This request came from Michelle Harrold in her application of MET to GMTB. They're running the GSI tools on global gdas diagnostic files and need to compute summary statistics.
They are processing 10 output times, all valid at 00Z on different days. They'd like to use only observations that fall within 30 minutes of 00Z. This task is to enhance stat-analysis to support that time of filtering by time.
Here are some options:
(1) Instead of a single, fcst_valid_beg and fcst_valid_end, support an array of them to handle multiple time windows. Would be flexible but difficult to set up for many entries.
(2) Add fcst_valid_hour_window and obs_valid_hour_window to define a beg and end time offset in seconds around the requested valid hours. This is easier to set up than the first option but less flexible.
(3) Consider generalizing the settings for fcst_valid_beg and fcst_valid_end to support the use of wildcards. For example, use "****0101_12" would mean ignore the year setting. This could be very powerful and concise but the parsing might be really tricky.
[MET-811] created by johnhg
This request came from Michelle Harrold in her application of MET to GMTB. They're running the GSI tools on global gdas diagnostic files and need to compute summary statistics.
They are processing 10 output times, all valid at 00Z on different days. They'd like to use only observations that fall within 30 minutes of 00Z. This task is to enhance stat-analysis to support that time of filtering by time.
Here are some options:
(1) Instead of a single, fcst_valid_beg and fcst_valid_end, support an array of them to handle multiple time windows. Would be flexible but difficult to set up for many entries.
(2) Add fcst_valid_hour_window and obs_valid_hour_window to define a beg and end time offset in seconds around the requested valid hours. This is easier to set up than the first option but less flexible.
(3) Consider generalizing the settings for fcst_valid_beg and fcst_valid_end to support the use of wildcards. For example, use "****0101_12" would mean ignore the year setting. This could be very powerful and concise but the parsing might be really tricky. [MET-811] created by johnhg