From 13c34e1dda4f5190cc6df63c1124add299f49bac Mon Sep 17 00:00:00 2001 From: Bente Tiedje <bente.tiedje@hereon.de> Date: Mon, 9 Dec 2024 11:55:00 +0000 Subject: [PATCH] streamlined README.md amnd merged with issue "Description of what we want to do " --- README.md | 95 +++++++++++++++++++++++++------------------------------ 1 file changed, 43 insertions(+), 52 deletions(-) diff --git a/README.md b/README.md index eb563ee..c7f6291 100644 --- a/README.md +++ b/README.md @@ -1,56 +1,47 @@ # heat2urbanimpact +This plugin extracts the development of the atmospheric temperature stratification during a specified heat/warm period from regional climate model data and tailors the data for the usage in urban impact models. For a selectable region of interest and period of time this plugin returns hourly temperature data at different layers/hights in a directly usable format (format specification). This plugin is based on the NUKLEUS ensemble of 3km horizontal grid. -## Work assignment - -Extension and composition of various existing plugins or program parts for the statistical processing of climate model data and barrier-free transfer of meteorological data (representative of certain climatological heat events) into an urban climate model +Note that we are developing prototype here and have to make compromises for the first development stage. ## Input: -* predefined RCM data basis: NUKLEUS ensemble (9 member) → output is calculated for all 9 members -* predefined RCM data basis - climate simulation and optionally observational data: NUKLEUS ensemble scenario runs (fixed reference 1961-1990, selectable GWL +2K, +3K), optionally: DWD gridded station data → do we need this data also to calculate/define events? → it depends on the method chosen - * choice of method: determined by whether bias-corrected data necessary/available (vertical air temperature profile) - * variant 0 (optimal) - bias-corrected data available: bias-corrected model data is used directly for picking out events and outputting data - * variant 1: bias is (proven) not significant: Procedure as variant 0 - * variant 2: bias-corrected data are available but only for daily data (mean, min, max): Difference: bias-corrected maximum - uncorrected maximum --> imprinting of this difference on the hourly values of the day under consideration or other method? Method applicable for all height levels? Method applicable for all variables? - * variant 3 (most unfavorable): bias is unknown, is judged to be significant and no bias-corrected data: The raw model data are used to derive the climate change signal (e.g. 2K-World - reference period). The climate change signal is suitably imposed on measurement data. Challenge: Derivation of vertical profiles (only measurements at a height of 2 m are available for temperature). -* predefined frequency of output: 1h -* selectable in configuration: - * scenario resp. GWL: 2K, 3K - * region, shape file, split by (see Climpact and PreprocNUK?): municipality (dropdown menu) or shapefile (file upload; the size of area must be limited to make sense for urban climate model!) → fieldmean (is OK for tasmax) - * event: Astrid is still researching how to define realistic time series of heat events using a statistical data analysis (which kind of statistics?, search of real time series of one model, which data matching the statisticsal analysis (e.g. extreme April: TX95p of 30a RCM data, ranking of x days with tasmax of the RCM April data and choose the best fitting day, with all ensemble members 9 time series results) or develop an artificial timeseries similar to Test Reference Years TRY from DWD using climate change signal of RCM output together with timeseries of measured station data?) of RCM model ensemble, e.g. - * representative summer day (this corresponds to an average summer day, tasmax median per month) - * extreme summer day (tasmax percentile per month, TX90p) - * very extreme summer day (tasmax percentile per month, TX95p) - * heat period (different length, heat wave typically at least 3d or 5d, longer period 14d relevant for indoor climate see option below) - * Period with high thermal stress (is postponed for the time being, as many variables are included in the definition of a thermal index like UTCI Universal Thermal Climate Index : tas, wind speed, air humidity, short/long-wave irradiation surface are used to calculate MRT mean radiation temperature (adaptable Fortran code available to calculate UTCI → Astrid) - * length of events: 1d - ~14d - * months of events (season is not differentiated enough): april, may, june, july, august, september - ◦ optional: impact model: PALM4U, Envimet (data format will be different for output) - -## Output: -* Data to download: 9 x timeseries (frequency: 1h, length: 1d - ~14d) of vertical temperature stratification (representative for selected event) which can be fed directly into an urban climate model -* optional add ons: - * timeseries of reference data and observational data for comparison of 2K/3K future with the past - * plot of all 9 timeseries (+ reference, observation) to visualize ensemble bandwidth (+ difference to past) - * future wish list: timeseries of additional variables: wind speed and direction (optimal: vertical profile or at least at AGS height geostrophic wind), shortwave (direct, diffuse) radiation at surface, air humidity (optimal: vertical profile), boundary conditon at ground surface (only initialisation): soil temperature and humidity - -## preliminary work plan: → will be put into gitlab at some point to manage tasks, issues etc. -1. create gitlab project -2. research background and collaboration, check existing tools/data: - * remapping und topography correction: PreprocNUK (Christoph) - * climpact with default settings to NUKLEUS-ensemble (Martin) - * calculation of heat events, meeting with Christoph, Marie, Christian, Miriam: - * PSI plugin, MHWId plugin? (Christoph, Marie) - * program to spot extreme events (Miriams PhD study?) - * KIT and BTU calculated various climate indices, also for heat waves (Christian) - * https://climpact-sci.org/indices/#index-HWM, /work/bb1203/ET-SCI - * number of individual heatwaves - * number of days that contribute to heatwaves - * length of the longest heatwave - * mean temperature of all heatwaves - * peak daily value in the hottest heatwave - * data processing for spatial and temporal format for selected impact model, there are converter COSMO and WRF to PALM4U → COSMO: [INIFOR preprocessor in Fortran](https://gitlab.palm-model.org/releases/palm_model_system/-/tree/74033244757f83ffd6452b44884b92c6c6ae1a6e/packages/dynamic_driver/inifor) (u, v, w, T, Qv, P, TS and WS Soil initial condition), [WRF_interface](https://gitlab.palm-model.org/releases/palm_model_system/-/tree/74033244757f83ffd6452b44884b92c6c6ae1a6e/packages/dynamic_driver/wrf_interface), [WRF4PALM](https://github.com/dongqi-DQ/WRF4PALM) py code -3. create plugin input mask → heat2urbanimpact_control.py -4. coding: convert heat event definitions into code (scientific core of the plugin) -5. coding: read needed data from freva data pool (see climpact; use of bias-corrected data) and consider PreProcNUK for hight correction of input data!!! -6. coding: convert (analysed/derived) RCM time series into impact model format -7. … +Input parameters for the plugin and help texts: +- **shape_file** (): "Select a geo reference file defining your region of interest, if None is selected (default), the whole region that is defined in the climate data will be taken. Note: You can either select a path on the HPC system, or a web url pointing to the file." +- **region** (selectfield): "Select a pre defined German municipality of interest. This selection has only effect if you don't chose a shape file." +- **split_by** (string): "If your selected geo reference file has multiple geometries and you whish to process each geometry separatly you can choose a split key that splits the geometries into different sub regions. The values for the split key will thereby used to distinguish the sub regions." +- **experiment** (selectfield): "Please select the model simulation of interest: historical or projection (global warming level +2K or +3K)." +- **event** (selectfield): "Please select the exceptionality of the heat/warm period by selecting percentiles of temperature probability distribution. Default is the 95th percentile." +- **length_of_event** (integer): "Please enter the number of days of the heat/warm period you would like to receive as output. An integer between 1 and 14 days would be plausible. Default is 3 days." +- **months_of_event** (string): "Please enter the month(s) when your heat/warm event is supposed to happen for your studies. E.g. enter for single month 'June' or for multiple months 'May:August'. Default is 'July'." +- **impact_model** (selectfield): "The format of the output will be tailored for the selected impact model (PALM or ENVI-met)." + +Predefined parameters for the prototype: +- project: nukleus +- product: ceu-3 +- models: all +- time_frequency: 1hr +- variable: tas, tasmin, tasmax ??? + +## Method --> still to be figured out! +### Data basis +It is still unclear if hourly bias corrected data of vertical air temperature profiles will be available. Therefore, we see three options to start the data anaylsis: +* option 0 (optimal) - bias-corrected data available: bias-corrected model data is used directly for picking out events and outputting data +* option 1: bias is (proven) not significant: Procedure as variant 0 +* option 2: bias-corrected data are available but only for daily data (mean, min, max): Difference: bias-corrected maximum - uncorrected maximum --> imprinting of this difference on the hourly values of the day under consideration or other method? Method applicable for all height levels? Method applicable for all variables? +* option 3 (most unfavorable): bias is unknown, is judged to be significant and no bias-corrected data: The raw model data are used to derive the climate change signal (e.g. 2K-World - reference period). The climate change signal is suitably imposed on measurement data. Challenge: Derivation of vertical profiles (only measurements at a height of 2 m are available for temperature). + +### "Filter out" heat/warm days and create times series of desired length +We are not looking for a specific heat/warm 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/warm, or 5 days or 10 days. The exceptionality of the heat/warm 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_. + +**We still researching how to define realistic time series of heat events using a statistical data analysis or how to develop an artificial timeseries.** + +## Output +As a result we get 9 timeseries (ensemble members) with a frequency of 1hr and a length of 1 to 14 days (selectable) of vertical temperature stratification (representative for selected heat/warm period) for the chosen historical, 2K, or 3K experiment which can be fed directly into an urban climate model. + +## Open issues: +* availability of hourly bias corrected data? +* availability of hourly topography corrected data? +* output options: + * offer plots of all 9 timeseries (+ reference, observation) to visualize ensemble bandwidth (+ difference to past) +* future wish list: + * timeseries of reference data and observation data for comparison of 2K/3K future with the past + * timeseries of additional variables: thermal stress (tas, wind speed, air humidity, short/long-wave irradiation surface), wind speed and direction (optimal: vertical profile or at least at AGS height geostrophic wind), shortwave (direct, diffuse) radiation at surface, air humidity (optimal: vertical profile), boundary conditon at ground surface (only initialisation): soil temperature and humidity \ No newline at end of file -- GitLab