Stream chemistry concentrations and fluxes using proportional sampling in the Andrews Experimental Forest, 1968 to present

ORIGINATOR: Richard L. Fredriksen
OTHER RESEARCHER: Julia A. Jones, Alba Argerich, Kate Lajtha, Stanley V. Gregory
ABSTRACTOR: Donald L. Henshaw
METHOD CONTACT: Craig Creel, Greg Downing, Cameron R. Jones
FORMER INVESTIGATOR: Frederick J. Swanson, Alfred B. Levno, W. Arthur McKee, C. Wayne Martin, Kristin L. Vanderbilt
Cooperative Chemical Analytical Laboratory (CCAL)
9 Oct 1984
26 Feb 2013
Discharge, Dissolved load, Hydrology/Water, Logging, Long-Term Ecological Research (LTER), Nutrients, Solution transport, Streamflow, Timber harvesting, Water, Water chemistry, Inorganic nutrients, Inorganic nutrients
Examine stream nutrient concentrations and outputs in small watersheds, evaluate the significance of nutrient losses from a forest soil-plant system after timber harvest compared to reference watersheds, and examine the nutrient and solute budgets for small watersheds.
Experimental Design - CF002:

Long term trends in stream water nutrient concentrations and fluxes from gaged experimental watersheds are monitored. Stream water samples are collected over a three week period in proportion to streamflow. Several of the watersheds have been harvested during the study period. Nutrient concentrations of analytes are conducted at the Chemical Cooperative Analytic Laboratory (CCAL) in Corvallis, Oregon.

The H.J. Andrews Experimental Forest is within the Lookout Creek Basin (6250 ha). Lookout Creek as well as nine smaller watersheds within the basin have gaging stations. Dominant overstory vegetation is Douglas-fir. Proportional stream water samples are collected at groups of small watersheds with differing harvest treatments: WS 1 and 2; WS 6, 7, and 8; WS 9 and 10. Watersheds 2, 8, and 9 are unharvested reference basins. Mack Creek is a larger basin (600 ha) with old-growth vegetation and Lookout Creek is a 5th order watershed with a USGS gaging station. Drainage areas for the small watersheds range from 9 to 100 ha and elevations range from 460 to 960 m.

Watersheds stream sampling history in the H. J. Andrews Forest:

  • Watershed 1: 100% clearcut 1962-1966, burned 1966, no roads; sampling record WY 2003 to present
  • Watershed 2: Control; sampling record WY 1981 to present
  • Watershed 6: 9% roaded 1963, 100% clearcut 1974; sampling record 1972 to 1987, restarted 2003 to present
  • Watershed 7: 60% shelterwood cut 1974, remaining 40% overstory cut in 1984, thinned 2001; sampling record 1972 to 1987, restarted 2002 to present
  • Watershed 8: Control; sampling record WY 1972 to present
  • Watershed 9: Control; sampling record WY 1969 to present
  • Watershed 10: 100% clearcut 1975, unburned; sampling record WY 1969 to present
  • Mack Creek: Control, third-order watershed; sampling record WY 1981 to present
  • Lookout Creek: Approximately 25% harvested; sampling record WY 2005 to present

See for more information.

Citation: Fredriksen, R. L. Study Plan--Nutrient cycle of a small watershed in the Douglas-fir region of the western slope of the Cascade range in Oregon. 1967. PNW-FSL-1602-22
Field Methods - CF002:

Streamwater is sampled from each stream at a rate proportional to the streamflow rate. Samples are collected in 5 gal. polyethylene carboys and stored in an insulated housing or cabinet at the gaging station. Beginning in 2005, the original site-developed samplers are being replaced with automated samplers (SIGMA type).

Composite samples, collected proportionally to streamflow over approximate 3-week periods, contain representative sampling of the sediment and dissolved constituents carried by the stream over the collection period. Beginning 1988, stream proportional samples are collected weekly and refrigerated. Refrigerated samples are stored at the Andrews and transported to Corvallis for analysis every 3 weeks. Beginning in WY 2009 due to resource limitations, cold-weather samples (Nov-Mar) are left in the gage house for the entire 3-week interval, but samples are collected weekly and refrigerated throughout the rest of the year. The weekly samples are combined and analyzed as a composite 3-week sample. Before 1988, samples remained at the gage houses until transported to Corvallis for analysis. During high flow periods the samples were collected at shorter intervals to prevent the carboys from overflowing.

Historically, samples were collected using a battery powered, in-house constructed sampler designed by Richard Fredriksen (Fredriksen 1969). The interface between the sampler and the recorders was an array of magnetic reed switches which were activated by magnets attached to the counter weights of the recorders. There were 20 segmented proportions equal to discharge increases of 1/20th of the expected maximum discharge. For each proportional increase in flow, the number of samples taken in the base time period increased by one. Streamflow is measured continuously with Leopold-Stevens A-35 recorders and Stevens PAT with Campbell CR-10 data logger.

The original sampling rate based on streamflow used in these original reed switch samplers is now represented in the new SIGMA samplers at our long-term sampling sites. Sampling at WS1 and Lookout Creek started in 2003 and 2005 using the SIGMA samplers. WS2 was converted to the SIGMA type sampler with first composite sample on 28 June 2004 after a brief period of running the original and new sampler concurrently. Mack Creek was converted to a SIGMA sampler with first composite sample on 28 August 2007.

Samples from each week within the 3-week period are typically combined at the Andrews, but partial samples (i.e., a week where operational problems with the sampler create a non-proportional or incomplete sample) and other very large samples may be combined proportionally at the lab. All samples are assigned sample type codes (TYPE). Handling of samples and assigning sample type codes have been generally consistent since Dec 1988, although modification was made in August 1991. The rules for combining samples and assigning sample type codes are as follows:

All three weekly proportional samples for a given watershed sampler, whether partial (incomplete) or full (complete), are combined for the 3-week period. Successful operation of the sampler over all 3 weeks represents a complete or full (TYPE="F") sample. Any partial collection will cause the sample type designation to be partial (TYPE="P") for that entire three week period. Grab samples should be taken when no sample has been collected through the proportional sampler for any single week, sample type 'Grab (TYPE="G"). Grab samples will not be combined with other composite samples and are submitted separately to the lab. Grab samples are only analyzed if the sampler otherwise fails for the entire 3-week period. A carboy with three combined grab samples should be designated as grab (TYPE="G"). Proportional samples with less than 2 liters of water will be augmented at the lab with any grab sample that has been submitted separately and designated as augmented (TYPE="A") sample type. Note that no sample has been augmented (Type='A') since 1998. Should no sample be collected for any week, sample type is missing and listed as no analyses (TYPE='N'). Biased sample (Type 'B') type has been used infrequently since 2007 and indicates samples that were proportionally combined based on streamflow to create a 3-week sample (when crews could not get out to collect the sample after the 3rd week). From Dec 1988- Aug 1991, partial and grab weekly samples were analyzed independently of full weekly samples. Flux values for these 3-week periods are based on the weekly samples and have been calculated mathematically (in Entity 2) using current rules for combining samples in the lab.

Before Dec 1988 samples were only collected on a 3-week basis. Successful operation of the sampler over the 3 week period represents a complete or full (TYPE="F") sample. When the sampler failed or was out of operation for the entire 3 week period, a grab (TYPE='G') sample was taken to represent the entire period. When the sampler failed or was out of operation for part of the 3 week period, the sample is designated as partial (TYPE="P"). When the partial samples contained insufficient water for analysis, they were "augmented" (TYPE="A") with a grab sample to a level considered sufficient for conducting all analytical measurements. There are 5 cases when the sample contained insufficient water and were saved (TYPE='S') and combined (TYPE='C') with the following sample for analysis. In these cases, there are no independent analytical values for the saved samples (TYPE='S'). Flux values are created for the saved sample interval using streamflow from this interval and using the analytical values from the combined sample representing the 6-week interval.

Date and times listed as data_time in the data files refer to the end date of a collection period. Data is summarized by water year: October 1 to September 30. To allow summarization by water year, an artificial end-of-year sample point (TYPE="YE") is inserted on Sept 30 every year (Entity 2 flux data only) but is not an actual sample . Nutrient concentrations from the next analyzed sample (in October) are used with the streamflow total ending Sep 30 , the end of the wateryear, to calculate flux for this interval, which is designated TYPE="YE". Similarly, the flux value for this first sample of the new wateryear in October is calculated based on concentration values from the October sample and on streamflow beginning Oct 1.

Quality control samples were taken biannually (one in the winter and one in the summer) at one site each year between 1989 and 2007. Grab QAQC samples were collected above and below the station, and two samples collected from streamwater flowing through the sampler. One of the through-sampler collections was stored at the gagehouse and transported to the lab with the composite sample the following week. The other through-sampler sample and the grab samples were refrigerated at the Andrews until the composite samples are brought to CCAL in Corvallis for analysis. Quality control samples were analyzed using the same methods as the composite samples. these samples are not made public online but are available on request.

Instrumentation: 1) Fredriksen locally-designed battery powered proportional stream water samplers with reed switch assembly (1968-Present); 2) Sigma Model 900 standard portable sampler and a Campbell Scientific CR-10X datalogger (2003-Present). The Sigma sampler is a commercially available and portable sampler powered by a 12 volt battery that has a peristaltic pump, and a programmable control unit. The controller is programmed for flow-based, composite sampling to produce outputs for events that are captured by a datalogger. The controller is programmed to produce repeatable and accurate sample volumes. A record of the program sample history is stored in the controller, and can be reviewed by the operator. The sample storage container is located in the insulated base housing of the Sigma. The sampling system CR-10X is also used to measure and record stream stage height at the gauging station; 3) Streamflow is measured continuously with Leopold-Stevens A-35 recorders and Stevens PAT with Campbell CR-10 data logger.
Citation: Fredriksen, R. L. 1969. A battery powered proportional stream water sampler. Water Resourc. Res. 5(6)1410-1413, illus.
Laboratory Methods - CF002:

Laboratory procedures: Official compilation of water analyses begins with the WY 1969 (10/1/68). Originally, samples are analyzed for suspended sediment, alkalinity, pH, ammonia nitrogen, nitrate, nitrite, dissolved kjeldahl nitrogen, ortho (reactive soluble) and total dissolved phosphorus, sodium, potassium, calcium, magnesium, and silica. Sulfate was determined for WY 1972 and sulfate and chloride in WY 1979. Sulphur and Chloride were analyzed at the University of Washington beginning 1988 until 2004, and are now conducted by CCAL at Oregon State. Analysis of specific conductivity was added in Dec 1974 and total P on an unfiltered samples in Oct 1974. Analysis of total kjeldahl N on an unfiltered sample was started in June 1978. Total N is calculated from the total Kjeldahl and NO3-N until May 2005. Direct analytical measurement of Total dissolved N and Total N (unfiltered sample) begins in May 2005. Dissolved organic carbon analytical measurement begins in 2003. Aluminum, iron, manganese, and nitrite were below levels of detection and were subsequently dropped from analysis. In general, duplicates were run on all analyses until Oct 1983, and are currently conducted randomly.

Other notes on analyses: 1) From 1969 - 1973 samples were analyzed for free ammonia (FNH3) prior to being digested for total nitrogen (DON). The FNH3 was driven off and collected and analyzed separately. Then the sample was digested and analyzed for DON. For these years TKN represents the sum of DON and FNH3. Listed Total Kjeldahl N (TKN) values from 6/1/1970-6/21/1971 for GSWS09 and GSWS10 do not include free ammonia (NH3-N) and are shown as estimated. 2) A combination of factors cause the results for silica prior to 17 March 1983 to be labeled as "Q"uestionable. Historic records indicate that many, if not all, of the samples prior to that date were frozen prior to analysis for silica. Freezing samples prior to analysis for silica will reduce the final analyzed concentration of reactive silica (the form we have historically determined). Another complicating factor is that the methodology changed about this same time. Until 1982, a stannous chloride method was used to determine silica, and this method is more prone to variability due to reagent instability.

Until May 2005, the CCAL lab analyzed for total dissolved Kjeldahl nitrogen (TKN) and also for total Kjeldahl nitrogen (UTKN), which is performed using an unfiltered sample and includes any particulates (sediment, biota) present. All dissolved results exclude particulates of 0.7 microns or greater. TKN includes NH3-N, but does not include NO3-N. The following are mathematically calculated in the data set and coded as 'D' when this occurs:

  • Dissolved organic nitrogen (DON) = TKN - NH3N
  • Total dissolved nitrogen (TDN) = TKN + NO3N (directly analyzed after May 2005)
  • Total nitrogen (UTN) = UTKN + NO3N (directly analyzed after May 2005)
  • Particulate nitrogen (PARTN) = UTKN-TKN, or UTN-TDN (after May 2005)
  • Particulate phosphorous (PARTP) = UTP - TDP

Currently, all samples are filtered upon arrival at the Corvallis lab (CCAL) and pH, alkalinity, conductivity, and autoanalyzer runs (NO3-N, NH4-N, SI) are made immediately. All analyses are completed within 6 weeks. Generally, samples are stored at 0 degrees C if not analyzed immediately. In the first year of sample collection (1968-1969), water was stored at 0 degrees C until operation of the PNW-1653 local lab in spring, 1969.

Generally, there is much greater trust and consistency to analytical measurements beginning in 1983.


Citations for all Cooperative Chemical Analytical Lab (CCAL) procedures are shown here: CCAL Methodology

Processing Procedure - CF002 :

The new process for creating monthly summaries are listed in: P:\FSDB\HF04\work_don\monthly_chem\status.txt and the work was performed here in Dec 2010. The key program is pio_month.prg, but other processing is necessary first, and the pio_true_mon.prg is used to produce mean annual concentration. In VFP, Entity 1 CF00201_final is the raw concentration data CF00201_month is the raw concentration/interval data, raw data with 'M' and 'Y' points CF00201_year is the original CF00201_TRUE, raw data with 'Y' points Entity 2 CF00202_final is new mean monthly concentration data Entity 3 CF00203_final is mean annual concentration data Entity 4 CF00204_final is new total monthly flux data Entity 5 CF00205_final is the old Entity 2 data (CF00202_TRUE) - calculated flux for each sample

Laboratory Methods - CCAL: Alkalinity:
Description: APHA 2320, titrate to pH 4.5. Modifications: Use 0.02N Na2CO3 and 0.02N H2SO4
Instrumentation: Radiometer type TTT lc auto-titrator with glass pH electrode, calomel reference electrode and temperature compensator electrode
Laboratory Methods - CCAL: Ammonia :
Description: APHA 4500-NH3 G; EPA 350.1
Instrumentation: Technicon Auto-Analyzer II
Laboratory Methods - CCAL: Calcium :
Description: APHA 3111 D; flame atomic absorption spectroscopy. Modifications: nitrous oxide/ acetylene flame. Addition of 1 ml 50,000 mg/1 lanthanum oxide to 10 ml sample to control ionization.
Instrumentation: Varian SpectrAA220; Metals by Flame Atomic Absorption Spectrometry; Nitrous Oxide-Acetylene Flame Method
Laboratory Methods - CCAL: Diss. Organic Carbon :
Description: APHA 5310 B
Instrumentation: Shimadzu TOC-VCSH Combustion Analyzer
Laboratory Methods - CCAL: Chloride :
Description: APHA 4110 B; EPA 9056A
Instrumentation: Dionex 1500 Ion Chromatograph
Laboratory Methods - CCAL: Specific conductance :
Description: APHA 2510; Wheatstone bridge.
Instrumentation: YSI model 3200
Laboratory Methods - CCAL: Magnesium :
Description: APHA 3111 B; flame atomic absorption spectroscopy.
Instrumentation: Varian SpectrAA220; Metals by Flame Atomic Absorption Spectrometry; Direct Air-Acetylene Flame Method
Laboratory Methods - CCAL: Nitrate :
Description: APHA 4500-NO3 F; EPA 353.3. Cadmium reduction method.
Instrumentation: Technicon Auto-Analyzer II
Laboratory Methods - CCAL: Total nitrogen :
Description: Persulfate digestion and analysis by automated colorimetric analysis
Instrumentation: Technicon Auto-Analyzer II
Laboratory Methods - CCAL: pH :
Description: APHA 4500 H; Calomel reference electrode, glass pH electrode, temperature compensator.
Instrumentation: Radiometer type TTT lc auto-titrator with glass pH electrode, calomel reference electrode and temperature compensator electrode
Laboratory Methods - CCAL: Ortho phosphate :
Description: APHA 4500-P E; EPA 365.2. Ascorbic acid method.
Instrumentation: Milton-Roy 601 Spectrophotometer with 10 cm pathlength
Laboratory Methods - CCAL: Total phosphorous:

APHA 4500-P B; APHA 4500-P E; EPA 365.2. Modifications: microwave digestion 60 minutes, 50 ml analysis volume, Ascorbic acid method.

The laboratory has automated the method used for analysis of total phosphorus (total phosphorus for unfiltered samples and total dissolved phosphorus for filtered samples) as of 16 June 2010. Samples submitted to the laboratory will now be analyzed using the Technicon Auto-Analyzer II (the same instrument used for analysis of nitrate, ammonia, phosphate, silica and total nitrogen). We have discontinued the routine use of the manual method using the Milton-Roy 601 spectrophotometer at this time.

We ran the manual method through the end of calendar year 2010 in an attempt to run comparisons for projects with historic results from the manual method. Results indicate there is expectation for reduced UTP/TDP due to the analytical change.

Instrumentation: Milton-Roy 601 Spectrophotometer with 10 cm pathlength and the Technicon Auto-Analyzer II beginning June 2010
Laboratory Methods - CCAL: Potassium:
Description: APHA 3111 B; flame atomic absorption spectroscopy.
Instrumentation: Varian SpectrAA220; Metals by Flame Atomic Absorption Spectrometry; Direct Air-Acetylene Flame Method
Laboratory Methods - CCAL: Silica :
Description: APHA 4500-SiO2 E; Technicon industrial method 105-71W/B.
Instrumentation: Technicon Auto-Analyzer II
Laboratory Methods - CCAL: Sodium :
Description: APHA 3111 B; flame atomic absorption spectroscopy.
Instrumentation: Varian SpectrAA220; Metals by Flame Atomic Absorption Spectrometry; Direct Air-Acetylene Flame Method
Laboratory Methods - CCAL: Sulfate :
Description: APHA 4110 B; EPA 9056A
Instrumentation: Dionex1500 Ion Chromatograph
Laboratory Methods - CCAL: Suspended sediment :
Description: APHA 2540 D
Models/Algorithms Methods - CF002:

Water samples are collected proportionally to streamflow. To obtain mean monthly concentration of analytes, streamflow totals must be broken into intervals that start with the beginning of each month and end at the end of each month. Streamflow is totaled in centimeters from the beginning of each month to the time of the first proportional sample, and from the time of first proportional sample to either the time of the next proportional sample or the end of the month, whichever comes first. Then for each interval, a streamflow total and set of analyte concentrations are available. For the monthly markers at the beginning and end of each month, analyte concentrations from the proportional sample being collected at the month change are assigned. FOR example, analyte concentration values for a 3-week sample collected on Feb 5 would also be assigned to the end of January marker and would therefore be included in the January mean. Analytes from each interval within the month are weighted by streamflow to determine mean concentrations.

If samples represent 25 or less days of the month (because one or more samples are missing), the sample is coded 'P'artial. If samples are only present for 5 or less days of the month, the sample is coded 'N' for no sample. Similarly, any analyte that is not analyzed for at least 25 days in a month will be given an 'I'ncomplete qualifier tag for mean monthly concentration values. Flux values calculated from 'I'ncomplete sample analyte concentrations, but are based on the complete streamflow for the month will be considered 'E'stimated. 'I'ncomplete tags will only be used in the flux data when streamflow itself is incomplete for a month. Analyte concentrations and calculated fluxes based on 5 or less days in any month will be considered missing and qualified with an 'N' code no analysis).

Sample type codes for the monthly summary values are assigned as follows: F=Full, G=Grab, P=partial. Partial ('P') will have precedence over 'G'rab sample type code. That is, if any sample in a given month is listed as partial, the code will be partial. If all samples are Grab for the month, the code will be Grab. Only if all samples IN month are 'F'ull will the type be labeled 'F'ull.

For other analysis codes, any monthly concentration OR flux based on derived calculations will be qualified with a 'D'erived as part of the code. Additionally, collective analysis code for each analyte concentration is assigned and up to two other qualifier codes will be displayed if present for any analysis within the month in the following order of precedence: Estimated, questionable, below detection (E, Q, *). Also note that F=S=C (Saved and Combined samples are treated as 'Full' samples. Saved samples use the 'C'ombined sample analytical concentrations for flux calculation.

Key conversion: Kg/ha= 0.1 * mg/l * cm where cm = centimeters of streamflow over the watershed area, and mg/l = milligrams per liter nutrient concentration.


Watersheds 1 and 2 comprise a low-elevation (442 m to 1082 m) pair of experimental basins with Watershed 2 serving as the control. Watersheds 6, 7, and 8 comprise a high-elevation (863 m to 1190 m) set of experimental basins with Watershed 8 as the control. Watersheds 9 and 10 are a low-elevation pair of small watersheds with Watershed 9 as the control. Mack Ck is a third order, old-growth watershed.

Vegetation on Watershed 9 and Watershed 10 (prior to logging) consisted primarily of 450-year-old Douglas-fir mixed with hemlock and younger Douglas-fir. Mid elevation watersheds (1500 ft and 3600 ft). Watershed 2 and 9 are first to second order old-growth watersheds (headwaters of watershed 9 were burned in the 1890's). Old-growth stands are approximately 450 yr old and are dominated by Psme and Tshe at lower elevations and also by Abam at the higher elevations. Watersheds 2, 8, and 9 are undisturbed controls.

H. J. Andrews Experimental Forest, Watersheds 1, 2, 6, 7, 8, 9, 10, Mack Creek, and Lookout Creek in the Willamette National Forest
1-3 weeks
Ground condition

Description of the Sigma sampler and proportional water sampling system (2003-Present)

Detailed 1988 memo describing changes in the stream and precipitation sampling program

A table showing the history of laboratory instrumentation for analytical measurements

Silica data quality, 1969-1983

StreamChem DB