Adding key metrics: Isla's NMSE metric

examples/key_metrics/config.yml

@@ -54,9 +54,9 @@ timeseries:
   case_name: 'b.e23_alpha17f.BLT1850.ne30_t232.092'
 
   atm:
-    vars: ['ACTNI', 'ACTNL', 'ACTREI', 'ACTREL', 'AODDUST']
-    derive_vars: [] # {'PRECT':['PRECL', 'PRECC'], 'RESTOM':['FLNT', 'FSNT']}
-    hist_str: 'h0'
+    vars: ['PSL']
+    derive_vars: []
+    hist_str: 'h0a'
     start_years: [1]
     end_years: [100]
     level: 'lev'
@@ -107,13 +107,14 @@ compute_notebooks:
         parameter_groups:
           none: {}
 
-#    atm:
-#      adf_quick_run:
-#        parameter_groups:
-#          none:
-#            adf_path: ../../../externals/ADF
-#            config_path: .
-#            config_fil_str: "config_f.cam6_3_119.FLTHIST_ne30.r328_gamma0.33_soae.001.yaml"
+    atm:
+      nmse_PSL:
+        parameter_groups:
+          none:
+            case: b.e23_alpha17f.BLT1850.ne30_t232.092
+            component: atm
+            start_date: '0001-01-01'
+            end_date: '0101-01-01'
 
     glc:
       LIWG_SMB_diagnostic:
@@ -181,13 +182,12 @@ book_toc:
 
     # Parts group notebooks into different sections in the Jupyter book
     # table of contents, so you can organize different parts of your project.
+    # Each chapter is the name of one of the notebooks that you executed
+    # in compute_notebooks above, also without .ipynb
 
-    # - caption: Atmosphere
-
-    #   # Each chapter is the name of one of the notebooks that you executed
-    #   # in compute_notebooks above, also without .ipynb
-    #   chapters:
-    #     - file: atm/adf_quick_run
+    - caption: Atmosphere
+      chapters:
+        - file: atm/nmse_PSL
 
     # - caption: Ocean
     #   chapters:

examples/nblibrary/atm/averaging_utils.py

@@ -0,0 +1,55 @@
+from __future__ import annotations
+
+import xarray as xr
+
+dpseas = {"DJF": 90, "MAM": 92, "JJA": 92, "SON": 91}
+
+
+def seasonal_climatology_weighted(dat):
+    """Calculate seasonal and annual average climatologies"""
+    days_in_month = dat.time.dt.days_in_month
+    mons = dat.time.dt.month
+    wgts = mons.copy(deep=True)
+    wgts = xr.where(
+        (mons == 1) | (mons == 2) | (mons == 12),
+        days_in_month / dpseas["DJF"],
+        wgts,
+    )
+    wgts = xr.where(
+        (mons == 3) | (mons == 4) | (mons == 5),
+        days_in_month / dpseas["MAM"],
+        wgts,
+    )
+    wgts = xr.where(
+        (mons == 6) | (mons == 7) | (mons == 8),
+        days_in_month / dpseas["JJA"],
+        wgts,
+    )
+    wgts = xr.where(
+        (mons == 9) | (mons == 10) | (mons == 11),
+        days_in_month / dpseas["SON"],
+        wgts,
+    )
+    datw = dat * wgts
+
+    wgts_am = days_in_month / 365.0
+    datw_am = dat * wgts_am
+
+    ds_season = (
+        datw.rolling(min_periods=3, center=True, time=3).sum().dropna("time", how="all")
+    )
+    dat_djf = ds_season.where(ds_season.time.dt.month == 1, drop=True).mean("time")
+    dat_mam = ds_season.where(ds_season.time.dt.month == 4, drop=True).mean("time")
+    dat_jja = ds_season.where(ds_season.time.dt.month == 7, drop=True).mean("time")
+    dat_son = ds_season.where(ds_season.time.dt.month == 10, drop=True).mean("time")
+    dat_am = datw_am.groupby("time.year").sum("time")
+    dat_am = dat_am.mean("year")
+
+    dat_djf = dat_djf.rename("DJF")
+    dat_mam = dat_mam.rename("MAM")
+    dat_jja = dat_jja.rename("JJA")
+    dat_son = dat_son.rename("SON")
+    dat_am = dat_am.rename("AM")
+
+    alldat = xr.merge([dat_djf, dat_mam, dat_jja, dat_son, dat_am])
+    return alldat