Table of Contents
           Reservoir 1 & 2.

•  Flow Data
•  Distributed Inflows
•  Temperature Data
•  Meteorological Data


    Lower Bull Run River
 

•  Inflows
•  Inflow Temperatures

One aspect of modeling the river-reservoir system is to incorporate the variability in the system due to inflows from various sources. Table 7 shows how much of the drainage basin to Reservoir #1 is accounted for by known gaged and ungaged streams based on their drainage areas. Overall, 92% of the Reservoir #1 basin can be characterized by a group of tributaries entering the reservoir which are either gaged or could be related to gaged streams. Four of the major tributaries to the reservoirs have gaged flows that provide flow measurements approximately every half hour. Additional tributaries had intermittent flow measurements made, approximately four times a year from 1980 to 1991. The data from the intermittent gaged stations was correlated with the continuously recorded USGS gage stations to generate a more complete record.

Table 8 indicates similar information for the Reservoir #2 drainage basin. When considering the flow contribution from Reservoir #1, 92% of the basin area can be accounted for by gaged streams or streams which can be correlated to gaged streams. The remaining streams which make up the rest of the drainage basin areas, 8% for each, were analyzed in more detail using a Geographic Information System (GIS) to identify their inflow contributions.
 
 

Figure 39 shows the variability in the stream flows from 4 USGS gaged streams which contribute to the reservoirs in both cfs and m³/s. Additionally, Figure 40 illustrates the inflows on a log scale. The largest contributor is the Main Stem of the Bull Run River and the second is the South Fork Bull Run River which contribute to Reservoir #1 and #2 respectively. The plots also illustrate that although the magnitudes of the flows vary their peaks and ebbs occur at the same time.

Only four of the nine stream which make up 92% of each drainage basin are gaged continuously the other five streams needed to be related to the gaged streams if possible. From 1979 to 1991 periodic flow measurements were made at Bear Creek, Camp Creek, Cougar Creek, Deer Creek, and Fivemile Creek. These measurements provided some information on stream flows but were not enough information to generate a continuous flow data file needed for the model.

Correlations were developed between the ungaged streams and the USGS gaged streams by taking the daily average flow from the USGS streams and correlating them with the individual measurements made at the other streams. Although relating daily averages with once-a-day flow measurements introduces the potential for error, relatively good correlations were achieved as illustrated in Table 11. If more frequent data were collected on these ungaged streams, the correlations may be improved.

Although the multiple regression correlations resulted in high correlation, the equations resulted in some negative flows calculated for the ungaged streams. To correct this the flow measurements recorded at the five ungaged streams were examined further. The data were collected by the U.S. Geological Survey and when measurements were made in the summer months it was believed the measured flows represented the base flow in the stream and was cited in the field notes. This information was then used to identify and calculate the average baseflow in each of the summer months from the years when the data were gathered, Table 9. The baseflows were then used as the lowest flow for each stream. Any flows which were calculated from the correlation equations which were lower than the baseflow were replaced by the baseflow values in Table 9 for their respective months. For the months of November through May the October baseflow were used since it most likely represented the baseflow in the winter months. For the month of June, the July baseflow was used since it was the closest month to have a baseflow measurement.

( 4 )

Based on the correlations and the baseflow adjustments, daily flow values were generated for the time period of 1996 to 1998 to use in the model, Figure 41. As expected the streams follow the same general shape as the USGS gaged stream flow plots, but with smaller magnitudes.
 
 

Distributed Inflows

As Table 7 and Table 8 above indicate, approximately 9.3% and 7.4% of the basin area is not considered in the inflows to Reservoirs #1 and #2 respectively. In order to approximate the flows which could be attributed to these ungaged areas correlations were first developed between existing subbasins where the flow and basin area were known. The correlations were conducted to relate two flows between basins using the basin areas as follows:

The results of conducting the correlations can be found in Table 12. The flow measurements used in these correlations were recorded on a half hourly basis. Graphs illustrating these correlations can be found in Appendix M.

The next step in the analysis was to identify the ungaged basins. This was done using a geographic information system of the Bull Run watershed. The ungaged subbasins were identified and their areas calculated as shown in Figure 43 and Figure 44 and Table 13.

Flows for the ungaged subbasins were then estimated by using the correlations developed above with adjacent or nearby subbasins. Table 10 shows the resulting equations for estimating flow in these ungaged basins. The flows were then summed over each reservoir to generate a distributed flow file for each reservoir in the Bull Run River - Reservoir Model. The inflows were incorporated into the model as distributed flows because the subbasins surround each reservoir and contribute to the reservoirs over wide areas, not in point source locations.

Temperature Data

In addition to flow measurements, the USGS gage stations also measure water temperature. Figure 45 illustrates the water temperature records for the streams from 1996 to 1999. The warmest contributors to the reservoirs seem to be the Main Stem of the Bull Run River and the South Fork Bull Run River with temperatures rising to 16^o C and higher in the summer. Slightly cooler temperatures are noted on Fir Creek and the North Fork Bull Run River during the summer. Overall the temperature variability is similar across all four streams.

Continuous flow records were generated for five streams that were not gaged but had several flow measurements made over several years. In order to generate appropriate temperature inflow files water temperatures were used from the closest gaged streams for the same time period. Table 15 shows the stream temperatures used for each ungaged stream.

In reviewing the water temperature data recorded at the USGS gage station on the Main Stem river flowing into Reservoir #1, two data gaps were identified: from 11/06/1997 to 12/23/1997 and from 01/11/1998 to 01/13/1998. In order to fill these gaps two correlations were developed between the water temperature measurements at the USGS site on the Main Stem and other tributaries flowing into the reservoirs based on continuous data recorded from 01/01/1996 to 06/03/1999.

The first correlation developed was used for the time period of 11/06/1997 to 12/23/1997 and resulted in the following relationship (R²=0.9859):

For the next time period a new correlation was developed because the water temperatures were colder in January and the above correlation, although well representing the water temperatures in the Main Stem, would result in water temperatures below zero for this time period. The new correlation used for the time period of 01/11/1998 to 01/13/1998 was (R²=0.9761):

The correlations were used to fill in the data gaps and were combined with existing data from the Main Stem to create a water temperature file for the Bull Run River - Reservoir Model. For a complete plot of the water temperatures at the Main Stem Bull Run River see Figure 45.

Table 16 shows the gaged tributary temperatures used to represent the inflow temperatures for each distributed flow from the subbasins surrounding Reservoir #1 and Reservoir #2.
 

Meteorological Data

An important forcing function for modeling the Bull Run River-Reservoir system is meteorological data for the modeling time period. In the case of the Bull Run, meteorological conditions vary over the watershed, making it important to include as much data as possible in the model. Model calibration will be focused on the time period from 1996 to 1998 and later to include 1999. Meteorological data such as wind speed and direction, air and dew point temperature and cloud cover, with as high a measurement frequency as possible, is needed in the model to capture forcing functions within each day.

Some of the needed meteorological data has been collected at the Headworks facility on a continuous basis since September of 1998 or on a daily basis for the rest of the time period. In order to generate a more detailed set of data (more frequent than daily) to use in the model, correlations were generated between the Headworks and several other meteorological stations to see if various weather parameters could be related. Additionally meteorological data files were also generated for the Portland International Airport and for a meteorological station higher in the watershed, Log Creek.

Figure 46 below illustrates the locations of several regional meteorological stations that were used in correlations. These sites consist of National Weather Service meteorological stations and a U.S. Forest Service RAWS site. Figure 47 shows the location of meteorological sites within the Bull Run watershed and just outside the watershed. These sites include some of the same sites shown in Figure 46 and also include SNOTEL monitoring sites within the watershed.

Meteorological Conditions

Meteorological conditions for several years were tabulated and compared for the summer period of May 1^st to September 30^th. Statistics for the Portland International Airport, the Headworks and the Log Creek Site are presented in Table 17 through Table 19.
 
 

Single Regression Correlations

The first set of correlations were conducted to examine if there were any relationships between any two given meteorological stations. The focus of this work was to try to develop a better understanding of whether meteorological sites could be related to the Headworks meteorological station. Additionally correlations were attempted for two sites monitored in the Lower Bull Run River to determine if they could be related to any other meteorological sites. Two sets of individual correlations were attempted: those using continuously collected data and sites collecting data on a daily basis. In some cases when daily correlations were conducted, any continuous data available during that time was averaged over each day and incorporated in the daily data correlation analysis.

Table 20 shows the results of conducting the correlations, and graphs illustrating are provided in Appendix A. Overall, some of the correlations were limited by the extent of data available both in terms of the time duration of the data and the data types collected at each location. Daily minimum and maximum air temperatures, relative humidity, and precipitation were recorded at most sites.

                                           Continuous Data

The continuously collected data showed good air temperature correlations between the Portland International Airport and the Headworks facility, and Larson's Bridge and the Rd 14 Bridge in the Lower Bull Run River. Unfortunately the time period over which these correlations were made was relatively short. Continuously collected data did not produce very good correlations for relative humidity.

The daily data correlations shown in Table 20 cover much longer time periods but sacrifice detail in less frequent measurements. Air temperature was the only parameter which could be correlated for the time periods and stations listed. Excellent minimum and maximum air temperature correlations were developed between the Portland International Airport, Eagle Creek, and Government Camp and the Headworks facility. Correlations between the Log Creek site, Larson's Bridge and Rd 14 Bridge and the Headworks facility were not quite as good

Multiple Regression Correlations

The first regression involved using hourly air temperature data from the Portland International Airport (PDX) and from the RAWS Log Creek Site from the time period of 09/17/98 to 02/10/99. The maximum temperature at PDX during this time period was 27.2^o C at 4 PM 9/30/98. The maximum temperature at Log Creek was 25.6^o C at 3 and 4 PM 9/30/98. The maximum at Headworks was 25.8^o C at noon on 9/22/98. (The temperature at Headworks at 4 PM 9/30/98 was 25.6^o C.) The correlation resulted in the following equation (R²=0.8958):

Another regression was conducted using the same two parameters and the daily minimum and maximum air temperatures recorded at the Headworks facility on the Reservoir #2 Dam. The minimum air temperature was assumed to occur at 4 am and the maximum air temperature was taken to occur at 4 p.m. The data for all three parameters used in the regression covered from 09/17/98 to 12/10/98. This generated the following equation for hourly air temperatures at the Headworks facility (R²=0.8839):

Hourly wind speed data from the Portland International Airport was then added into the regression analysis to determine if a better relationship between hourly air temperature at the Headworks and the airport could be achieved. The resulting equation, using data from 09/17/98 to 02/10/99, (R²=0.8989) was

Hourly wind speed data from the RAWS Log Creek site was then incorporated into the regression analysis for the same time period and resulted in the following equation (R²=0.9001):

The last two equations result in no real improvement over the previous regressions. Additionally, given the added parameters to each regression analysis conducted there is very little improvement over the original regression which included the air temperature data from the Log Creek and Portland Airport sites. When the first multiple regression presented above is compared against individual regressions in Table 20 the analysis for air temperature shows only a 3% improvement using the multiple regression over the direct correlation between the Headworks and the Portland Airport.
 
 

Meteorological Files Generated for CE-QUAL-W2

The model uses air temperature, dew point temperature, wind speed and direction, and cloud cover in the model as meteorological forcing functions. Precipitation data is not used in the model but varying inflows to the reservoirs from tributaries represent its influence. In order to better understand the influence of the meteorological conditions on the model and to make sure the conditions most accurately represent what is occurring in the Bull Run System, several meteorological files were generated for the model based on different weather stations.

PDX Meteorological Data File

Hourly meteorological data from the Portland International Airport were used to develop one of the meteorological files. The data included wind speed and direction, air temperature, dew point temperature, and cloud cover recorded on a half hourly basis.

Log Creek Meteorological Data File

Another meteorological data file consisted of data from the Log Creek RAWS weather station high in the Bull Run watershed. This site is closest to Reservoir #1 and the proposed Reservoir #3. The U.S. National Forest Service RAWS data consisted of wind speed and direction, air temperature, relative humidity and barometric pressure. For the time period under consideration, the air temperature, wind speed and direction were used in generating the meteorological data file. Cloud cover data was not available at the Log Creek site so cloud data was used from Portland International Airport. Since dew point temperature was not available at this site the relative humidity and air temperature data were used to calculate the dew point temperature using Equation ( 5 ) (Singh, 1992):
 

( 5 )

where T_a and T_d are the air temperature and dew point temperature respectively and RH is the relative humidity. The dew point temperature was calculated from 1/1/96 to 2/17/99 with the exception of the time period of 1/1/96 to 6/5/96 because of errors in the Log Creek relative humidity data. In order to derive the dew point temperature for this time period a correlation was developed between the Log Creek air temperature and calculated dew point temperature and the air and dew point temperatures at Portland International Airport (PDX). The correlation resulted in the following equation (R²=0.7456) :

This resulted in a meteorological file for the model that was based predominately on data from the Log Creek site with cloud cover data supplied by Portland International Airport.

Headworks Meteorological Data File

Most of the meteorological data recorded at the Water Bureau Headworks is recorded daily. More frequent data is needed in order to more accurately capture the meteorological dynamics for the CE-QUAL-W2 model. The data record consists of daily minimum and maximum air temperatures, wind run and precipitation. For the time period of 9/17/98 to present more frequent data were recorded at the Headworks facility with additional parameters. The data consisted of minimum, maximum, and average air temperatures, relative humidity, wind speed, and direction, barometric pressure, and solar radiation on a continuous basis. The wind instrument was found to be not working properly so wind speed and direction data for some of this time duration were invalid. For the meteorological file required in the model, the air temperature and relative humidity were again used to calculate the dew point temperature at the Headworks using the equation by Singh, 1992. Cloud cover data was used from Portland International Airport since no cloud cover data was available at the Headworks.

Precipitation - Single Regression Correlations

It is important to understand how the precipitation measurements at the Headworks facility in the Bull Run can be related to other gauging stations for several reasons. The precipitation data record at the Headworks is extensive on a daily basis and there is considerable data recorded on a continuous basis as well. But the Headworks location may not represent the precipitation levels received across the watershed. In order to more accurately model the river-reservoir system a better understanding of the precipitation levels across the watershed needs to be achieved. Depending on the precipitation variability (and other meteorological data variability) across the watershed more than one meteorological file may be used in the model.

The first aspect of this analysis was to try to relate the precipitation record from the Headworks to other precipitation gages in the area. Figure 46 and Figure 47 indicate the station locations used in this analysis. Continuous precipitation records and daily records were correlated against the Headworks gage station as shown in Table 21. Correlation graphs can be found in Appendix B.

                                                     Continuous Data
 

                                                     Daily Data
 

Table 21 indicates that there is relatively little correlation between the Headworks and other local precipitation gage locations. The best precipitation correlation achieved was between the Brightwood site and the Headworks facility, but this station was discontinued in 1981 so more recent data is not available.

Precipitation - Multiple Regression Correlations

One regression conducted to try to generate a continuous precipitation record for the Headworks involved using precipitation data from the Government Camp and Brightwood precipitation gage stations since these are the only two locations close to the Bull Run that have continuous monitoring. Hourly data from both sites and Headworks were used from 10/01/75 to 11/27/95 and resulted in the following relationship (R²=0.236):

Based on the calculated correlation coefficient, even with a multiple regression, there is not much of a relationship between the precipitation stations.

The meteorological data analysis shows the correlations between stations were close for some parameters but not for others. Additionally, the water quality model uses air temperature, dew point temperature, wind speed and direction, and cloud cover. Although there were a lot of data available for the region, not all of these parameters can be used by the model or have parameters useable for the model measured at other sites. In examining the system and conducting the analysis there are differences in meteorological conditions between the Portland International Airport, the Headworks facility and the Log Creek site. Longer term comparisons could not be made with the Headworks facility or the Lower Bull Run River due to the recent installation of new continuous weather stations at both locations. As more data is collected additional comparisons might be possible between these sites within the watershed. Due to the various forcing functions influencing the water quality model, having more frequent meteorological data than daily data is valuable. Unfortunately, before the new weather station was installed at the Headworks, only daily data existed. In general, the meteorological analysis has provided a much better insight into the spatial variability of the meteorological data and its importance in modeling the river - reservoir system.

Bull Run Databases

Databases were created to store water quality, meteorological and hydrological data on the Bull Run watershed. Four databases were created: the Lower Bull Run River, Meteorology, Reservoirs and Tributary Inflows. The databases also store monitoring site information and locations for use in GIS. Details of the databases' contents can be found in Appendix J.
 
 

Lower Bull Run River Inflows

The Lower Bull Run River subbasin was delineated in order to determine where surface water enters the lower river. Figure 78 and Figure 79 show maps of the delineated watershed. Subbasin 12 (the Little Sandy River) extends well off of Figure 78. It was cut in order to show the rest of the watershed in more detail. The subbasins were created in a geographic information system using a USGS map of the area and a digital contour map.

Flows were associated with these subbasins based upon their basin areas. Correlations were developed between the measured flows on the Little Sandy River divided by the watershed area and several USGS gage flows in the Bull Run Watershed divided by their respective watershed areas. The correlation with flow on the South Fork of the Bull Run River gave the best correlation coefficient (R² = 0.9297). The equation used in the correlation is

This relationship was solved for alpha (a =0.80596) by drawing a correlation between each side of the equation, and was used to create flow coefficients for each of the subbasins.

Using the subbasin map for the Lower River, Figure 78, areas were associated with each subbasin using a geographic information system with results shown in Table 42. The subbasins were then grouped into logical area sets which would be contributing flow to the Lower River as either a point source such as a creek or as a distributed flow into the river. Table 43 shows the subbasins that were grouped together and their surface area summed and divided by the area of the South Fork basin (15.72 mi²) times a to obtain the flow coefficients. The flow coefficients were then used with the flows at the South Fork gage station to calculate the flows for each group of subbasins.
 

Lower Bull Run River Inflow Temperatures

Water temperature measurments were made at only a few tributaries entering the Lower Bull Run River. Therefore water temperatures were assigned to the inflows based on the proximity of the subbasins to the station where data was collected. The station used for each inflow is shown in Table 44.

Because the temperature records for these stations were only recorded in 1998 and 1999, correlations between the inflow temperatures and water temperature data at the South Fork gage station (USGS14139800) were developed to fill in the gaps between 1/1/96 and present. Table 45 shows the correlation equations that were used for each station.