Skip Navigation

Testing The Hydrologic Unit Water Quality Tool In The Tensas Basin Louisiana

EXECUTIVE SUMMARY

The Hydrologic Unit Water Quality (HUWQ) Tool was developed by the Natural Resources Conservation Service (NRCS) Information Resource Management group in Fort Collins, Colorado. The NRCS in Louisiana agreed to provide a test of the HUWQ Tool as a part of the beta testing process of the software

The software known as the HUWQ Tool was conceived as a common interface for four of the pollutant loading models that are approved by the NRCS. These models are Groundwater Loading Effects of Agricultural Management Systems (GLEAMS), Erosion/Productivity Impact Calculator (EPIC),  Simulator for Water Resources in Rural Basins-Water Quality (SWRRBWQ), and Agricultural Non-Point Source Pollution (AGNPS). GLEAMS and EPIC are field scale models, and SWRRBWQ and AGNPS are for larger sized watersheds. The current version of HUWQ Tool does not include either of the field scale models. HUWQ Tool runs in an UNIX environment, using Geographic Resources Analysis Support System (GRASS) to provide a geo-spatial representation of the watershed with individual maps or "layers" such as soils, stream network, land use, and topography. The attributes related to the Geographical Information Systems (GIS) layers are entered and stored in an INFORMIX data base.

The Tensas Basin located in northeast Louisiana is recognized as containing some of the most productive farmland in the Mississippi Delta. The accelerated conversion of forestland to cropland during the 1960's and 1970's has resulted in increased environmental problems in the basin. One of the more visible problems is diminish water quality in the basin's 460 miles of streams and bayous. The actual test site is a 2,500 acre sub-watershed of the Tensas River located approximately 15 miles west of Tallulah, Louisiana. The sub-watershed has gentle relief and contains landforms and soils typical to the basin.

The Water Management Center (WMC) was requested in 1996 to assist the Louisiana NRCS staff in testing of the HUWQ Tool. The Louisiana staff had previously installed HUWQ Tool on one of their computers, but due to hardware problems, they no longer could access the software. Starting in 1997 the GRASS data layers for the Tensas was transferred from Louisiana to the WMC. During several visits to the WMC, the team of Joe Conti, Alice Weeks, and David Moffitt completed the necessary attribute files for "present" conditions and one alternative treatment. Initial simulations for the "present" condition and alternative were made using both AGNPS and SWRRBWQ.

The test of the Hydrologic Unit Water Quality Tool for a watershed in Louisiana was a success in that we were able to install the software, integrate geo-spatial data with their attributes, and use the data to simulate pollutant loading in a portion of the Tensas Basin. The team found the Tool difficult to maintain in working order and use on a consistent basis; however, much of the HUWQ Tool use was in a series of limited duration activities followed by longer periods on inaction. Help was available from NRCS personnel in Fort Collins, Colorado; and their support is gratefully acknowledged.

We recommend:

  1. Use of HUWQ Tool by states or other units in a planning setting that have adequate on-site expertise to deal immediately with UNIX-INFORMIX issues. On-site expertise should also include those familiar with GRASS GIS and pollutant loading models.
  2. Continued assistance to states in the use of HUWQ Tool by the NWMC if proper support of the Tool can be obtained as described in 1. above.
  3. Exploring other tools integrating geo-spatial data with pollutant loading models. These tools could include the Disk Operating System (DOS) version of SWAT with BLACKLANDS GRASS, and the revision of HUWQ Tool for the WINDOWS environment to include the daily precipitation version of AGNPS. NRCS should also look beyond the commonly used Agricultural Research Service (ARS) technology for models applicable to our needs.

The HUWQ Tool was developed by the NRCS Information Resource Management group in Fort Collins, Colorado. The NRCS in Louisiana agreed to provide a test of the Hydrologic Unit Water Quality (HUWQ) Tool as a part of the beta testing process of the software. Previously a number of states had tested a prototype of the Tool, and their comments, along with those of invited prospective users of the Tool guided the final configuration of the software. The report that follows provides a brief summary of the testing of the tool. For additional information, contact Joe Conti, Resource Conservationist, with NRCS in Alexandria, Louisiana; or David Moffitt or Alice Weeks with the National Water Management Center in Fort Worth, Texas, and Little Rock, Arkansas, respectively.

DESCRIPTION OF THE HYDROLOGIC UNIT WATER QUALITY TOOL

The software known as the HUWQ Tool was conceived as a common interface for four of the pollutant loading models that are approved by the NRCS. These models are GLEAMS, EPIC, SWRRBWQ, and AGNPS. GLEAMS and EPIC are field scale models (developed to evaluate field sized watersheds with a common soil, land use, and management), and SWRRBWQ and AGNPS are for larger sized watersheds with complex soils, land use, and management; and are often referenced as watershed scale models. Although not a firm rule, the watershed scale models are most appropriate for watersheds composed of first, second, and third order streams. The current version of HUWQ Tool does not include either of the field scale models. AGNPS is a single storm event model (a single rainfall event is simulated as it would be routed through the watershed), where SWRRBWQ is a continuous simulation model where daily precipitation values drive the water budget portion of the simulation.

HUWQ Tool runs in an UNIX environment, using GRASS to provide a geo-spatial representation of the watershed with individual maps or "layers" such as soils, stream network, land use, and topography. The attributes related to the GIS layers are entered and stored in an INFORMIX data base. The attributes assigned by the user are used directly as data entry into either AGNPS or SWRRBWQ, or they are used to compute the model inputs. The concept of HUWQ Tool is that once entered, the attributes of the GIS "layers" can be used as data entry into any of the four models without the user having to do additional work. Climatic data is integrated into the Tool, and the near-by weather station is brought into the Tool keying into the GIS location. Ideally the soils data is prepared for the watershed outside HUWQ Tool and entered into the data base via Field Office Computer System (FOCS) download.

In addition to data input, the HUWQ Tool also provides an organized display of graphical and tabular output from each of the models. One of the unique features of the Tool is the "compare" function that provides either a graphical or tabular comparison of the output of various runs using the same model on the same watershed. This is particularly useful when identifying the benefits of alternative treatments.

The actual models are operated within the environment of the HUWQ Tool, but they are not modified by the Tool. In other words, the limitations of the models are not changed because of the operation of the Tool. This was particularly important in this test as will be described below. The versions of the models included in HUWQ Tool are the versions which were current in the mid 1990's when the technology was "frozen" to allow the completion of the Tool.

The Agricultural Non-Point Source Pollution (AGNPS) model was developed at USDA-Agricultural Research Service (ARS), North Central Soil Conservation Laboratory, Morris, Minnesota. AGNPS is a single-event-based computer model that was originally developed to simulate sediment and nutrient transport from agricultural watersheds in Minnesota. However, the principles are not limited to a specific geographic area.

The basic components of the model are hydrology, erosion, sediment transport, nutrient transport, and chemical oxygen demand. The model works on a cell basis. The cells are uniform square areas that collectively represent the watershed. Contaminants are routed from the headwaters of the watershed to the outlet in a stepwise fashion so that you can examine flow through any cell.

AGNPS can be used to evaluate non-point source pollution from agricultural watersheds. It can compare the effects of implementing various conservation alternatives within the watershed. The user can model cropping systems, fertilizer application rates and timing, point source loads, and the effect of terraced fields.

The model partitions soluble nitrogen and phosphorus between surface runoff and infiltration. Chemical oxygen demand and nutrient contributions from feedlots are assumed to be soluble and are transported with runoff. Once the soluble pollutants reach concentrated flow, they are conservative and accumulate in the flow. The model determines sediment-transported nitrogen and phosphorus. The model uses a modified universal soil-loss equation (USLE), adjusting for slope shape, to predict local sediment yield within the originating cell. The user can enter an estimate of gully erosion occurring in a cell, and the model adds this estimate to the total amount of sediment yield in the cell. The model also simulates sediment and runoff routing through impoundment terrace systems.

The Simulator for Water Resources in Rural Basins-Water Quality (SWRRBWQ) model was developed at USDA-ARS and Temple Agricultural Experiment Station, Temple, Texas, and the National Soil Erosion Research Laboratory, Purdue University, West Lafayette, Indiana. SWRRBWQ is a continuous simulation model that simulates hydrologic and related processes in rural basins.

The objective of the model is to predict the effect of management decisions on the hydrologic cycle, pond and reservoir storage, sedimentation, crop growth, nutrient yield, and pesticide fate at the sub-basin or basin outlet. A basin can be divided into a maximum of 10 sub-basins to account for differences in soils, land use, crops, topography, vegetation, or weather. SWRRBWQ simultaneously computes each sub-basin and routes the water, sediment, and chemicals from the sub-basin outlets to the basin outlet. It also has a lake water quality component that tracks the fate of pesticides and phosphorus from their initial application on the land to their final deposition in a lake.

SWRRBWQ can model the effect of farm-level management systems such as crop rotations, tillage, planting date, irrigation scheduling, and the rates and timing of fertilizer and pesticide applications.

The model partitions nitrate loss between surface runoff, lateral subsurface flow, percolation, and crop uptake. It also calculates soluble and sediment-attached phosphorus, computes the amount of pesticide lost with runoff and leached through the top 1 cm of soil, and calculates sediment yield using the modified universal soil loss equation (MUSLE). The model determines nutrient, pesticide, and sediment yields at the basin outlet by adding sub-basin yield.

Version 2.2.1 (Nov. 1996) of HUWQ Tool was used in the test in the Tensas Basin. This is the version of the tool on which training was provided by the Fort Collins, CO, Information Resource Management (IRM) staff in 1997. Since that time two newer versions of HUWQ Tool have been released. These later versions offer improved capabilities from version 2.2.1, but there are no assurances the problems noted below would have been avoided if the later versions were used.

DESCRIPTION OF THE WATERSHED

The Tensas Basin located in northeast Louisiana is recognized as containing some of the most productive farmland in the Mississippi Delta. Historically, over 90% of the 718,000 acre basin was forested with bottomland hardwood species. An estimated 85% of these forests have been cleared and converted to row crop agriculture. The accelerated conversion of forestland to cropland during the 1960's and 1970's has resulted in increased environmental problems in the basin. One of the more visible problems is diminish water quality in the basin's 460 miles of streams and bayous.

An interdisciplinary team of NRCS personnel selected the Tensas Basin as Louisiana's test site location based on the following rationale:

  • According to the Louisiana Department of Environmental Quality water sampled from the Tensas River only partially meets its designated uses which are primary and secondary contact and fish and wildlife propagation. The major contributors to water quality impairment are activities associated with the production of agricultural commodities. The average levels of nitrogen, phosphorus, and sediment in the Tensas River exceed those in the Mississippi River. Also, an informational advisory on fish consumption due to pesticide contamination exists for portions of the river.
  • NRCS has recently completed the Tensas River Basin Study and a case study titled Selecting Sites For Wetland Restoration In The Tensas River Basin. These studies provided insight on the existing resources, problems and opportunities in the basin.
  • The Tensas Basin contains soils and geomorphology that are similar to those found in other areas of the Mississippi Alluvial Plain. Results from the model test can be applied to other states in the Delta.

The actual test site is a 2,500 acre sub-watershed of the Tensas River located approximately 15 miles west of Tallulah, Louisiana. The sub-watershed has gentle relief and contains landforms and soils typical to the basin.

GIS DATA LAYERS

GRASS is the GIS format utilized by the HUWQ Tool. For the Tensas Basin, the GRASS layers were prepared by the Louisiana State University (LSU) Agricultural Economics Geographical Information Lab. Spatial data layers were developed for soils, stream networks, field boundaries, watershed boundaries, and elevations. The layers were digitized initially in ARCINFO and then converted to GRASS.

Two problems were encountered during the development of the GIS data layers. The first problem is related to the geography of the test site; the second problem deals with the software used to digitize the layers. The test site is located in the Mississippi River Alluvium, consequently the landscape is nearly level with only subtle changes in elevation. The United States Geological Survey (USGS) 7.5' quadrangle map for the site was found to be inadequate to accurately develop the required digital elevation model (DEM). A detailed field survey of the entire test site had to be made in order to provide enough data to develop an accurate DEM. The GRASS software package used by the model is not the GIS software of choice for most entities. The LSU lab that digitized the required layers using ARCINFO. This necessitated the lengthy development of a process to convert the data layers from ARCINFO to GRASS.

ATTRIBUTE DATA

The attribute data for the HUWQ Tool includes land use and the management of the specific land use, particularly crop land. In the Tensas Basin most of the land use was cropland, and cropping systems included cotton, corn, and soybeans. The attribute file for each field included tillage type, equipment, and tillage dates; rates and timing of fertilizers and application methodology; and pesticide application including type, amounts, and application methods. In the Tensas Basin the field by field inventory of land use was made by the Tallulah Field Office. Alternative conservation treatments for the various land uses were estimated from those commonly used in the area.

PROCESS

The Water Management Center (WMC) was requested in 1996 to assist the Louisiana NRCS staff in testing of the HUWQ Tool. The Louisiana staff had previously installed HUWQ Tool on one of their computers, but due to hardware problems, they no longer could access the software. The decision was the HUWQ Tool would be installed on the WMC SUN System, and the testing be done in Little Rock, with access available to the WMC computer from both Alexandria and Fort Worth. HUWQ Tool was installed on the South Central Regional Office SUN as well, but the Tensas data was never added. The Fort Collins IRM staff in conjunction with the Water and Climate Center staff provided training to WMC and Louisiana personnel in the use of the HUWQ Tool.

Starting in 1997 the GRASS data layers for the Tensas was transferred from Louisiana to the WMC. During several visits to the WMC, the team of Joe Conti, Alice Weeks, and David Moffitt completed the necessary attribute files for "present" conditions and one alternative treatment. Initial simulations for the "present" condition and alternative were made using both AGNPS and SWRRBWQ.

During review of the completed simulations, it became apparent the numerous pesticides used in the Tensas were not being correctly addressed by the SWRRBWQ model. After assistance from the IRM unit in Fort Collins it was discovered the SWRRBWQ would only accommodate five pesticides in any one simulation, even though the Tool would allow any number of pesticides in the attribute file. In addition, the pesticide data supplied by the Tool to the SWRRBWQ data base was not accurate; i.e. the Tool didn't take the first five pesticides in the attribute file, but seemingly selected pesticides at random. To rectify the situation, the attribute files dealing with pesticide use were modified so only five pesticides were used each simulation. The rationale was that since SWRRBWQ degraded and routed each pesticide separately (in reality there could be interactions between the chemicals, but to assume no interaction is not a gross error). The separation of the data files to accommodate the limit on number of pesticides was done from Fort Worth through remote access using the Telnet process. Doing the conservation alternative simulations using the attributes with the pesticide groups was discontinued as will be explained below.

RESULTS AND ANALYSIS

The discussion that follows focuses not on the results of the actual model simulations; rather will focus on the use of the HUWQ Tool. The numerical results of the simulation were reviewed by the team, and seemed to be consistent with observations of water yields and sediment yields common in the area. Chemical losses were consistent with losses from other studies using similar application rates.

HUWQ Tool allowed data entry, managed data files for SWRRBWQ and AGNPS input, and provided both graphical and tabular displays of output. The user's guide and other documents provided by the IRM group were helpful in defining the level of input needed (since only the watershed models are presently operational in the Tool, data for the field scale models is not needed). In these respects the test of the Tool was a success. The results of the SWRRBWQ and AGNPS simulations would be useful in a planning setting, particularly in comparing the impacts of alternative treatments to existing conditions.

It must be noted that data entry using the Tool is only slightly more time consuming than that required to use SWRRBWQ outside the Tool, and less than the time required to use AGNPS on a watershed; all assuming geo-spatial data is to be collected for other purposes. Stated differently, the use of HUWQ Tool could save time if AGNPS was the model of choice, and the user could use two simulation tools rather than one. As will be noted below, the largest variable in the time required to run HUWQ Tool will be the detail of the field descriptions.

SENSITIVITY ANALYSIS

As the test of the Tool was underway, the question was raised as to the opportunity to reduce the amount of data input at the same time preserving the magnitude of the results. The concept is one of sensitivity. What parameters really make the simulated output change dramatically in response to changes in parameter value? In most modeling efforts the crucial task is to correctly simulate hydrology, with all other model functions correlating to the hydrology component. With data such as climate, land use, topography, and soils being either a part of the Tool, or supplied to the Tool prior to analysis, land cover is one of the few factors the user can influence. With this in mind, reduced input must come from the field attributes that define the cropping system, crop operations, and management options such as chemical applications. The user can alter runoff patterns, and ultimately runoff, by imposing features such as ditches and dikes on the landscape, however, this feature doesn't impact sensitivity.

One approach for reducing input data is assuming a common cropping system for all "cropland" land use; one that would be most common in the area. This would not eliminate the data entry, but would simplify the process. Using a copy of the Sycamore data file with the Tool installed in Fort Worth, a new simulation was created with all cropland as a single year rotation of corn, and a second simulation where all cropland was a single year simulation of soybeans. The results of the simulations with limited cropping systems when compared to the actual simulated data were numerically different, but of the same magnitude. More importantly, though, the three or four areas with highest sediment yield were the same for all three simulations, and the areas with the lowest sediment yield were the same. It would appear that from the perspective of identifying priority watersheds, less field-level data could be used.

DISCUSSION

We found the Tool to be particularly temperamental and sensitive to unknown factors. During all phases of the data input and analysis we had numerous times where the Tool would not operate, often immediately after a successful operation. The Fort Collins IRM did provide support to us (the IRM no longer has the personnel to provide the level of support provided last year) to get the Tool back in operation, but often at the expense of several hours waiting. The team used HUWQ Tool over a period of several months in a series of several day sessions. Often there was weeks elapsed before the tool was used again. This intermittent use of the Tool and other uses being made of the same system may have contributed to the problems we experienced. During the re-analysis of the alternatives using the multiple pesticide files, the "compare" function of the Tool quit operating; again with no apparent reason. The test was terminated at that point. In short, we found the Tool to be extremely powerful and potentially beneficial in a planning setting, but impractical to operate without better support.

Remote access to the Tool operating at the WMC from Fort Worth was successful, but painfully slow. Data base manipulation was in the form of graphical tables that transferred very slowly through the Telnet connection. In addition, the table size at the remote access site was larger than at the host site to the point that movement around in the tables was difficult at the best. Technical support personnel in Fort Worth and at Fort Collins were never able to determine the reason or the solution for the large graphics.

A discussion of the team's experience with the HUWQ Tool must be tempered with the relatively trouble- free operation of the Tool in Virginia and New York. In discussions with the staff at both these locations, we learned there were instances when the tool would not perform as expected, but the cause of the trouble was recognized locally, and steps taken to prevent it from repeating. It appears with both of these two successful uses of the Tool, the staff was more in tune with the UNIX-INFORMIX environment than they were in the use of the pollutant loading models.

CONCLUSIONS AND RECOMMENDATIONS

The test of the Hydrologic Unit Water Quality Tool for a watershed in Louisiana was a success in that we were able to install the software, integrate geo-spatial data with their attributes, and use the data to simulate pollutant loading in a portion of the Tensas Basin. The team found the Tool difficult to maintain in working order and use on a consistent basis; however, much of the HUWQ Tool use was in a series of limited duration activities followed by longer periods on inaction. Help was available from NRCS personnel in Fort Collins, Colorado; and their support is gratefully acknowledged.

In looking at the test in total, the team concluded the HUWQ Tool provides useful data for watershed planning efforts1, and is suitable for use in a setting where a great deal of expertise is available to deal with the UNIX - INFORMIX issues that will arise. The opportunity exists for the use of the tool at a lesser intensity where the product of the use of HUWQ Tool is a general ranking of sub-watersheds.

We recommend:

  1. Use of HUWQ Tool by states or other units in a planning setting that have adequate on-site expertise to deal immediately with UNIX - INFORMIX issues. On-site expertise should also include those familiar with GRASS GIS and pollutant loading models.
  2. Continued assistance to states in the use of HUWQ Tool by the NWMC if proper support of the Tool can be obtained as described in 1. above.
  3. Exploring other tools integrating geo-spatial data with pollutant loading models. These tools could include the DOS version of SWAT with BLACKLANDS GRASS, and the revision of HUWQ Tool for the WINDOWS environment to include the daily precipitation version of AGNPS. NRCS should also look beyond the commonly used ARS technology for models applicable to our needs.

1 To use or not to use pollutant loading models in a planning setting continues to be controversial, primarily because of the large time input it requires to use the models, and the lack of "modeling expertise" at the location where the plan is prepared. Often a portion of the time can be off-set by data collection and analysis that goes on whether models are used or not.

The issue is primarily one of; "Does the model provide enough additional information to justify the time requirements?" Expressed another way; "Does the integrity of the planning process require the use of modeling techniques to support decisions that can not be supported otherwise." These issues are site specific decisions that must be made based on level of resource commitment discussed in the plan, amount of controversy in the planning area surrounding a proposed project or program, and the availability or access to the expertise required to conduct the modeling tasks.

< Back to Watershed Planning and Rehabilitation

< Back to AGNPS Other Water Quality Models