Frequently Asked Questions#

What is nwm_region_mgr?

nwm_region_mgr is a command line Python utility for identifying optimized NextGen formulations and parameter values in ungauged catchments.

What is a formulation?

A NextGen formulation is any single model or combination of models and/or modules running in a basin to simulate that location’s hydrology. A formulation may include one or more hydrologic models and modules, plus a streamflow routing module, a coastal model, and other supporting routines.

For more information on the available models/modules, please see NOAA-OWP/NextGen-Info

What are parameters in this context?

Every model or module utilizes parameters to reflect physical attributes of a location when making predictions of hydrologic behavior. Parameters might measure physical quantities such as drainage area, they could be dimensionless indices such as topographic wetness index, or they may be non-interpretable quantities such as weights and activation functions in a neural network. Each model or module will require different sets of parameters to make predictions.

During calibration in gauged catchments, parameter values are tuned to create the “best” recreation of observed data for a given model.

In what areas of the world can I run this?

Regionalization is currently supported for CONUS, Alaska, Hawaii, and Puerto Rico and the Virgin Islands. The spatial domain is set in config_general.yaml under the general:domain field, where values can be conus, ak, hi, prvi, respectively.

What watershed delineations are used?

Regionalization relies on the NextGen Hydrofabric, which is set in config_general.yaml under the general:ngen_hydrofabric_file field.

What spatial unit are parameters assigned to?

Regionalization assigns parameters to hydrofabric divides (i.e., catchments), which serve as the fundamental spatial units.

How does nwm_region_mgr find optimal formulations?

Optimal formulations are first determined for calibrated catchments based on computational cost and user-specified performance metrics (e.g. Nash–Sutcliffe efficiency (nse), Kling-Gupta efficiency (KGE), bias, or correlation (cor)). For uncalibrated catchments, optimal formulations are then assigned by selecting the best-performing formulation from calibrated catchments within the same region based on the user-defined HUC level (e.g., HUC8).

How does nwm_region_mgr find optimal parameter values?

Parameter values are assigned to each hydrofabric divide by identifying calibrated catchments that are close to the divide in both mapped location and physiographic characteristics. Users can choose from multiple clustering or distance-based algorithms and multiple attribute datasets (e.g., ngen: NextGen divide attributes, hlr: Hydrologic Landscape Region attributes, and streamcat: StreamCat attributes).

Which attribute datasets are supported for each domain?

The following attribute datasets are supported for each domain:

  • CONUS (conus): ngen, hlr, streamcat

  • Alaska (ak): ngen, hlr

  • Hawaii (hi): ngen, hlr

  • Puerto Rico and Virgin Islands (prvi): ngen

Does the donor/receiver pairing distinguish between snowy and non-snowy regions?

Currrently, the snowiness of catchments is determined using the mean annual snowfall fraction derived based on the HydroATLAS dataset, based on “snow_cover.threshold” defined in the configuration file. If the “snow_cover.consider_snowness” option is set to True, parameter regionalization will be performed separately for snowy and non-snowy basins.

Note HydroATLAS-based mean annual snowfall fraction data is not available for PRVI.

How do I start using this tool?

Check out the User Guide to get started.

This Page