Description of what we want to do
Streamline Climpact, apply statistical method for constructing "heat" events on climpact output data (spatially averaged time series), and prepare data output for urban impact model
Activity
Streamline Climpact
We provide a script that provide the whole functionality of
Climpactin your plugin.apply statistical method for constructing "heat" events
We need more detail to get what are steps to get
heatevents. For example, when processing temperature and flux or radiation data from Climpact on a clipped region, the detection of a heat event could be based on specific criteria. If the spatially averaged temperature across all grid points exceeds a defined threshold of 30°C, combined with additional indicators such as specific thresholds of shortwave or longwave radiation fluxes, we can classify the event as a "heat" event. We need something like this ...prepare data output for urban impact model
Not a big deal
Please only provide us with more detail for the second step then we can go forward faster together
-
let's remove this repo and continue the further steps from the plugin below which seems to be more furnished ...
https://gitlab.dkrz.de/ch1187/plugins4freva/heat2urbanimpact
Streamline climpact
climpact's configuration table looks like this:
- shape_file
- region
- split_by
- project
- product
- experiment
- models
- time_frequency
- variable
- start
- end
- output_units
- output_file_type
- mask_type
- mask_method
- plot_map
- pre_process_module
- pre_process_function
heat2urbanimpact's configuration table looks like this (see heat2urbanimpact-test-wrapper-api.py):
- shape_file
- region
- split_by
- experiment: 2K, 3K, historical
- event
- length_of_event
- months_of_event
- impact_model
The list of input parameters is so much shorter because we agreed to concentrate on the nukleus ensemble and temperature development for the prototype of heat2urbanimpact (or we just go with the default to not confuse users). So, "not needed" input parameters are:
- project: nukleus
- product: ceu-3
- models: *
- time_frequency: 1hr
- variable: ['tas', 'ta300', 'ta500', 'ta700', 'ta950']
- start: *
- end: *
- output_units: true
- output_file_type: nc
- mask_type: none
- mask_method: centers
- plot_map: true
- pre_process_module: none
- pre_process_function: pre_proc
Intermediate data product
As a result we get an intermediate data product with dimensions 9 (ensemble members) x 5 (variables) x 262800 (time= 24 (hr) x 365 (days) x 30 (years)) for the chosen historical, 2K, or 3K experiment.
Statistical method
We are not looking for a specific heat event or a concrete number of heat or summer days in the data. For the chosen experiment (historical, 2K, or 3K) we would like to know how the temperature stratification typically evolves when it is e.g. 2 days exceptionally hot, or 5 days or 10 days. The exceptionality of the heat period is defined with the input parameter event (percentiles of temperature probability distribution) and the length of the heat period is defined with the input parameter length_of_event. To consider also the seasonality of heat periods the user can select different months with the input parameter months_of_event.
- First step: season/months filter
- Second step: analyse temperature evolution of days where the temperature (tas,ta300,ta500,ta700,ta950? all of them?) lies within e.g. the 95th percentile to put together a characteristic evolution for x days (defined by length_of_event) --> how do we do that? Pattern comparison methods? cross-correlation? Literature research is needed, too many question marks and traps here ...
The output of the plugin will be an artificially constructed time series with the dimensions 9 x 5 x 24 x 1 to 14 days. (Probably we will have to think about offering an ensemble mean or calculating more temperature levels in between ... later).
Prepare data output for urban impact model
Astrid is the expert here and can provide an example file how the input for PALM or ENVImet should look like.
moved to heat2urbanimpact#2