Replace gfs_cyc with interval

To facilitate longer and more flexible GFS cadences, the `gfs_cyc`
variable is replaced with a specified interval. Up front, this is
reflected in a change in the arguments for setup_exp to:

```
--interval <n_hours>
```

Where `n_hours` is the interval (in hours) between gfs forecasts.
`n_hours` must be a multiple of 6. If 0, no gfs will be run (only
gdas; only valid for cycled mode). The default value is 6 (every
cycle).

In cycled mode, there is an additional argument to control which
cycle will be the first gfs cycle:

```
---sdate_gfs <YYYYMMDDHH>
```

The default if not provided is `--idate` + 6h (first full cycle).

As part of this change, some of the validation of the dates has
been added. `--edate` has also been made optional and defaults to
`--idate` if not provided.

During `config.base` template-filling, `INTERVAL_GFS` (renamed from
`STEP_GFS`) is defined as `--interval` and `SDATE_GFS as
`--sdate_gfs`.

Some changes were necessary to the gfs verification (metp) job, as
`gfs_cyc` was being used downstream by verif-global. That has been
removed, and instead workflow will be responsible for only running
metp on the correct cycles. This also removes "do nothing" metp
tasks that exit immediately, because only the last GFS cycle in a
day would actually process verification.

Now, metp has its own cycledef and always runs at 18z, regardless
of whether gfs is running at 18z or not. This is simplier than
trying to determine the last gfs cycle of a day when it could
change from day to day. To facilitate this change, support for the
undocumented rocoto dependency tag `taskvalid` is added, as the
metp task needs to know whether the cycle has a gfsarch task or not.
metp will trigger on gfsarch completing (as before), or gdasarch
completing if there is no gfsarch.

metp tasks are no longer generated for forecast-only, as the pgbanl
files (copied of the 1p00 pgbanl files) are not generated for f-o
anyway. If metp is needed for f-o, additional work will be needed.

Additionally, a couple EE2 issues with the metp job are resolved
(even though it is not run in ops):
- verif-global update replaced `$CDUMP` with `$RUN`
- `$DATAROOT` is no longer redefined in the metp job

Depends on NOAA-EMC/EMC_verif-global#137
Resolves NOAA-EMC#260
Refs NOAA-EMC#1299

gempak/ush/gfs_meta_comp.sh

@@ -24,7 +24,7 @@ device="nc | ${metaname}"
 
 export COMIN="gfs.multi"
 mkdir "${COMIN}"
-for cycle in $(seq -f "%02g" -s ' ' 0  "${STEP_GFS}" "${cyc}"); do
+for cycle in $(seq -f "%02g" -s ' ' 0  "${INTERVAL_GFS}" "${cyc}"); do
     YMD=${PDY} HH=${cycle} GRID="1p00" declare_from_tmpl gempak_dir:COM_ATMOS_GEMPAK_TMPL
     for file_in in "${gempak_dir}/gfs_1p00_${PDY}${cycle}f"*; do
         file_out="${COMIN}/$(basename "${file_in}")"

gempak/ush/gfs_meta_mar_comp.sh

@@ -15,7 +15,7 @@ cp "${HOMEgfs}/gempak/fix/datatype.tbl" datatype.tbl
 
 export COMIN="gfs.multi"
 mkdir -p "${COMIN}"
-for cycle in $(seq -f "%02g" -s ' ' 0 "${STEP_GFS}" "${cyc}"); do
+for cycle in $(seq -f "%02g" -s ' ' 0 "${INTERVAL_GFS}" "${cyc}"); do
     YMD=${PDY} HH=${cycle} GRID="1p00" declare_from_tmpl gempak_dir:COM_ATMOS_GEMPAK_TMPL
     for file_in in "${gempak_dir}/gfs_1p00_${PDY}${cycle}f"*; do
         file_out="${COMIN}/$(basename "${file_in}")"

jobs/JGFS_ATMOS_VERIFICATION

@@ -16,12 +16,6 @@ source "${HOMEgfs}/ush/jjob_header.sh" -e "metp" -c "base metp"
 ## METPCASE : METplus verification use case (g2g1 | g2o1 | pcp1)
 ###############################################################
 
-# TODO: This should not be permitted as DATAROOT is set at the job-card level.
-# TODO: DATAROOT is being used as DATA in metp jobs.  This should be rectified in metp.
-# TODO: The temporary directory is DATA and is created at the top of the J-Job.
-# TODO: remove this line
-export DATAROOT=${DATA}
-
 VDATE=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} - ${VRFYBACK_HRS} hours")
 export VDATE=${VDATE:0:8}
 

parm/config/gefs/config.base

@@ -224,7 +224,8 @@ export FHOUT_OCN=3
 export FHOUT_ICE=3
 
 # GFS cycle info
-export gfs_cyc=@gfs_cyc@ # 0: no GFS cycle, 1: 00Z only, 2: 00Z and 12Z only, 4: all 4 cycles.
+export INTERVAL_GFS=@INTERVAL_GFS@ # Frequency of GFS forecast
+export SDATE_GFS=@SDATE_GFS@
 
 # GFS output and frequency
 export FHMIN_GFS=0
@@ -243,11 +244,6 @@ export FHOUT_WAV=3
 export FHMAX_HF_WAV=120
 export FHOUT_HF_WAV=1
 export FHMAX_WAV=${FHMAX_GFS}
-if (( gfs_cyc != 0 )); then
-    export STEP_GFS=$(( 24 / gfs_cyc ))
-else
-    export STEP_GFS="0"
-fi
 export ILPOST=1           # gempak output frequency up to F120
 
 export FHMIN_ENKF=${FHMIN_GFS}

parm/config/gfs/config.base

@@ -280,7 +280,8 @@ export FHOUT_ICE=3
 export EUPD_CYC="@EUPD_CYC@"
 
 # GFS cycle info
-export gfs_cyc=@gfs_cyc@ # 0: no GFS cycle, 1: 00Z only, 2: 00Z and 12Z only, 4: all 4 cycles.
+export INTERVAL_GFS=@INTERVAL_GFS@ # Frequency of GFS forecast
+export SDATE_GFS=@SDATE_GFS@
 
 # GFS output and frequency
 export FHMIN_GFS=0
@@ -299,11 +300,7 @@ export FHMAX_HF_WAV=120
 export FHOUT_HF_WAV=1
 export FHMAX_WAV=${FHMAX:-9}
 export FHMAX_WAV_GFS=${FHMAX_GFS}
-if (( gfs_cyc != 0 )); then
-    export STEP_GFS=$(( 24 / gfs_cyc ))
-else
-    export STEP_GFS="0"
-fi
+
 # TODO: Change gempak to use standard out variables (#2348)
 export ILPOST=${FHOUT_HF_GFS}           # gempak output frequency up to F120
 if (( FHMAX_HF_GFS < 120 )); then

parm/config/gfs/config.wave

@@ -117,7 +117,7 @@ if [[ "${RUN}" == "gdas" ]]; then
   export WAVNCYC=4
   export WAVHCYC=${assim_freq:-6}
   export FHMAX_WAV_CUR=48 # RTOFS forecasts only out to 8 days
-elif [[ ${gfs_cyc} -ne 0 ]]; then
+elif (( INTERVAL_GFS > 0 )); then
   export WAVHCYC=${assim_freq:-6}
   export FHMAX_WAV_CUR=192 # RTOFS forecasts only out to 8 days
 else

scripts/exgfs_aero_init_aerosol.py

@@ -11,13 +11,13 @@
 ---------
 This script requires the following environment variables be set beforehand:
 
-CDATE: 		Initial time in YYYYMMDDHH format
-STEP_GFS: 	Forecast cadence (frequency) in hours
-FHMAX_GFS: 	Forecast length in hours
-RUN: 		Forecast phase (gfs or gdas). Currently always expected to be gfs.
-ROTDIR: 	Rotating (COM) directory
-USHgfs: 	Path to global-workflow `ush` directory
-PARMgfs: 	Path to global-workflow `parm` directory
+CDATE:        Initial time in YYYYMMDDHH format
+INTERVAL_GFS: Forecast cadence (frequency) in hours
+FHMAX_GFS:    Forecast length in hours
+RUN:          Forecast phase (gfs or gdas). Currently always expected to be gfs.
+ROTDIR:       Rotating (COM) directory
+USHgfs:       Path to global-workflow `ush` directory
+PARMgfs:      Path to global-workflow `parm` directory
 
 Additionally, the following data files are used:
 
@@ -66,7 +66,7 @@
 def main() -> None:
     # Read in environment variables and make sure they exist
     cdate = get_env_var("CDATE")
-    incr = int(get_env_var('STEP_GFS'))
+    incr = int(get_env_var('INTERVAL_GFS'))
     fcst_length = int(get_env_var('FHMAX_GFS'))
     run = get_env_var("RUN")
     rot_dir = get_env_var("ROTDIR")

workflow/applications/applications.py

@@ -68,6 +68,7 @@ def __init__(self, conf: Configuration) -> None:
 
         self.nens = base.get('NMEM_ENS', 0)
         self.fcst_segments = base.get('FCST_SEGMENTS', None)
+        self.interval_gfs = to_timedelta(f"{base.get('INTERVAL_GFS')}H")
 
         if not AppConfig.is_monotonic(self.fcst_segments):
             raise ValueError(f'Forecast segments do not increase monotonically: {",".join(self.fcst_segments)}')
@@ -109,9 +110,6 @@ def _init_finalize(self, conf: Configuration):
         # Save base in the internal state since it is often needed
         base = self.configs['_no_run']['base']
 
-        # Get more configuration options into the class attributes
-        self.gfs_cyc = base.get('gfs_cyc')
-
         # Get task names for the application
         self.task_names = self.get_task_names()
 
@@ -199,19 +197,6 @@ def get_task_names(self, run="_no_run") -> Dict[str, List[str]]:
         '''
         pass
 
-    @staticmethod
-    def get_gfs_interval(gfs_cyc: int) -> timedelta:
-        """
-        return interval in hours based on gfs_cyc
-        """
-
-        gfs_internal_map = {'1': '24H', '2': '12H', '4': '6H'}
-
-        try:
-            return to_timedelta(gfs_internal_map[str(gfs_cyc)])
-        except KeyError:
-            raise KeyError(f'Invalid gfs_cyc = {gfs_cyc}')
-
     @staticmethod
     def is_monotonic(test_list: List, check_decreasing: bool = False) -> bool:
         """

workflow/applications/gefs.py

@@ -42,7 +42,6 @@ def _get_app_configs(self):
     def _update_base(base_in):
 
         base_out = base_in.copy()
-        base_out['INTERVAL_GFS'] = AppConfig.get_gfs_interval(base_in['gfs_cyc'])
         base_out['RUN'] = 'gefs'
 
         return base_out

workflow/applications/gfs_cycled.py

@@ -128,7 +128,7 @@ def _get_app_configs(self):
     @staticmethod
     def _update_base(base_in):
 
-        return GFSCycledAppConfig.get_gfs_cyc_dates(base_in)
+        return base_in
 
     def get_task_names(self):
         """
@@ -297,7 +297,7 @@ def get_task_names(self):
             tasks['enkfgdas'] = enkfgdas_tasks
 
         # Add RUN=gfs tasks if running early cycle
-        if self.gfs_cyc > 0:
+        if self.interval_gfs > to_timedelta("0H"):
             tasks['gfs'] = gfs_tasks
 
             if self.do_hybvar and 'gfs' in self.eupd_runs:
@@ -307,49 +307,3 @@ def get_task_names(self):
                 tasks['enkfgfs'] = enkfgfs_tasks
 
         return tasks
-
-    @staticmethod
-    def get_gfs_cyc_dates(base: Dict[str, Any]) -> Dict[str, Any]:
-        """
-        Generate GFS dates from experiment dates and gfs_cyc choice
-        """
-
-        base_out = base.copy()
-
-        sdate = base['SDATE']
-        edate = base['EDATE']
-        base_out['INTERVAL'] = to_timedelta(f"{base['assim_freq']}H")
-
-        # Set GFS cycling dates
-        gfs_cyc = base['gfs_cyc']
-        if gfs_cyc != 0:
-            interval_gfs = AppConfig.get_gfs_interval(gfs_cyc)
-            hrinc = 0
-            hrdet = 0
-            if gfs_cyc == 1:
-                hrinc = 24 - sdate.hour
-                hrdet = edate.hour
-            elif gfs_cyc == 2:
-                if sdate.hour in [0, 12]:
-                    hrinc = 12
-                elif sdate.hour in [6, 18]:
-                    hrinc = 6
-                if edate.hour in [6, 18]:
-                    hrdet = 6
-            elif gfs_cyc == 4:
-                hrinc = 6
-            sdate_gfs = sdate + timedelta(hours=hrinc)
-            edate_gfs = edate - timedelta(hours=hrdet)
-            if sdate_gfs > edate:
-                print('W A R N I N G!')
-                print('Starting date for GFS cycles is after Ending date of experiment')
-                print(f'SDATE = {sdate.strftime("%Y%m%d%H")},     EDATE = {edate.strftime("%Y%m%d%H")}')
-                print(f'SDATE_GFS = {sdate_gfs.strftime("%Y%m%d%H")}, EDATE_GFS = {edate_gfs.strftime("%Y%m%d%H")}')
-                gfs_cyc = 0
-
-            base_out['gfs_cyc'] = gfs_cyc
-            base_out['SDATE_GFS'] = sdate_gfs
-            base_out['EDATE_GFS'] = edate_gfs
-            base_out['INTERVAL_GFS'] = interval_gfs
-
-        return base_out

workflow/applications/gfs_forecast_only.py

@@ -42,9 +42,6 @@ def _get_app_configs(self):
             if self.do_genesis_fsu:
                 configs += ['genesis_fsu']
 
-            if self.do_metp:
-                configs += ['metp']
-
             if self.do_bufrsnd:
                 configs += ['postsnd']
 
@@ -78,7 +75,7 @@ def _get_app_configs(self):
     def _update_base(base_in):
 
         base_out = base_in.copy()
-        base_out['INTERVAL_GFS'] = AppConfig.get_gfs_interval(base_in['gfs_cyc'])
+        base_out['RUN'] = 'gfs'
 
         return base_out
 
@@ -122,9 +119,6 @@ def get_task_names(self):
             if self.do_genesis_fsu:
                 tasks += ['genesis_fsu']
 
-            if self.do_metp:
-                tasks += ['metp']
-
             if self.do_bufrsnd:
                 tasks += ['postsnd']
 

workflow/rocoto/gefs_xml.py

@@ -14,19 +14,19 @@ def __init__(self, app_config: AppConfig, rocoto_config: Dict) -> None:
         super().__init__(app_config, rocoto_config)
 
     def get_cycledefs(self):
-        sdate = self._base['SDATE']
+        sdate = self._base['SDATE_GFS']
         edate = self._base['EDATE']
-        interval = self._base.get('INTERVAL_GFS', to_timedelta('24H'))
+        interval = self.interval_gfs
         sdate_str = sdate.strftime("%Y%m%d%H%M")
         edate_str = edate.strftime("%Y%m%d%H%M")
         interval_str = timedelta_to_HMS(interval)
         strings = []
         strings.append(f'\t<cycledef group="gefs">{sdate_str} {edate_str} {interval_str}</cycledef>')
 
-        sdate = sdate + interval
-        if sdate <= edate:
-            sdate_str = sdate.strftime("%Y%m%d%H%M")
-            strings.append(f'\t<cycledef group="gefs_seq">{sdate_str} {edate_str} {interval_str}</cycledef>')
+        date2 = sdate + interval
+        if date2 <= edate:
+            date2_str = date2.strftime("%Y%m%d%H%M")
+            strings.append(f'\t<cycledef group="gefs_seq">{date2_str} {edate_str} {interval_str}</cycledef>')
 
         strings.append('')
         strings.append('')

workflow/rocoto/gfs_cycled_xml.py

@@ -24,19 +24,30 @@ def get_cycledefs(self):
         sdate_str = sdate.strftime("%Y%m%d%H%M")
         strings.append(f'\t<cycledef group="gdas">{sdate_str} {edate_str} {interval_str}</cycledef>')
 
-        if self._app_config.gfs_cyc != 0:
+        interval_gfs = self._app_config.interval_gfs
+
+        if interval_gfs > to_timedelta("0H"):
             sdate_gfs = self._base['SDATE_GFS']
-            edate_gfs = self._base['EDATE_GFS']
-            interval_gfs = self._base['INTERVAL_GFS']
+            edate_gfs = edate
             sdate_gfs_str = sdate_gfs.strftime("%Y%m%d%H%M")
             edate_gfs_str = edate_gfs.strftime("%Y%m%d%H%M")
             interval_gfs_str = timedelta_to_HMS(interval_gfs)
             strings.append(f'\t<cycledef group="gfs">{sdate_gfs_str} {edate_gfs_str} {interval_gfs_str}</cycledef>')
 
-            sdate_gfs = sdate_gfs + interval_gfs
-            sdate_gfs_str = sdate_gfs.strftime("%Y%m%d%H%M")
-            if sdate_gfs <= edate_gfs:
-                strings.append(f'\t<cycledef group="gfs_seq">{sdate_gfs_str} {edate_gfs_str} {interval_gfs_str}</cycledef>')
+            date2_gfs = sdate_gfs + interval_gfs
+            date2_gfs_str = date2_gfs.strftime("%Y%m%d%H%M")
+            if date2_gfs <= edate_gfs:
+                strings.append(f'\t<cycledef group="gfs_seq">{date2_gfs_str} {edate_gfs_str} {interval_gfs_str}</cycledef>')
+
+            if self._base['DO_METP']:
+                # Run verification at 18z, no matter what
+                sdate_metp = sdate_gfs.replace(hour=18)
+                edate_metp = edate_gfs.replace(hour=18)
+                interval_metp = to_timedelta('24H')
+                sdate_metp_str = sdate_metp.strftime("%Y%m%d%H%M")
+                edate_metp_str = edate_metp.strftime("%Y%m%d%H%M")
+                interval_metp_str = timedelta_to_HMS(interval_metp)
+                strings.append(f'\t<cycledef group="metp">{sdate_metp_str} {edate_metp_str} {interval_metp_str}</cycledef>')
 
         strings.append('')
         strings.append('')

workflow/rocoto/gfs_forecast_only_xml.py

@@ -12,15 +12,16 @@ def __init__(self, app_config: AppConfig, rocoto_config: Dict) -> None:
         super().__init__(app_config, rocoto_config)
 
     def get_cycledefs(self):
-        sdate = self._base['SDATE']
+        sdate = self._base['SDATE_GFS']
         edate = self._base['EDATE']
-        interval = self._base.get('INTERVAL_GFS', to_timedelta('24H'))
+        interval = self._app_config.interval_gfs
         strings = []
         strings.append(f'\t<cycledef group="gfs">{sdate.strftime("%Y%m%d%H%M")} {edate.strftime("%Y%m%d%H%M")} {timedelta_to_HMS(interval)}</cycledef>')
 
-        sdate = sdate + interval
-        if sdate <= edate:
-            strings.append(f'\t<cycledef group="gfs_seq">{sdate.strftime("%Y%m%d%H%M")} {edate.strftime("%Y%m%d%H%M")} {timedelta_to_HMS(interval)}</cycledef>')
+        date2 = sdate + interval
+        date2_str = date2.strftime("%Y%m%d%H%M")
+        if date2 <= edate:
+            strings.append(f'\t<cycledef group="gfs_seq">{date2.strftime("%Y%m%d%H%M")} {edate.strftime("%Y%m%d%H%M")} {timedelta_to_HMS(interval)}</cycledef>')
 
         strings.append('')
         strings.append('')