frequency/origin_year uses actual year, the values are still very different. Looking at a single year, take 1986: 33 TCs are specified for the calendar year 1986 in frequency. This would mean there would need to be about 33 TCs in the TC seasons 1985/1986 and 1986/1987 to give 33 TCs in the 1986 calendar year. In region_frequency, there are 35 in these two seasons, which is OK, but we would need to have some TCs left over for the 1987 calendar year (where there are also 23 TCs in frequency). The numbers just don't seem to match up well.
DataProcess reports the total number of events in the input dataset, regardless of the chosen track domain. TrackGenerator reports the mean number of events to be simulated in the track domain. Are you using an input track dataset that covers only the South Pacific?
I have a query regarding the genesis frequency calculated during the DataProcess module and that used to determine the number of tracks in the TrackGenerator module.
When doing the Data Process step, I have set restrictToWindfieldDomain=True, so that my historical data stats all only relate to the domain I am processing. I have also tried to set my domain to cover the majority of the SP basin:
gridLimit={'xMin':130.0,'xMax':205.0,'yMin':-35.0,'yMax':0.0}
Domain and historic genesis density shown here:
I would not expect there to be very many TCs tracking through this domain that do not form there, so I would not expect the number of TC events to be very different for the genesis frequency.
During the DataProcess, in the _frequency function, the following is output:
17:50:11: Extracting annual frequency of events
17:50:11: Mean annual frequency: 21.6984126984127
17:50:11: Standard deviation: 6.7132537013173765
17:50:11: Calculating annual distribution of observations
From this function, the "frequency" file is generated.
At the start of TrackGenerator, to determine how many cyclones will be generated, the mean genesis frequency is calculated, using the CalcFrequency function from DataProcess/CalcFrequency.py. The following is output:
18:10:31: No genesis frequency specified: auto-calculating
18:10:31: Calculating annual frequency of TC events
18:10:31: Estimated annual genesis frequency for domain: 13.75
From this function, the "region_frequency" file is generated.
I am confused by the difference between the DataProcess and TrackGenerator frequency results. I have plotted up the data in the frequency and region_frequency files:
The CalcFrequency function uses the origin_year and origin_lon_lat data output during the DataProcess module. Given I have set restrictToWindfieldDomain=True, these data should only relate to TCs in my region of interest. Looking at the data in the origin_year file, I have plotted up a histogram of the frequency of years:
Allowing for the fact that region_frequency relates to season counts, while frequency/origin_year uses actual year, the values are still very different. Looking at a single year, take 1986: 33 TCs are specified for the calendar year 1986 in frequency. This would mean there would need to be about 33 TCs in the TC seasons 1985/1986 and 1986/1987 to give 33 TCs in the 1986 calendar year. In region_frequency, there are 35 in these two seasons, which is OK, but we would need to have some TCs left over for the 1987 calendar year (where there are also 23 TCs in frequency). The numbers just don't seem to match up well.
There does not appear to be a large number of TCs moving into the region that did not form there in these two TCs seasons (note the total number of cyclones shown here agrees well with the region_frequency values):
Image from http://www.bom.gov.au/cyclone/tropical-cyclone-knowledge-centre/history/tracks/
So why is there the disparity between the numbers from DataProcess and those from TrackGenerator? Is there something else going on that I have missed?
The text was updated successfully, but these errors were encountered: