Return period and bankfull attributes data for Pin2Flood May 11 exercise

Yan Y. Liu <yanliu@ornl.gov>
April 20, 2022

Using the NOAA NWM Reanalysis V21 streamflow archive, we derive three return period attributes and the 67% bankfull attribute for each stream reach. 

From NOAA NWM Reanalysis, we get 42 annual max streamflow data points, one for each of the previous 42 years (1979-02-01 to 2020-12-31). We sort these data points first. Suppose n is the number of data points (i.e., the number of years; n=42) and rank is the rank of each data point in the sorted data array, starting from 1. Below is how we derive the three return period attribute lookup tables.

Cumulative probability estimate: CPE = rank / (n + 1) . We add 1 year to denote future.

Exceedance probability estimate: EPE = (n + 1 - rank) / (n + 1) = 1 - CPE

Return period (as the number of years): RP = (n + 1) / (n + 1 - rank) = 1 / EPE

On each lookup table, the x-axis is the data points and the y-axis is CPE, EPE, or RP. Using this data array, we also compute the bankfull to be the value at the 67% percentile.

The above attributes are computed in two different ways: based on the streamflow (Q) or the water depth (H). Q-based attributes are computed for each foreach reach, i.e., feature_id used in NOAA NWM Reanalysis. The feature_ids should be consistent with the COMIDs in the NHDPlus MR dataset. Using the rating curve table of each HUC, we can also interpolate Qs into Hs and compute return period and bankfull attributes by using the interpolated water depth data points. H-based attributes, however, are computed at HydroID level, which is called model reach. Each model reach is about 1.5km long. Usually, a feature_id is split into multiple HydroIDs.

Given multiple HUC units, we produce the following output files. Suppose ${h} denotes the HUC code.

COMIDs_RP_Q_table_42years.csv : return period attribute lookup tables for all unique feature_ids in HUCs.
header: feature_id,Q_cms,RP_Q,EPE_Q,CPE_Q

COMIDs_BF_Q_table.csv : bankfull attribute for all unique feature_ids in HUCs.
header: feature_id,BF_Q_cms_67

HydroIDs_RP_H_table_42years.csv : return period attribute lookup table for all unique HydroIDs in HUCs.
header: HydroID,feature_id,huc,Q_cms,H_m,RP_H,EPE_H,CPE_H

HydroIDs_BF_H_table.csv : bankfull attribute for all unique HydroIDs in HUCs.
header: HydroID,feature_id,huc,BF_H_m_67

${h}/hydroTable_rp_bf.csv : HUC rating curve table with return period and bankfull columns. This table has both Q- and H-based attributes.
header: HydroID,feature_id,NextDownID,order_,stage,discharge_cms,HydraulicRadius (m),WetArea (m2),SLOPE,ManningN,HUC,LakeID,RP_Q_years,CPE_Q,EPE_Q,BF_Q_cms_67,RP_H_years,CPE_H,EPE_H,BF_H_m_67

${h}/hydroTable_rp_bf_hindex.csv : the same as hydroTable_rp_bf.csv but has an extra column ``hindex`` to number the stages from 1 to 84
header: HydroID,feature_id,NextDownID,order_,hindex,stage,discharge_cms,HydraulicRadius (m),WetArea (m2),SLOPE,ManningN,HUC,LakeID,RP_Q_years,CPE_Q,EPE_Q,BF_Q_cms_67,RP_H_years,CPE_H,EPE_H,BF_H_m_67

This dataset covers 6 HUC8 units in Texas Region 6:
12070102  12070201  12070202  12070203  12070205 12060203 12060204

There are 88 COMIDs missing NOAA NWM Reanalysis data
[2568394, 2568396, 2568398, 2568400, 2568402, 2568404, 2568406, 2568414, 2568536, 2568538, 2569840, 2569842, 2569844, 2569846, 2569850, 2569852, 2572278, 2572280, 2572282, 2572288, 2572292, 2572294, 2572296, 2572298, 2572302, 2572304, 2572308, 2572310, 2572312, 2572314, 2572316, 2572318, 2572320, 2572322, 2572326, 2572328, 5531586, 5531590, 5531592, 5531596, 5531600, 5531602, 5531604, 5531606, 5531608, 5531610, 5531612, 5531640, 5532450, 5532454, 5570777, 5570779, 5570781, 5570783, 5570787, 5570789, 5570791, 5570793, 5570795, 5570797, 5570799, 5570805, 5570807, 5570809, 5570811, 5570813, 5570815, 5570817, 5570819, 5587562, 5587564, 5587566, 5587568, 5587570, 5589064, 5589066, 5589068, 5589074, 5671641, 5671643, 5671645, 5671649, 5671655, 5671659, 5671661, 5671663, 5671665, 5671673]