Skip to content

Commit

Permalink
Per #1448, switch from tech1/tech2 to dev/ops methods. Update log mes…
Browse files Browse the repository at this point in the history
…sages and add lots of details to the tc_gen documentation.
  • Loading branch information
JohnHalleyGotway committed Dec 30, 2020
1 parent f9ea178 commit 4ba89a5
Show file tree
Hide file tree
Showing 7 changed files with 218 additions and 86 deletions.
12 changes: 9 additions & 3 deletions met/data/config/TCGenConfig_default
Original file line number Diff line number Diff line change
Expand Up @@ -161,6 +161,12 @@ genesis_init_diff = 48;
//
////////////////////////////////////////////////////////////////////////////////

//
// Scoring methods to be applied
//
dev_method_flag = TRUE;
ops_method_flag = TRUE;

//
// Confidence interval alpha value
//
Expand All @@ -170,9 +176,9 @@ ci_alpha = 0.05;
// Statistical output types
//
output_flag = {
fho = NONE;
ctc = BOTH;
cts = BOTH;
fho = NONE;
ctc = BOTH;
cts = BOTH;
}

//
Expand Down
77 changes: 63 additions & 14 deletions met/docs/Users_Guide/tc-gen.rst
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,7 @@ The TC-Gen tool provides verification of tropical cyclone genesis forecasts in A
Statistical aspects
___________________

The TC-Gen tool populates a contingency table with hits, misses, and false alarms. As with other extreme events (where the event occurs much less frequently than the non-event), the correct negative category is not computed the non-events would dominate the contingency table. Therefore, only statistics that do not include correct negatives should be considered for this tool. The following CTS statistics are relevant: Base rate (BASER), Mean forecast (FMEAN), Frequency Bias (FBIAS), Probability of Detection (PODY), False Alarm Ratio (FAR), Critical Success Index (CSI), Gilbert Skill Score (GSS), Extreme Dependency Score (EDS), Symmetric Extreme Dependency Score (SEDS), Bias Adjusted Gilbert Skill Score (BAGSS).
The TC-Gen tool populates a contingency tables with hits, misses, and false alarms. As with other extreme events (where the event occurs much less frequently than the non-event), the correct negative category is not computed the non-events would dominate the contingency table. Therefore, only statistics that do not include correct negatives should be considered for this tool. The following CTS statistics are relevant: Base rate (BASER), Mean forecast (FMEAN), Frequency Bias (FBIAS), Probability of Detection (PODY), False Alarm Ratio (FAR), Critical Success Index (CSI), Gilbert Skill Score (GSS), Extreme Dependency Score (EDS), Symmetric Extreme Dependency Score (SEDS), Bias Adjusted Gilbert Skill Score (BAGSS).

Other considerations for interpreting the output of the TC-Gen tool involve the size of the contingency table output. The size of the contingency table will change depending on the number of matches. Additionally, the number of misses is based on the forecast duration and interval (specified in the configuration file). This change is due to the number of model opportunities to forecast the event, which is determined by the specified duration/interval.

Expand Down Expand Up @@ -59,19 +59,33 @@ Optional arguments for tc_gen

The TC-Gen tool implements the following logic:

* Parse the genesis data and identify forecast genesis events separately for each model present.
* Parse the forecast genesis data and identify forecast genesis events separately for each model present.

* Parse the BEST and operational track data and identify observed genesis events.
* Parse the BEST and operational track data, and identify BEST track genesis events.

* Loop over the filters defined in the configuration file and apply the following logic for each.

* For each forecast genesis event, search the BEST genesis events for a match that is close enough in time and space. If not found, search the operational genesis events for a match. If a match is found, classify the forecast genesis event as a **hit**. Otherwise, classify it as a **false alarm**.
* For each BEST track genesis event meeting the filter critera, determine the initialization and lead times for which the model had an opportunity to forecast that genesis event. Store an unmatched genesis pair for each case.

* For each forecast genesis event, search for a matching BEST track. A BEST track matches if the valid time of one of its track points matches the forecast genesis time and is within a configurable radius of the forecast genesis location. If a BEST track match is found, store the storm ID.

* In no BEST track match is found, apply the same logic to search the 0-hour operational track points. If an operational match is found, store the storm ID.

* If a matching storm ID is found, match the forecast genesis event to the BEST track genesis event for that storm ID.

* If no matching storm ID is found, store an unmatched pair for the genesis forecast.

* For each BEST track genesis event, determine the initialization and lead times for which the model had an opportunity to forecast that genesis event. If the model opportunity is not classified in the previous step, then classify as a **miss**.
* Loop through the genesis pairs and populate contingency tables using two methods, the developement (dev) and operational (ops) methods. For each pair, if the forecast genesis event is unmatched, score it as a dev and ops FALSE ALARM. If the BEST track genesis event is unmatched, score it as a dev and ops MISS. Score each matched genesis pair as follows:

* Do not count any correct negatives.
* If the forecast initialization time is at or after the BEST track genesis event, DISCARD this case and exclude it from the statistics.

* Compute the difference between the forecast and BEST track genesis events in time and space. If they are both within the configurable tolerance, score it as a dev HIT. If not, score it as a dev FALSE ALARM.

* Compute the difference between the BEST track genesis time and model initialization time. If it is within the configurable tolerance, score it as an ops HIT. If not, score it as an ops FALSE ALARM.

* Report the contingency table hits, misses, and false alarms separately for each forecast model and configuration file filter.
* Do not count any CORRECT NEGATIVES.

* Report the contingency table hits, misses, and false alarms separately for each forecast model and configuration file filter. The development (dev) scoring method is indicated in the output as *GENESIS_DEV* while the operational (ops) scoring method is indicated as *GENESIS_OPS*.

tc_gen configuration file
~~~~~~~~~~~~~~~~~~~~~~~~~
Expand All @@ -90,6 +104,14 @@ The **init_freq** variable is an integer specifying the model initialization fre

______________________

.. code-block:: none
valid_freq = 6;
The **valid_freq** variable is an integer specifying the valid time of the track points to be analyzed in hours, starting at 00Z. The default value of 6 indicates that only track points with valid times of 00Z, 06Z, 12Z, and 18Z will be checked for genesis events. Since BEST and operational tracks are typically only available at those times, a match to a forecast genesis event is only possible for those hours.

______________________

.. code-block:: none
lead_window = {
Expand Down Expand Up @@ -128,14 +150,16 @@ ______________________
vmax_thresh = NA;
mslp_thresh = NA;
}
oper_genesis = {
technique = "CARQ";
category = [ "DB", "LO", "WV" ];
vmax_thresh = NA;
mslp_thresh = NA;
}
The **best_genesis** and **oper_genesis** dictionaries define genesis criteria for the BEST and operational tracks, respectively. Like the **fcst_genesis** dictionary, the **vmax_thresh** and **mslp_thresh** thresholds define required genesis criteria. In addition, the **category** array defines the ATCF storm categories that should qualify as genesis events. The **technique** string defines the ATCF ID for the BEST and operational tracks.
The **best_genesis** dictionary define genesis criteria for the BEST tracks. Like the **fcst_genesis** dictionary, the **vmax_thresh** and **mslp_thresh** thresholds define required genesis criteria. In addition, the **category** array defines the ATCF storm categories that should qualify as genesis events. The **technique** string defines the ATCF ID for the BEST track.

______________________

.. code-block:: none
oper_technique = "CARQ";
The **oper_technique** entry is a string which defines the ATCF ID for the operational track data that should be used. For each forecast genesis event, the BEST tracks are searched for a track point valid at the time of forecast genesis and within the search radius. If no match is found, the 0-hour operational track is searched for a match.

______________________

Expand Down Expand Up @@ -235,6 +259,31 @@ The **genesis_radius** entry defines a search radius, in km, relative to the for

______________________

.. code-block:: none
genesis_init_diff = 48;
The **gensis_init_diff** entry is an integer which defines a maximum allowable time difference in hours. This option applies only to the ops matching methodology. For each matching forecast and BEST track genesis event, if the difference between the BEST track genesis time and the forecast initialization time is less than or equal to this value, then the pair is counted as a HIT. Otherwise, it is counted as a FALSE ALARM.

______________________

.. code-block:: none
dev_method_flag = TRUE;
ops_method_flag = TRUE;
The **dev_method_flag** and **ops_method_flag** entries are booleans which indicate whether the development and operational scoring methods should be applied and written to the output. At least one of these flags must be set to true.

______________________

.. code-block:: none
basin_file = "MET_BASE/tc_data/basin_global_tenth_degree.nc";
The **basin_file** entry defines the path to the NetCDF basin data file that is included with MET. When a BEST track storm moves from one basin to another, the BEST track dataset can include two tracks for the same storm, one for each basin. However, both tracks have the same genesis point. When this occurs, this basin data file is read and used to determine the basin in which genesis actually occurred. The corresponding BEST track is retained and the other is discarded.

______________________

.. code-block:: none
ci_alpha = 0.05;
Expand Down
2 changes: 2 additions & 0 deletions met/src/basic/vx_config/config_constants.h
Original file line number Diff line number Diff line change
Expand Up @@ -1069,6 +1069,8 @@ static const char conf_key_mslp_thresh[] = "mslp_thresh";
static const char conf_key_genesis_window[] = "genesis_window";
static const char conf_key_genesis_radius[] = "genesis_radius";
static const char conf_key_genesis_init_diff[] = "genesis_init_diff";
static const char conf_key_dev_method_flag[] = "dev_method_flag";
static const char conf_key_ops_method_flag[] = "ops_method_flag";

//
// TC-RMW specific parameter key names
Expand Down
Loading

0 comments on commit 4ba89a5

Please sign in to comment.