SPARROW Decision Support System

Frequently Asked Questions for the SPARROW DSS Tool

The Frequently Asked Questions listed here are specifically related to the SPARROW Decision Support System (SPARROW DSS).  FAQs related to SPARROW models and how they work can be found here.  SPARROW models are created by hydrologists to model a river system.  Those models can then be submitted to the SPARROW DSS to be viewed and explored online.  Other links:

FAQ Questions and Answers



FAQ 1: What is 'DSS'? How did you come up with that weird name?!

Answer: 'DSS' is short for Decision Support System. The SPARROW Decision Support System (DSS) provides access to national, regional, and basin-wide SPARROW models for water managers, researchers, and the general public. Decision support is provided by generating predictions of long-term, average water-quality conditions based on existing or hypothetical source contributions. The prediction results of existing conditions or source-reduction scenarios can be mapped over variety of detailed map layers, tracked downstream to receiving waters, exported and compared.

FAQ 2: Why do the predicted values in the DSS not match the prediction values in some of the published reports?

Answer: SPARROW models are expressed in the form of a mass balance. The law of conservation of mass implies that certain basic accounting rules must apply to a mass balance water-quality model, such as: (1) the sum of fluxes entering the confluence of two streams equals the flux leaving the confluence; (2) the sum of the fluxes attributable to each source must equal total flux; and (3) a doubling of all sources in the model results in an exact doubling of the predicted flux at each location. Mass balance provides a basis for flux accounting, whereby flux can be allocated to its various sources, both spatially and topically (that is, according to the location and type of source—for example, fertilizer, atmospheric deposition, etc.). For example, mass balance makes it possible to attribute nutrients discharged to the Gulf of Mexico to specific sources within the Mississippi basin, thereby providing guidance in managing the reduction of this discharge. Predictions from a SPARROW model employing mass balance are referred to as unadjusted predictions. When mass balance is employed, predictions of flux in monitored reaches (those with calibration sites) are unlikely to exactly match the measured flux in monitored reaches.

It is possible in SPARROW to condition predictions on measured flux in monitored reaches, so that predictions of flux in monitored reaches exactly match the measured flux in monitored reaches. This capability implies that predictions for all reaches at or below the monitored reach will be adjusted to be consistent with the measured flux. It is important to understand, however, that this capability does not necessarily imply that predicted flux immediately downstream of a monitored reach will be close in value to the measured flux, because of the assumed error structure of the model. Predictions that are conditioned on measured flux in monitored reaches are referred to as adjusted predictions. Adjusted predictions do not preserve mass balance, and thus they do not provide the ability to trace predicted in-stream flux in a given stream reach to the individual sources in each of the upstream reach watersheds.

The use of unadjusted versus adjusted predictions depends upon the objective. More accurate predictions may be obtained through the use of adjusted predictions, but identification of important upstream sources can only be obtained through the use of unadjusted predictions. Additionally, it is not appropriate to use adjusted predictions when the SPARROW model is run in prediction mode for purposes of simulating water-quality conditions under alternative management scenarios. The measured flux is consistent with management actions actually taken. As such, the conditioning of predictions on measured flux would not incorporate the changes in water quality associated with the alternative management scenario. Because one of the main functionalities of the DSS is simulating alternative management scenarios, all predictions available in the DSS are unadjusted.

Some published reports may include adjusted predictions. These will not match the unadjusted predictions in the DSS.

FAQ 3: Why does it seem like the New England / Mid-Atlantic Regional models run slower than the other SPARROW models?

Answer: The New England / Mid-Atlantic Regional models are based on the National Hydrography Dataset Plus (NHDPlus) stream network, which includes a stream network based on the medium resolution NHD (1:100,000 scale). The other SPARROW models are based on the Enhanced River Reach File (ERF1_2), which has a 1:500,000 scale. Because the New England / Mid-Atlantic models are based on a stream network with higher resolution, there are many more reaches to model and to display on the map, making them run slower in the DSS.

FAQ 4: Are loads coming from streams in Canada, which drain into stream in the U.S., included in the SPARROW model load predictions?

Answer: There are various assumptions involved in load predictions that may be affected by streams in Canada and it varies from model to model. See documentation on the individual SPARROW models for more information.

FAQ 5: I see reach lines drawn in unexpected places: Across lakes, along the sides of rivers, along the the ocean shore, straight lines in the desert, and some reaches don't even seem to connect to anything. Are these errors?

Answer: The stream networks include lines and catchment areas to account for non-stream features in the SPARROW model. Lines across lakes, or inter-lake reaches, transport flow from the inlet of a lake to the outlet of a lake. Lines alongside lakes and oceans are shore reaches, which represent flow entering directly into a lake or ocean from the immediate catchment area alongside the waterbody. Straight lines represent network pipes, pseudo-reaches, and other oddities that redirect flow from the natural stream network.

FAQ 6: Why are the DSS predicted values for Total Yield different then the published reports?

Answer: Total yield is derived from Total Load divided by Watershed Area. The watershed area can have some variability in how it is calculated, see the _Watershed areas_ Q & A.

FAQ 7: Why are the DSS reported Watershed areas (the total drainage area upstream of a reach) different than published values?

Answer: Several potential issues may be involved with the calculation of the watershed area. First, it is important to understand how the watershed area in SPARROW actually is calculated - See the Watershed Area section of [this help page | http://cida.usgs.gov/sparrow_qa/report/terminal_report_help.jsp] _(this needs to be updated to point to prod docs when available)_ for a complete explanation. The key point is that the watershed area used by SPARROW DSS is *not* the same as the geographical upstream area of a reach and comparisons to other published drainage areas would not be expected to match.

Second, in some cases. the modelers have split the flow in a reach using the fraction value in order to simulate flow leaving the reach (perhaps a water utility pulling water from the river). This modification affects the DSS watershed area calculation as described above. However, the modelers may have reported the original watershed areas prior to the split in a published report.

Finally, in some cases, the modelers have made specific changes to certain watershed areas by intentionally ignoring the river topology in the reach network to provide a more accurate watershed area. The DSS currently cannot incorporate these changes. If you believe this issue is affecting your data, please contact the system administrator and we may be able to resolve it.

FAQ 8: Why do the DSS predictions for some shore reaches differ from the published reports?

Answer: An error was discovered in some of the models where some shore reaches were accumulating loads from other connected reaches. Shore reaches, by definition, have no upstream reaches and exist only to account for load that directly enters a waterbody without entering a river reach. A good example of this would be a beach, where load goes directly into a lake, but never enters any type of river. This error existed at the time the MRB models were published, but was later discovered and fixed in the DSS. Modelers may have since published an errata or corrected their models.

FAQ 9: How do I match the reach ID's in the New England / Mid-Atlantic models from the DSS to the reach ID's in NHDPlus? How do I match the reach ID's of other models to the reach ID's in the appropriate stream network?


Answer: Reach ID's for the reaches in the DSS come from the stream network on which the SPARROW model is based. 

For models based on the NHDPlus stream network, including the New England / Mid-Atlantic models, reach ID's assigned to streams in the DSS match up to the NHDPlus GRIDCODE not the COMID (see http://water.usgs.gov/nawqa/modeling/nhdplusattributes.html). 

For models based on the Enhanced River Reach File (ERF1_2) stream network, including the National models, reach ID's assigned to the streams in the DSS match up to the E2RF1 identifier (see http://water.usgs.gov/GIS/metadata/usgswrd/XML/erf1_2.xml). 

For models based on the MRB Enhanced River Reach File (MRB_ERF\!_2) stream network, including the MRB regional models, reach ID's assigned to the streams in the DSS match up to the MRB_E2RF1 identifier (see http://water.usgs.gov/nawqa/modeling/rf1attributes.html).