Downsampling Time-series with cut? Pandas .resample?

[dmidlo]

Say I have the following dataset; this is water data measuring the water-level and flow (cubic feet per second) of a river used to provide early warning of flood conditions among many other uses. This is data taken from one of thousands of many other water gauges.

# gauge-id     timestamp         level flow
USGS-05286000, 2007-11-11 00:00, 2.90, 420
USGS-05286000, 2007-11-11 01:00, 2.90, 420
USGS-05286000, 2007-11-11 02:00, 2.90, 420
USGS-05286000, 2007-11-11 03:00, 2.89, 415
USGS-05286000, 2007-11-11 04:00, 2.89, 415
...
USGS-05286000, 2022-11-17 02:00, 2.94, 466
USGS-05286000, 2022-11-17 02:15, 2.94, 466
USGS-05286000, 2022-11-17 02:30, 2.95, 472
USGS-05286000, 2022-11-17 02:45, 2.94, 466
USGS-05286000, 2022-11-17 03:00, 2.95, 472
USGS-05286000, 2022-11-17 03:15, 2.94, 466
USGS-05286000, 2022-11-17 03:30, 2.93, 460

Originally, this particular gauge was configured to take samples once per hour, later it was configured to take samples once every fifteen minutes. Other gauges have different sample intervals which likewise have been changed per need as time has gone by.

Assuming the timestamp column (as a datatable frame) is a time64 column, how could I bin and downsample this data into 24hr or weekly (7day) measurements summarized by the first and last measurements of the day or week, as well as high and low during each period?

example output would be something like:

# gauge-id     timestamp   level_start,level_end,level_high,level_low,flow_start,flow_end,flow_high,flow_low
USGS-05286000, 2007-11-01, 2.90,       2.81,     2.91,      2.80,      420,       ...,     ...,      ...
USGS-05286000, 2007-11-08, 2.81,       2.95,     3.05,      2.81,      420,       ...,     ...,      ...
USGS-05286000, 2007-11-22, 2.95,       2.85,     3.01,      2.85,      420,       ...,     ...,      ...
USGS-05286000, 2007-11-29, 2.85,       2.91,     ....,      ....,      415,       ...,     ...,      ...