Jcampinsp patch 3 (#40)

* Rename plotvarbcpred.py to plotvarbccoeff.py

* Update plotvarbccoeff.py

* Create varbccoeff.md

* Update make.jl

* Update CMakeLists.txt

* Update varbccoeff.md

* Update varbccoeff.md

* Update README.md

---------

Co-authored-by: Eoin Whelan <[email protected]>

README.md

@@ -71,15 +71,15 @@ Example:
 varbcdiagnose -i /path/to/directory/with/varbc_files/ -o varbc_diag_out
 ```
 
-#### VarBC predictor time-series plotting
+#### VarBC coefficient time-series plotting
 Help/usage:
 ```
-plotvarbcpred -h
+plotvarbccoeff -h
 ```
 
 Example:
 ```
-plotvarbcpred -i varbc_diag_out/VARBC_3_16_3309_210000 -b
+plotvarbccoeff -i varbc_diag_out/VARBC_3_16_3309_210000 -b
 ```
 
 ### DFS

doc/make.jl

@@ -2,11 +2,11 @@ using Documenter
 
 
 pages = [
-   "jbdiagnose" => "jbdiagnose.md"
+   "Jb diagnose" => "jbdiagnose.md",
    "cv header list" => "cvheaderlist.md"
+   "VarBc Coeff" => "varbccoeff.md"
 ]
 
-
 prettyurls = get(ENV, "CI", nothing) == "true" 
 
 format = Documenter.HTML(prettyurls = prettyurls)   

doc/src/varbccoeff.md

@@ -0,0 +1,58 @@
+# VarBC coefficient diagnostic tool
+
+
+## 1. Purpose
+
+This diagnostic tool is intended for use in the evaluation of the performance of the variational bias correction (VarBC) used for satellite observations in the context of HARMONIE-AROME data assimilation systems. In particular for evaluating the performance in time of the VarBC coefficients.
+
+
+## 2. Content
+
+The tool comprises two scripts:
+
+- **varbcdiagnosis** for data extraction
+
+- **plotvarbccoeff** for VarBC coefficient time-series plotting 
+
+## 3. Usage
+
+**varbcdiagnosis** extracts VarBC coefficients for a certain satellite/instrument and all the channels with data and for 
+a certain period of time. VarBC coefficients are extracted from VARBC.cycle files.
+
+Script **varbcdiagnosis** arguments:
+
+        -i input-directory 
+           where VARBC.cycle files are stored, usually in the form ./EXP/YYYY/MM/DD/HH (EXPERIMENT, Year, Month, Day and                Hour)
+
+        -o output-directory
+           
+        -S list-sat
+           colon separted list of satellites to process. For example, -S 3:4:5 to process Metop-A, Metop-B and Metop-C. 
+           See https://apps.ecmwf.int/odbgov/satelliteidentifier/ for more etails.
+
+        -s list-sen
+           colon separted list of sensors to process. For example, -s 3:15 to process AMSU-A and MHS. 
+           See https://apps.ecmwf.int/odbgov/sensor/ for more details.
+
+        -h Help! Print usage information.
+
+
+**plotvarbccoeff** produces a plot for a selected satellite, sensor, channel and hour (optional, if not present all the assimilation cycles with observations are included). 
+
+Script **plotvarbccoef** arguments:
+
+       -h, --help  show this help message and exit
+       -i IPATH    Input file name
+       -l LLOC     Legend location using matplotlib syntax
+       -d LABELS   Optional experiment description
+       -b          Batch mode, produce png only
+
+For example for METOP-B (3), MHS (15), Channel 9 at 21 UTC (or at all assimilation cycles with data), IPATH is
+output-directory/VARBC_3_15_9_210000 (or output-directory/VARBC_3_15_9).
+
+
+## 4. Interpretation
+
+Time-series of VarBC coefficients can be used for monitoring the performance of variational bias correction, specially at the initial steps of the bias correction, when usually initial bias is not corrected (cold-start) and satellite observations are assimilated in passive. Coefficients usually stabilize in time (it depends in many factors), and when they are stable it is assumed that satellite observations should be evaluated for active assimilatedo.
+
+

scripts/CMakeLists.txt

@@ -10,7 +10,7 @@ endfunction()
 
 install_scripts (jbdiagnose.sh plotjbbal.py plotjbdiag.py)
 install_scripts (cv_header_list.sh)
-install_scripts (varbcdiagnose.sh plotvarbcpred.py)
+install_scripts (varbcdiagnose.sh plotvarbccoeff.py)
 install_scripts (rundiacov.sh diag_diacov.sh plotdiacov.py)
 install_scripts (dfscomp.sh plotdfs.py)
 #install_scripts (diacov.sh festat.sh)

scripts/plotvarbcpred.py → scripts/plotvarbccoeff.py

@@ -133,7 +133,7 @@ def plot_varbc_pred_ts(datetime,data,labels,lloc,batch) :
 
   fig, ax = plt.subplots(figsize=(8.27,3.6))
 
-  title_string='VarBC Predictors for '+sat_map[nsat]+': ' +sen_map[nsen]+ ' Channel '+str(nchn)
+  title_string='VarBC Coefficients for '+sat_map[nsat]+': ' +sen_map[nsen]+ ' Channel '+str(nchn)
   plt.title(title_string)
 
   ax2=ax.twinx()
@@ -169,7 +169,7 @@ def plot_varbc_pred_ts(datetime,data,labels,lloc,batch) :
 
   # Ensure a major tick for each week using (interval=1) 
   ax.xaxis.set_major_locator(mdates.WeekdayLocator(interval=1))
-  ax.set_ylabel('Normalised Predictor Value',fontsize=10)
+  ax.set_ylabel('Coefficient Value',fontsize=10)
   ax2.set_ylabel('Number of observations',fontsize=10)
 
   #plot legend
@@ -179,7 +179,7 @@ def plot_varbc_pred_ts(datetime,data,labels,lloc,batch) :
   plt.tight_layout()
 
   figsuffix=str(nsat)+'_'+str(nsen)+'_'+str(nchn)
-  figname = 'varbc_pred_'+figsuffix+'.png'
+  figname = 'varbc_coeff_'+figsuffix+'.png'
   plt.savefig(figname)
   print("Output:",figname)
   if not batch :
@@ -209,7 +209,7 @@ def read_data(filename):
 #############################################################################################
 def main(argv) :
 
-  parser = argparse.ArgumentParser(description='Plot VarBC predictor time-series')
+  parser = argparse.ArgumentParser(description='Plot VarBC coefficient time-series')
   parser.add_argument('-i',dest="ipath",help='Input file name',default=None,required=True)
   parser.add_argument('-l',dest="lloc",help='Legend location using matplotlib syntax',default=None,required=False)
   parser.add_argument('-d',dest="labels",help='Optional experiment description',default=None,required=False)