Three of the projects are on the analysis of wind and vegetation data that we will be using later on during the fire behavior course. We focus here on reading data and processing simple statistical analysis. The project consists in reading raster files (e.g. tif, netcdf) or vector files (e.g. shapefile), running GIS task and processing extracted data.
The Fire occurred in El pont de Vilomara on the 17th of July 2022. Information on the fire can be found here
A fourth project is more "FDS" orientated. Using a vegetation fire from the validation test case, we compute various fire behavior metrics using the fdsreader library to extract directly FDS data into python.
When developing your python code, keep in mind to use functions and classes when advantageous.
The deliverables are:
- the code that will be shared in a git repository (see 04-usingGit.md)
- a Short presentation that you could base on a notebook to show the main output of your code (ie answer to questions listed in the tasks section of the project description). Build your presentation on the figure that will be displayed in the notebook and when interesting highlight some specs of your code. A task request will be open en
ateneawhere you could give the link to your notebook.
objectives:
The objective of this project is to compare atmospheric data from 3 run of the atmospheric model MESONH run at 3 different resolutions: M1 (2km), M2(400m), M3(80m). The 3 models come from the same two-way nested simulation. A two-way nested simulation involves simulation models that interact with each other, allowing information to flow in both directions. This is common in atmospheric models where different scales or processes are interdependent, and each model can influence the other. Here the three model are nested with a focus on the area around el Pont de Vilomara.
We want:
- to observe the differences between the models on variable such as: wind at 10m, wind gust at 10m, temperature at 2m, relative humidity at 2m, or boundary layer height.
- And compare to weather station data near the location of the fire (see Castellnou de Bages).
tasks: a non-exhaustive list of task to perform:
- read raster data from the MESONH netcdf output files
- mask the area of the fire to extract atmospheric data in the area of the fire. - output the atmospheric variables in geopandas data frame format.
- perform comparison between M1, M2 and M3 ambient condition on the burn area.
- highlight the gain of higher resolution model on wind data.
- show difference in variance and diurnal variation.
- perform comparison between M3 and data from the weather station located in Castellnou de Bages.
data are available on andromeda at:
- MesoNH files:
/data/IMFSE/PythonCourse/MNH/ - burn area:
/data/IMFSE/PythonCourse/PontdeVilomara/burntArea/ - weather station data in Castellnou de Bages: here as reported in page 2 of the fire report. A csv file with this data copiedpasted is here
/data/IMFSE/PythonCourse/PontdeVilomara/weatherStation_CastellouDeBages.csvIt also include the location of the weather station.
objectives
The objectives are to compare high resolution simulated atmospheric data of the 16 and 17th of July 2022 from MesoNH (see project 1) against 21 years of Reanalysis data spanning from 2000 to 2021.
The Copernicus European Regional ReAnalysis (CERRA) was uploaded on andromeda, see data section below for location. 4 variables are available every 6h at 5.5km resolution for all europe:
wdir10: 10m wind direction ,si10': 10m wind speed,r2: relative humidity at 2mt2m: temperature at 2m.
The objective is to see how much the ambient condition on the day of the fire were exceptional.
tasks: a non-exhaustive list of tasks to perform:
- read MESONH and CERRA data
- resample to CERRA resolution,
- mask the zone of the fire,
- output CERRA data and MESONH data for the fire area in geopandas data frame
- compare CERRA and MESONH data and evaluate the anomaly.
data are available on andromeda at:
- MesoNH files:
/data/IMFSE/PythonCourse/MNH/ - burn area: `/data/IMFSE/PythonCourse/PontdeVilomara/burntArea/
- CERRA files:
/data/paugam/CDS/CERRA. see .py file in directory for example f how to load CERRA data.
objectives The objectives are to compare atmospheric data, cover map/topography use in the atmospheric model, and fuel/topography map data used in fire model. In particular, we are interested in plotting covariance matrix to observe variables dependence.
tasks: a non-exhaustive list of tasks to perform:
- in MesoMNH files:
- read atmospheric data (wind, temperature, relative humidity) from M3 model
- read cover map used in MESONH. you need to concatenate all coverXXX variable from the PGDM3.nc file in one map. The cover map are from the ecoclimap dataset (see here for the list of the cover map).
- read topography and compute aspect and slope. use topography from both the MesoNH file (srtm 250m) and the high resolution 30m srtm file (see data section below for where to find the data).
- read data that will be used in fire model. The map are available in LCP tif file, see data section below.
- resample LCP data to MESONH grid using rasterio.warp.reproject, see example in 05a-GIS-rasterio.ipynb
- mask the burn area
- output in a geopandas data frame for all pixel in the burn area:
- from MESONH: cover map , altitude, slope, atmospheric data
- from the fuel model (LCP file): fuel model (Scott and Burgan 2005), altitude, slope
- study covariance between MESONH and LCP data.
- study covariance of atmospheric data in the burn zone with fuel, cover map and orography data.
- use the
seabornlibrary with for example pairplot and heatmap function that can be applied to pandas data frame.
data are available on andromeda at:
- MesoNH files:
/data/IMFSE/PythonCourse/MNH/ - topography:
/data/IMFSE/PythonCourse/LCP/srtm30m_pdV.tifis a high resolution (30m) terrain map for the same zone as in the MesoNH files. The MesoNH terrain information input are based on the 250m product from SRTM. - burn area:
/data/IMFSE/PythonCourse/PontdeVilomara/burntArea/ - LCP data:
/data/IMFSE/PythonCourse/LCP/lcp_pgd80.tifsee.pyfile in same directory to loadlcp_pgd80.tif
objectives: the objectives are to load fds device and slice files to run analysis directly in python. 3 simulation are available on andromeda, see data section below.
tasks: a non-exhaustive list of tasks to perform:
- read FDS simulation using
fdsreader. see 06-fdsreader.ipynb for example of how to usefdsreader. - From the device data (see device
ID='x'andID='D'), compute the velocity of the flame front, or as we will see later in the fire behavior course the Rate Of Spread (ROS) along the center line of the front. Compare with results from the validation guide page 1102-1103. - Using slice of
HRRPUVavailable adt=5s, develop an algorithm to compute 2D map of arrival time. In this map, each pixel mark the time of arrival of the fire front as observed in the slice file. To get a continuous map you need to interpolate between available points (seescipyfunction griddata ). - From the arrival time map, compute ROS.
- perform the 1D and 2D ROS calculation above for the 3 simulations.
- compare ROS between 1D and 2D, and in between simulation (i.e. effect of wind and fuel load).
data: FDS simulations are on andromeda at /data/IMFSE/PythonCourse/FDS/. Three simulations are available. There are one reference simulation, the same as the one in the validation guide and two others with different wind speed and fuel load:
Case_C064_Validation677- the reference simulation.Case_C064_Validation677_windPlusCase_C064_Validation677_fuelLoadPlus