... | @@ -25,15 +25,17 @@ If you wish to use a different domain, you need to add a `[domain]` section to y |
... | @@ -25,15 +25,17 @@ If you wish to use a different domain, you need to add a `[domain]` section to y |
|
|
|
|
|
# Using the Domain for Data Thinning
|
|
# Using the Domain for Data Thinning
|
|
|
|
|
|
Observations whose stations are located outside the adopted domain's area are discarded even before performing clustering. Moreover, the `select` command discards all but one observation for each point of a support grid based on the domain's original grid.
|
|
Observations whose stations are located outside the adopted domain's area are discarded even before performing clustering. Moreover, for each point of a support grid based on the domain's original grid, the default behaviour of the `select` command is to discard all but one of the observations that were accepted during the clustering process.
|
|
|
|
|
|
To define the support grid used by the `select` command for the final data thinning, set, under the config file's `[commands.select]` section, the parameter `thinning_grid_coarse_factor`. This parameter should be an integer larger than zero and defines how much coarser the coarse grid is w.r.t. the original one. The default is `thinning_grid_coarse_factor=4`.
|
|
To define the support grid used by the `select` command for the post-clustering data thinning, set, under the config file's `[commands.select]` section, the parameter `thinning_grid_coarse_factor`. This parameter should be an integer larger than zero and defines how much coarser the support grid is w.r.t. the original one. **The default is `thinning_grid_coarse_factor=4`**.
|
|
|
|
|
|
|
|
**NB.:** **The removal of observations by the `select` command happens only after clustering**, at the very end of the process, right before writing the output files. **It has nothing to do with the clustering process**. Setting **`thinning_grid_coarse_factor=0` will disable this thinning.**
|
|
|
|
|
|
# Using the Domain to Reduce Memory Usage and Improve Performance
|
|
# Using the Domain to Reduce Memory Usage and Improve Performance
|
|
|
|
|
|
By default, NetAtmoQC divides the domain into smaller, regular subdomains and performs a preliminary clustering run on the separate subdomains before it performs the main, whole-domain clustering. The main run uses only those observations that were accepted at the pre-clustering step.
|
|
By default, NetAtmoQC divides the domain into smaller, regular subdomains and performs a preliminary clustering run on the separate subdomains before it performs the main, whole-domain clustering. The main clustering run uses only those observations that were accepted in the aforementioned pre-clustering step.
|
|
|
|
|
|
To control how the domain is split into subdomains, add the config `split.lon` and `split.lat` to the `[domain]` section of the config file. These two parameters should be integers, larger than zero, and, respectively, divisors of `nlon` and `nlat`. The defaults are `split.lon=3` and `split.lat=4`.
|
|
You can control how the domain is split into subdomains by adding the configs `split.lon` and `split.lat` to the `[domain]` section of your config file. These two parameters should be integers, larger than zero and, respectively, divisors of `nlon` and `nlat`. **The defaults are `split.lon=3` and `split.lat=4`**.
|
|
|
|
|
|
# Visualising the Configured Domain
|
|
# Visualising the Configured Domain
|
|
|
|
|
... | | ... | |