3.1.2 Surface Water Temperature

icon_water_123.jpg

Goal: A. Maintain and improve water quality and supply to sustainably meet the needs of natural and human communities

Objective: 1. Maintain water quality for healthy aquatic systems

WAF Attribute: Physical/Chemical

What is it?

Surface water temperature is a constant variable of any waterbody and can be measured and summarized in various ways. Maximum seven day average daily maximum (7DADM) was calculated for each year and for each subwatershed. Each year was represented by the highest seven-day average maximum water temperature within the year. This metric is identical to the Maximum Weekly Maximum Temperature metric in the literature.

Why is it Important?

One of the consequences of increased withdrawal of river water for human uses is an increase in water temperature due to lowered volume. Increase of river temperatures from their natural levels has far-reaching effects on local ecology, including alteration of community processes and facilitating invasion by exotic species (Poole & Berman 2001). Restoring natural flow regimes and thus natural temperatures is critical to restoring a healthy natural system.

Native salmonid species are of great ecological, economic, and cultural importance to local communities. They also serve as strong indicators of habitat quality and integrity in river systems, particularly with regard to water temperature, sediment load, and barriers to passage. They are well-studied, including behavioral and physiological responses to temperature extremes. The Central Valley spring-run Chinook salmon is listed as a threatened species under the Endangered Species Act (ESA), giving them a high priority for restoration. The main threats to the remaining populations are loss and degradation of habitat. In particular, rising water temperature combined with upstream dams has greatly diminished available juvenile summer habitat. Within the Feather River Watershed, only two populations persist. One, in the Feather River itself, is completely dependent on the Feather River Fish Hatchery to maintain itself. The other, in the Yuba River, is of unknown status. There are occasionally spawning salmon in the Lower Bear River.

Maximum water temperature is a critical part of habitat quality for salmonids. Temperature affects every aspect of salmonid biology, from feeding and growth rates to migration and spawning, and stress levels and survival (Carter 2005). Rainbow trout, for example, are more severely impacted by temperatures in excess of 20°C than by fishing pressure (Runge & Peterson 2008). Upstream diversion of water for human usage increases downstream temperatures, as the lower remaining volume warms more quickly. Due to upstream barriers such as dams, only less-suitable, high-temperature regions are available for spawning and summer feeding. Anthropogenic temperature increases have been identified as key contributors to salmon decline (USEPA 2003).

What is the target or desired condition?

USEPA suggests as a guideline that a river sustaining salmonid populations should not have 7DADM temperatures over 18°C to avoid impairment of salmon health. Similarly, migratory portions of the river should not exceed Maximum Weekly Maximum Temperatures of 20°C and temperatures greater than 22°C will cause broad mortality (USEPA 2003). For core rearing areas in mid-to-upper parts of the river basin, a maximum of 16°C may be appropriate. Experimental studies indicate that spawning temperatures up to 16.5°C do not have deleterious effects on juvenile salmon, but mortality increases markedly after that point (Geist et al. 2006). These temperature guidelines, along with additional information from Brett et al. (1982), were used to convert monthly maximum 7DADM into a 0-100 scale.

A score of 100 is equivalent to the USEPA’s stated protective criteria of 18°C 7DADM for secondary foraging/rearing areas. A score of 0 will be equivalent to 25°C 7DADM, the lethal point for juvenile Chinook salmon. Intermediate scores were scaled using an adaptation of the Brett et al. (1982) growth curve (Figure 1). Only the right side of the curve was used; temperatures below the USEPA protective criterion were still scored as 100. Brett et al. (1982) estimate that natural populations of Chinook feed at roughly 60% of saturation (or R=0.6, the lowest growth curve). Because of daily temperature fluctuation, 7DADM temperatures are equivalent to constant laboratory temperatures roughly 1-2°C colder (USEPA 2003).

Figure 1: Chinook salmon growth curve (Brett et al. 1982)
Growth rates at different temperatures for three feeding levels (R=0.6, 0.8, and 1.0). Rmax (R=1.0) represents satiation feeding, with R=0.6 closer to natural feeding levels

The scaling curve is shown in Figure 2. The scaling curve does not exactly match the growth curve, due to the temperature thresholds for 0 (25°C) and 100 (18°C). Temperatures for the growth curve were adjusted upward by 1.5°C to adjust for the use of 7DADM measurements. These scores apply only to summer maximum temperatures.

Figure 2: Water Temperature Scaling Curve
This curve is approximated from salmonid growth/survival data in Brett et al. 1982. It converts 7-day average daily maximum temperature to a 0-to-100 score. The formula for temperature (x) conversion to score is 100 – r(x-K)2, where r = 2.041 and K = 18°C.

What can influence or stress condition?

The major factor which raises water temperature is decreased flow within the river. Low water volume allows the sun to warm the river much faster, and temperatures increase rapidly as the water moves downstream. Prolonged decreased flow (as opposed to seasonal variations) is most often due to human water use; water is retained in reservoirs and diverted to urban centers or for agricultural use, and only a small fraction is released into the original channel. Land-uses can contribute to higher temperatures, including logging, agriculture, and urban development. Increasing temperature due to climate change is another possible factor.

What did we find out/How are we doing?

The current states of the subwatersheds are shown in Table 1 and Figure 3. Scores ranged from a low of 20 for the Lower Yuba to 100 for Deer Creek and the Lower Feather. Many of the sampling sites were excluded from analysis due to data limitations. Many sites had only one year of data, sometimes represented by a single point. Most problematic was the prevalence of monthly samples at irregular times, which clearly do not represent temperature maxima. However, there were sufficient daily datasets from each subwatershed to perform trend analyses. Due to the aforementioned data limitations, only current state assessments and annual Mann-Kendall analyses were performed for each subwatershed. More details are available in later tables.

Table 1: Report Card scores for water temperature for subwatersheds

Goal Measurable Objective Subwatershed Score
A. Maintain and improve water quality and supply to sustainably meet the needs of natural and human communities 1) Maintain water quality for healthy aquatic systems — water temperature NFF 90
EBNFF 32
MFF 28
LF 100
NY 87
MY 87
SY 88
DC 100
LY 26
UB 79
LB 82

Figure 3: Water temperature condition scores across subwatersheds

Trends Analysis

The majority of subwatersheds had positive temperature trends, but only one was statistically significant ( Table 2). This was the Lower Feather, which had the second highest score for current state. Deer Creek had a significant negative slope, but the limited data available indicate an unreliable trend. Regional Mann-Kendall analysis (section 4.3) was conducted on each subwatershed, using the individual sites as separate regions. Only datasets with three or more years were included, to minimize noise and avoid a disproportionate contribution from the many 2008-2009 datasets. As suggested by the results from the individual regions, the overall trend for the entire watershed was positive but not significant (tau-b = 0.011, p = 0.728).

Table 2: Regional Mann-Kendall trend analysis. “Tau-b” is a Mann-Kendall test statistic. “S.N.” refers to number of sites.

Subwatershed Tau-b Significant? p-value Slope Magnitude Years S.N. Confidence Remarks
DC -0.150 Yes 0.050 -0.156 2001-2009 13 Medium Many datasets, but false maxima
ENFF 0.149 No 0.107 0.234 1999-2008 9 High All true maxima
LB -0.326 No 0.081 -0.065 1968-2003 1 Low Insufficient data
LF 0.262 Yes 0.026 0.195 1964-2003 4 Low Few data, false maxima
LY -0.021 No 0.890 -0.004 1973-2009 7 Low False maxima
MFF 0.000 No 0.699 -0.198 2001-2007 4 Medium Few data, but true maxima
MY 0.105 No 0.466 0.06 2001-2009 6 Low False maxima
NFF -0.333 No 0.564 -0.184 2001-2003 3 Medium Few datasets, but true maxima
NY 0.042 No 0.771 0.04 2000-2009 6 Low Few datasets and false maxima
SY 0.020 No 0.721 0.018 1966-2009 47 Medium Many datasets, some true maxima
UB 0.238 No 0.296 .450 2000-2003 5 Low Few datasets and false maxima

Most datasets, particularly the true maxima, were gathered within the last 10 years. Consequently, these analyses rely heavily on the significance of recent trends. However, there are some historical monthly datasets available through USGS sites which extend back to the early 1960s. These are shown in Figure 4. There are large gaps in the data, and the temperatures are not true maxima. Nonetheless, in the interest of historical comparison a Mann-Kendall analysis was performed on each site. The results are given in Table 3. Two sites (out of seven) showed significant positive trends, both in the Lower Feather.

Figure 4. Annual maximum temperature (°C) from seven long-term USGS datasets

Table 3: Regional Mann-Kendall analyses for seven USGS datasets. “Tau-b” is a Mann-Kendall test statistic.

RU Site Years Tau-b Significant? p-value Slope Magnitude
LB Bear near Wheatland 1968-2003 -0.325 No 0.082 -0.084
LF Feather near Nicolaus 1978-2000 0.071 No 0.902 0.021
LF Feather at Oroville 1957-1978 0.481 Yes 0.003 0.493
LF Feather near Gridley 1964-2002 0.186 No 0.269 0.125
LF Feather at Yuba City 1964-2002 0.467 Yes 0.018 0.281
LY Yuba at Marysville 1973-2002 0.000 No 0.917 0.041
SY Yuba at Jones Bar 1966-2004 0.066 No 0.784 0.078

Temporal and spatial resolution

Temperature is monitored throughout the Feather River Watershed, but not consistently over time or space. For example, there are few sites in the very large North Fork Feather and many in the comparatively small South Yuba. In addition, temperatures are collected using a combination of monthly grab samples and continuously monitoring Hobo-temps (thermometers left in the waterways). Continuous monitoring provides the most consistent source of temperature data and indicator calculation, but it is conducted on only a few sites. A critical feature of watershed-wide monitoring would be the establishment of a network of continuous temperature measuring devices that covers all important waterways and times of the year.

Figure 5: Distribution of temperature monitoring stations used as sources of data from throughout the watershed

How sure are we about our findings?
(Things to keep in mind)

The overall condition assessment based on 7DADM tells part of the story, but is best calculated based on complete data-sets. Water temperature is a straightforward parameter to measure, but is complicated to interpret. For example, the North Yuba received a lower condition score than the South Yuba (Table 1), but maximum temperatures in the South Yuba ( Figure 6B) can get higher than in the North Yuba (Figure 6A, Table 4). Similarly, Deer Creek received a score of 100, but has also experienced high maximum temperatures (Table 4). The most significant problem is the lack of long-term continuous temperature data for all sites. Although continuous data exists for many sites, it is limited to only the most recent years. The majority of sites have only single daily or monthly measurements, which makes it impossible to estimate a reliable maximum temperature. Scores based on these “false maxima” are likely to under-represent temperature problems in the watershed. The lack of reliable long-term time series makes trend estimation difficult as well.

The sites included were not edited to provide balanced spatial resolution, and some monitoring stations may not be representative of the subwatershed. For instance, the Lower Yuba score includes temperatures from Dry Creek. Dry Creek has higher temperatures than the Lower Yuba, and excluding its data would give a final score for the Lower Yuba of 41, not 26. Due to these and other factors, subwatersheds varied broadly in the confidence in findings (see Table 4). Subwatersheds with the lowest confidence scores are: Deer Creek, Lower Bear, Lower Feather, and North Fork Feather. These regions typically had fewer sampling sites, or lacked data for true temperature maxima. Highest confidence subwatersheds are: East Branch North Fork Feather, Middle Yuba, South Yuba, and Upper Bear. Differences in quality of data may explain some of the results, such as the high score for the Lower Feather compared to the Lower Yuba.

Figure 6: Sample data from remote electronic thermometers (“Hobo-Temps”) in A) the North Yuba River and B) the South Yuba River

A. North Yuba below Fiddle Creek:

B. South Yuba at Bridgeport:

Technical Information

Data sources:

  • Feather River: USGS daily temperatures 1957-2009, FRCRM/DWR daily temperatures 2000-2008, Nevada Irrigation District (NID), Federal Energy Regulation Commission (FERC) summer temperatures 2008-2009
  • Yuba River: SYRCL monthly data 2000-2009, daily data 2007-2009, USGS daily temperatures 1957-2009, NID FERC summer temperatures 2008-2009
  • Bear River: USGS daily temperatures 1957-2009, Nevada Irrigation District (NID), Federal Energy Regulation Commission (FERC) summer temperatures 2008-2009

Data Manipulation:

Units: Temperature data were converted into Celsius where necessary.

Type of data: Analysis is meant to be on temperature maxima only. Non-maximum data, such as daily averages or single measurements, were treated in one of two ways:

  • If data set was inferior or redundant, data were excluded from analysis. Generally excluded were: data sets with only one year of data; data sets with no data within the last 10 years; and false maxima data in regions with 5+ sites with true maxima.
  • If data set was desirable (i.e. the particular region or time period had little alternative data), then data were be treated as if they represented daily maxima, but this was noted in the analysis and the data confidence evaluation.
  • Unrealistically extreme values (>>30°C) were assumed to be erroneous readings from exposed temperature probes, and were removed.

Temporal aggregation:

Data were aggregated temporally as follows:

Sub-daily data (i.e. hourly, 15min, etc.): included only daily maximum.

Daily data: Daily maxima were averaged over seven-day periods to form a rolling 7DADM. Standard deviations calculated for each average. Averages started with the seventh day in a series, and then moved forward until the final day. Some averages extended into two months, but the 7DADM was associated with the final day in the average. For months with fewer than seven days of reporting, a shorter average was used, though this was reflected in the standard deviation. Missing days were accommodated in a similar manner, though single missing days were treated as an unbroken average.

Weekly data: Whether this was a single measure of weekly temperature, or a 7DADM point, only the maximum value was used to represent the month. Standard deviation for the maximum point was preserved.

Monthly data: Seasonal Kendall and month-by-month trend analysis were carried out on monthly maxima data, when data permitted. Trends were reported for each season or month, as well as the overall yearly trend.

Annual data: For annual analysis, the Mann-Kendall trend analysis was used on yearly maximum data. Standard deviations were maintained from the 7DADM averages when possible.

Subwatershed aggregation: Data within a subwatershed was assumed to represent independent sampling of the subwatershed, and was used to calculate aggregate scores for that subwatershed. Number of sites was considered when assessing confidence measures.

Table 4: Basic statistics for subwatershed condition assessments. “SD” refers to standard deviation; “95% C.I.” refers to 95% confidence intervals. “S.N.” refers to the number of monitoring sites.

Subwatershed Name Mean 7DADM (°C) Minimum 7DADM Maximum 7DADM SD S.N. 95% C.I. Score Confidence Remarks
Deer Creek 18.38 13.12 28.89 4.36 17 +/-2.07 99.70 Low False maxima (score could be lower)
East Branch North Fork Feather 23.77 17.03 29.81 3.18 18 +/-1.47 31.98 High All true maxima
Lower Bear 21.00 16.50 25.50 6.36 2 +/-8.82 81.63 Low Few data, no true maxima
Lower Feather 18.42 11.30 24.00 5.07 6 +/-4.05 99.65 Low Few data, no true maxima
Lower Yuba 24.01 18.21 28.66 4.78 5 +/-4.19 26.25 Medium True maxima, but few data
Middle Fork Feather 23.93 17.33 29.13 4.82 5 +/-4.22 28.13 Medium True maxima, but few data
Middle Yuba 20.53 11.90 27.76 5.69 16 +/-2.79 86.89 High Plentiful true maxima
North Fork Feather 20.24 14.06 26.22 6.09 3 +/-6.89 89.81 Low True maxima, but very few data
North Yuba 20.54 16.03 24.01 3.72 6 +/-2.98 86.82 Medium True maxima, but few data
South Yuba 20.40 10.91 29.75 4.49 56 +/-1.18 88.27 High Very extensive sampling, true maxima
Upper Bear 21.17 13.09 29.68 4.15 27 +/-1.56 79.44 High Good sampling, many true maxima

Trends Analysis Reporting

The primary values reported were Mann-Kendall trends and Kendall B estimated trend slope, with confidence intervals. When performing regional analysis, trends for subunits were reported along with overall trend. Only time series with 3+ years of data were used for trend analysis.

Citations

Brett JR, Clarke WC, Shelbourn JE. 1982. “Experiments on thermal requirements for growth and food conversion efficiency of juvenile chinook salmon, Oncorhynchus tshawytscha.” Can Tech Rep Fish Aquat Sci 1127. 29 p.

Carter, K. 2005. “The effects of temperature on Steelhead trout, Coho salmon, and Chinook salmon biology and function by life stage.” California Regional Water Quality Control Board

Geist, D.R., C.S. Abernethy, and K.D. Hand. 2006. “Survival, development, and growth of fall Chinook salmon embryos, alevins, and fry exposed to variable thermal and dissolved oxygen regimes.” Transactions of the American Fisheries Society 135:1462-1477.

Poole, G.C. and C.H. Berman. 2001. “An ecological perspective on in-stream temperature: natural heat dynamics and mechanisms of human-caused thermal degradation.” Environmental Management 27:787-802.

Runge, J.P. and J.T. Peterson. 2008. “Survival and dispersal of hatchery-raised rainbow trout in a river basin undergoing urbanization.” North American Journal of Fisheries Management 28:745-767.

U.S. Environmental Protection Agency (USEPA). 2003. “USEPA Region 10 Guidance for Pacific Northwest State and Tribal Water Quality Standards.” Region 10, Seattle, WA. USEPA 910- B-03-002. 49pp. Available online at: http://www.epa.gov/r10earth/temperature.htm.

U.S. Environmental Protection Agency (USEPA). 2001. “Issue Paper 5: Summary of Technical Literature Examining the Physiological Effects of Temperature on Salmonids.” Region 10, Seattle, WA. USEPA 910-D-01-005. 119pp.