Evaluating Cumulative Effects of Logging and Potential Climate Change on Dry Season Flow in a Coast Redwood Forest

Silviculture techniques such as clear-cutting and selective logging are known to enhance runoff over a short time period. However, recent studies indicate that post-logging flow levels may instead drop below pre-logging flow levels as the forest re-establishes. These changes in flow levels can affect downstream water supply and modify habitat conditions for stream biota. As climate change alters rainfall regimes, scientists have become concerned that climate change may strengthen the effects of silviculturally induced flow changes. Using pretreatment calibrations between summer flows and antecedent precipitation indices (APIs), Reid and Lewis (2011) modeled the effects of six altered rainfall regimes on dry-season flows when combined with flow changes caused by selective and clear-cut logging within watersheds at Caspar Creek. The authors found that summer flows increased at a faster rate after selective logging when compared to clear-cutting. Yet, clear-cutting resulted in a larger increase in dry-season flow. Afterwards, the selective logging watershed experienced depressed flows while the summer flow in a partially clear-cut watershed remained higher than expected after logging. After a 22% reduction in annual rainfall, the area experienced a 23% reduction in flow under unlogged conditions, and a 41% flow decline after selective logging. Under the current climate regime, lower flows would be expected twice as often during the post-logging period. Additionally, the results indicate that a shift in seasonal rainfall distribution may increase or decrease dry-season flows when combined with logging effects even if annual rainfall levels remain constant. Overall, the authors’ findings suggest that forest managers should employ watershed-scale silvicultural strategies to reduce the risks of adverse dry-season flow changes when combined with the effects of varying rainfall levels.—Megan Smith
Reid, L.M., and Jack Lewis., 2011. Evaluating Cumulative Effects of Logging and Potential Climate Change on Dry-Season Flow in a Coast Redwood Forest. < http://www.fs.fed.us/psw/publications/reid/psw_2011_reid001.pdf>.

Water flow has been measured since 1962 at gaging weirs in the North and South Forks of Caspar Creek using a sequence of float gages (recorded using strip charts) and data-logged pressure transducers. Between 1971 and 1973, the 424ha South Fork watershed underwent 67% volume-selection logging. Throughout 1985 and 1986, 13% of the 473ha North Fork watershed was clear-cut. In 1989–92, an additional 37% of this watershed was clear-cut. Marine sandstones and siltstones underlie both watersheds, and most slopes are covered by 0.5–1.5 m deep clay-loam to loam soils. Annual rainfall averages 1,170 mm, and half runs off as stream flow. Approximately 95% of the rain falls between the months of October and May. This period also accounts for 95% of runoff. Minimum flow occurs in early October, but most streams are dry by June. As of 1960, the watersheds support 60–100 year old second growth stands that are dominated by coast redwood and Douglas fir. A map displaying the Caspar Creek Experimental Watersheds was constructed.
Calibrations established between North and South Fork flows for the pre-logging period were used to estimate expected South Fork flows after logging, and the observed deviations from expected values characterized the initial South Fork dry-season flow response to selective logging. However, because logging began in the North Fork after 1985 and South Fork flows had not returned to pretreatment levels at this time, the North Fork now lacked a paired control watershed. Therefore, treatment effects from the new North Fork experiment were unable to be evaluated at the weir gage. As a result, subwatersheds were utilized as controls. Gaging flumes were installed within these subwatersheds in 1984, and the area was logged soon after. However, subwatersheds were not gaged during summer months because the sites run dry. So, dry-season flow analysis depended on gaging records from the North and South Fork weirs.
The authors derived a method to estimate the expected dry-season flows at the weirs after the 1985 logging by using rainfall levels to predict pretreatment flows. An antecedent precipitation index (API) was used to predict these pretreatment flows. Reid and Lewis collected rain data from the Fort Bragg gage for summer rain records and from the South Fork gage for winter records. The records were combined to construct a continuous rainfall record from 1962 through 2008. Then, a suite of APIs with recession coefficients ranging from 0.993 to 0.600 was calculated from the rainfall record.
Because late-summer flow data were unavailable for years when weir ponds were drained, the authors selected three to five dates in August and September (for years with dry-season data) that had no rainfall during the previous 3 days, had greater than 9 mm of rain during the previous 30 days, and were more than 6 days a part. Mean daily flows on the selected dates during the pre-logging periods were regressed against the suite of APIs to identify the API that best predicted observed flows at each site. These calibrations resulted in estimations of expected August and September flows at the South Fork weirs (Les, L/km2-s) and the North Fork weirs (Len) for pretreatment (unlogged) conditions. Les = 0.0143API0.985 – 0.0320 and Len = 0.0272API0.977 + 0.0366. API0.985 was calculated using a recession coefficient of 0.985 and API0.977 was calculated using a recession coefficient of 0.977. The flow changes after logging treatments were calculated using ratios of observed flows to those expected for forested conditions. Graphs displaying calibration relations between late-summer flows and APIs in the North Fork and South Fork were constructed.
To evaluate interactions between logging-related flow changes and those arising from potential climate change, the authors constructed six plausible rainfall regimes by modifying the existing rainfall record to reflect altered annual rainfalls and changes in the seasonal rainfall distribution. Scenarios were selected within the observed range of variability so that the API model could describe them. Indirect interactions between altered rainfall effects and other changes climatic attributes were not considered.
The 24 wettest years on record showed an annual average 22% higher than the 48 year average. Therefore, the authors constructed one 48-year record by multiplying the recorded daily rainfalls by 1.22 and by 0.78. Rainfall in April and May accounted for an average of 10.4 percent of the annual rain over the 48-year record and 14.9 percent during the 24-year of that record with the highest percentages. Another record was constructed by increasing April–May daily rainfalls by a factor of 14.9/10.4 and 5.8/10.4 while multiplying rainfalls in other months by 85.1/89.6 and 94.2/89.6.
June and July account for 1% of the annual rainfall at Caspar Creek. Years with lower than the median proportion of summer rain show a mean percentage of 77% lower than average. Two additional records were constructed that reflected a 77% increase and decrease in June and July rainfall without modifying annual rainfall levels.
For each rainfall sequence, rainfall in August was set to 0 and values of API0.985 and API0.977 were calculated for September 1 of each year. Equations 1 and 2 were used to estimate expected flow under unlogged conditions at each weir on that date for each of the six API sequences. The proportional logging-related changes in flow, having been defined as a function of time after clear-cutting or selective logging, were applied to the climatically altered flows predicted for unlogged conditions to estimate the combined effects of logging and hypothetical changes in rainfall.
Reid and Lewis found that late summer flows increased after selective logging in the South Fork and remained high after eight years. Fifteen years after selective logging, flows dropped below levels expected for unlogged conditions. These levels continued to drop until 1992, twenty-one years after logging. Although flows have increased since then, the flow levels have remained slightly lower than pre-logging levels.
Flows within the North Fork watershed took longer to respond after clear-cutting. After 11 years, the proportional increase in flows reached a level similar to the maximum at the South Fork. The maximum mean increase at 11 years is equivalent to a 1.57% increase per percent of forest logged for the 50% clear-cut North Fork. This rate was 1.2 times the maximum 1.33% increase per percent of forest removed by 67% selection logging at the South Fork. North Fork flow levels dropped to pretreatment levels 19 years after clear-cutting. A graph displaying flow changes after South Fork selective loggin and North Fork clear-cutting was constructed.
The difference between selective logging flow levels and clear-cutting flow levels can be attributed to a difference in the distribution of trees that remain after logging. In second-growth redwood forests, clusters of trees share a common root system. When neighboring trees are selectively logged, the remaining trees have the root system in place to take advantage of soil moisture that is no longer utilized by the logged trees. Dry-season flow dropped quickly as the remaining trees used up excess moisture, and continued to drop as the newly established young trees grow larger. However, on a clear-cut slope, the nearest trees are off site. As a result, a large amount of regrowth must occur onsite before the excess soil moisture can be fully used. Thus, dry-season flow remains elevated longer than in selectively logged watershed.
The authors’ findings also reveal that the 22% change in annual rainfall produced the largest effect on flow. A reduction in rainfall would lead to a 23% decrease in the 10th percentile September 1 flow under unlogged conditions. Twenty-one years after selective logging, the 10thpercentile flow declined to 41% of unlogged levels, compared to 54% under the present rainfall regime. For unlogged conditions under the current climatic regime, a September 1 flow <0.55 L/km2-s could occur once in five years. Yet, lower flows would be expected twice as often during the 36-year post-logging period. A decrease in average flow for the post-logging period would be similar to that expected form a 10% decrease in mean annual rainfall under unlogged conditions.
Recent rains affect APIs more strongly than earlier ones. So, a shift in seasonal rainfall distribution could influence dry-season flows even if the overall annual rainfall levels remained constant. A 44% reduction of spring rainfall (and corresponding increases of rainfall in other months) would reduce the 10th percentile flow at 21-years after selection logging to 47% of forested levels. Additionally, a 2.5% increase in annual rainfall, if it only occurred in May, would affect September 1 flows by 10%.  Similar results were obtained when rainfall was increased by 46% and 83% in March and January. Overall, rainfall occurring after February had more influence on dry-season flow than rainfall occurring earlier in the wet season.
A 77% reduction of June and July rainfall (with no change in annual rainfall), produced one half the flow reduction as that caused by the 44% decrease in April–May rain. However, the results also suggest that summer rainfall may have a smaller effect on dry-season flows than the API models predicted. After a seasonal soil moisture deficit accumulates, a higher proportion of rainfall may be stored in the soil and transpired before it contributes to runoff.
 Modeling the same changes in rainfall regimes in combination with clear-cut logging can only be assed for the initial 19-year period of flow increase at the North Fork. Seasonal redistribution of rainfall affected the post-logging response even without a change in annual rainfall. The influence of rainfall late in the wet season was more pronounced for the North Fork than for the South Fork. Graphs displaying the modeled responses of September 1st weir flows to logging and hypothetical rainfall changes were constructed.
Overall, these results indicate that forest managers should design watershed-scale silvicultural strategies that could reduce the risk of unfavorable dry-season flow alterations. 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s