Getting NAN for catchment attributes for certain segments

[jsadler2]

There are 40 segments in the DRB that have nan's for all segment attributes. Here are the special segments:

[1446, 1485, 1488, 1499, 1564, 1576, 1589, 1590, 1601, 1650, 1653, 
1660, 1679, 1691, 1715, 1716, 1719, 1720, 1778, 1779, 1791, 1800, 
1811, 1812, 1818, 1819, 1820, 1822, 1823, 1825, 2312, 2326, 2327, 
2328, 3558, 4187, 4188, 4190, 4195, 4196]

[jsadler2]

No obvious reason they shouldn't have attributes just from looking at the map:

[jsadler2]

I tried to use this data again and I'm figuring out what the problem is. Here is one example of what is going on:

There are three segments that are "in" this one basin (that's actually split into three). All of these three sub-basins are associated with the segment 1721 (regional id 287). None of them are associated with 1719 or 1720 which is why in the aggregated dataset the catchment attributes are nan.

[jsadler2]

I can't think of a way to easily resolve these cases, so I think what I'll do is just not use these sites in the training for the moment.

For the record, these segments have 4.7% of the total flow data and 1.6% of the total temp data from 1980-2016 (starting at water years).

seg_id_nat	n_obs_flow
1485	13150
1601	13150
1653	13150
1715	13150
1719	13150

seg_id_nat	n_obs_temp
1601	3289
1650	1
1653	37
1715	109
1719	26
1720	22
1800	65
1812	156
1820	67
1823	158
2326	1
2327	171

[jsadler2]

This came up again recently. Not all segments have HRU's! - #40