Shallow Landslide Erosion Rates on Industrially Managed Timberlands: Key Factors Affecting Historical and Contemporary Rates

Timberland management can negatively impact the landscape and can lead to increased shallow landslide incidences and erosion rates. The correlation between increased landsliding and timber harvesting is well documented in the literature (Croft and Adams, 1950; Bishop and Stevens, 1964; Swanston, 1974; Sidle, 1992; and Cafferata and Spittler, 1998). These studies were based on reviews of historical management practices and methods, many of which were generally unregulated. Although forestry boards and regulations have been in existence in California since 1885, in general, a lack of enforcement prevented any meaningful environmental protections (Lundmark, 1975). Forest management has changed substantially over time, so it is appropriate to look at historical and long-term trends compared with modern-day practices and corresponding modern-era erosion rates.

Particularly in California, management practices have changed dramatically over the last four decades and probably most significantly over the last two. Regulations have changed, harvest methods have changed, and geologic hazard awareness and oversite have become commonplace. Managing industrial timberlands is no longer as simple as cutting trees for money; it has become more about responsibly managing a forest for multiple resource values.

This study was based on long-term monitoring projects associated with an aquatic habitat conservation plan (AHCP). The development of the AHCP (Green Diamond Resource Co., 2006) that is specific to this study was a collaboration between the private landowner, Green Diamond Resource Co., and federal regulatory agencies (National Marine Fisheries Service and U.S. Fish and Wildlife Service) and included a consistency determination with the California Department of Fish and Wildlife. The data was collected as part of a mass wasting assessment embedded within the AHCP. Shallow-seated landslides were the focus of data collection between 2008 and 2016, covering over 121,500 hectares (300,000 acres) in northern coastal California. Deep-seated landslides were not evaluated for this study. Randomly selected survey locations included hillside areas adjacent to more than 6.5 percent of the perennial flowing streams within the study area.

The study area is in a tectonically active area just north of the Mendocino Triple Junction (MTJ), where the North American, Gorda, and Pacific Plates collide. Seismogenic fault systems in the area are part of the MTJ and include the north end of the San Andreas Fault zone to the southwest, the Mendocino Fracture Zone to the southwest, and the southern end of the Cascadia subduction zone to the west, just off the coastline. As a result of the compressional forces exerted on the region due to the converging North American, Pacific, and Gorda Plates, there are numerous on-land upper-plate thrust faults throughout the region that are also considered potential sources for seismic shaking (Cao et al., 2003; Kelsey, 2001). They include but are not limited to Little Salmon fault, Mad River fault zone, Bald Mountain-Big Lagoon faults, and Grogan and Surpur Creek faults. The structural orientation of these upper-plate thrust faults is typically northwest-trending.

Earth materials vary throughout the study area because of the highly active tectonic regime described previously. At the southern extent of the study area, the bedrock is dominated by Miocene to Late Pleistocene deposits of the Wildcat Formation (Ogle, 1953). The Wildcat Formation is thought to be a coarsening-upward regressional sediment sequence deposited in the ancestral Eel River basin. To the north, the remainder of the property is dominated by deposits of the Coastal, Central, and Eastern Belts of the Franciscan Complex, which range in age from Pliocene to Early Jurassic (McLaughlin et al., 2000). Bedrock within the Franciscan Complex includes sedimentary, igneous, and metamorphic rock types; the most common earth materials encountered (from north to south) are sandstone and metasandstone, greenstone, mélange, and schist. A simplified illustration of the distribution of these materials, modified from Jennings et al., 2010, is shown in Figure 2. These units are somewhat specific to watersheds within the study area and are typically characterized by: (a) broken to sheared moderately indurated sandstone and metasandstone (largely in the central portion of the study area), (b) highly sheared siltstones and mudstones in an argillaceous matrix (mostly found in the central portion of the study area), (c) quartz-mica schist (primarily found in the eastern portion of the study area), and (d) moderate- to well-indurated fractured graywacke (mainly found in the northern portion of the study area). Throughout the study area, bedrock is found to be capped by Pleistocene to Holocene alluvial sediments or marine terrace deposits (Irwin, 1997).

Geomorphology varies across the study area and is characterized as more subdued in the south, becoming more rugged and incised to the north and inland. Landsliding is prevalent throughout, and types of landslides are typically associated with or attributed to the underlying bedrock. Debris slides and debris flows are the most dominant types of landslides seen across the study area. However, in the south where there are more low-gradient slopes and younger less-consolidated deposits of the Wildcat Group, an increase is seen in earth slides and translational landslides compared with other areas. Inner gorges (Kelsey, 1988) are prevalent in the central portion of the study area where examples are found of the steepest terrains that are commonplace in the Klamath River watershed.

The study methods discussed in this section are based on or modified from previous work and literature (Wieczorek, 1984; Keaton and DeGraff, 1996; Washington Forest Practice Board, 1997; and Brardinoni et al., 2003).

Historical aerial photographs were assessed by using SOKKISHA MS-27 (Sokkia Co., Ltd., Atsugi, Japan) and Abrams CB-1 stereoscopes (Abrams Instrument Co., Lansing, MI). Attributes were recorded of active landslides while reviewing aerial imagery and were mapped into Esri ArcMap based Geographic Information System (GIS) using light detection and ranging (LiDAR) bare earth 1-meter digital elevation model (DEM) as a base. Post processing of the raw LiDAR data and development of the DEM was done at Green Diamond Resource Co in 2008 & 2009. Age classifications were based on Keaton and DeGraff (1996). Most of the aerial photographs in the study's collection were at a scale of 1:12,000, with some as small as 1:38,000. Stereo-paired aerial imagery years included 1942, 1948, 1954, 1958, 1962, 1966, 1969, 1975, 1978, 1984, 1988, 1997, and 2001. Orthographically rectified aerial imagery was also reviewed and included the years 2005, 2006, 2009, 2010 2012, 2014, and 2016. In total, the study covered a 74-year period. The availability of orthographically rectified imagery is changing rapidly. In the past, aerial photo flights had to be contracted and purchased. Flights were typically flown every three years or so. More recently, however, orthographically rectified imagery has been available every one to two years, either through public access sources or through contracted flights. Although the photo resolution is not as good in some cases, landslide detection using orthographically rectified photos is still adequate, in part due to the greater frequency of photo sets available.

The earliest aerial photographs reviewed were flown in 1942 and, as available, at least one set from each decade thereafter was reviewed. Aerial photo coverage across the sample area was good. However, because of the aerial extent of the project and changes in ownership over the years, not all photo sets covered the entire study area. In most cases, there was at least one photo set that covered each watershed or group of watersheds for each decade. In several cases, aerial photographs were used from two different photo years to fully review an area for a particular decade; hence, the extensive list of aerial photographs used.

The data for the study was composed of information gathered as part of a mass wasting assessment. This process included surveying hillsides for shallow-seated landslides between 2008 and 2016. Surveys for landslides were completed on hillslopes adjacent to half-mile long perennial flowing Class I (fish bearing) and Class II (non-fish bearing) stream segments (Woodward et al., 2017), as well as Steep Streamside Slope (SSS) buffers that were retained for landslide prevention. Survey locations were randomly selected as described in Woodward et al. (2012) and they are shown in Figure 1. Following a review of aerial photographs, two- to three-person crews reviewed each of the survey locations. The general purpose of each of these field surveys was to confirm any landslides identified in aerial imagery and review the sample area in the field for any additional landslides that may have been missed through remote sensing. In all, hillslopes were surveyed adjacent to 298 km (185 mi) of stream segments (over six percent of the perennial stream network), as well as 37 hectares (92 acres) of SSS buffers (15 percent of the total SSS buffers). Primary data collected for each landslide included dimensions (length, width, and depth) for both the source area and the displaced landslide debris remaining on the slope, and topographic profiles, cross sections, activity levels, delivery estimates, average slope gradients, and distance to nearest watercourse. Landslide depths were estimated using information from scarp heights and field-developed topographic profiles of each landslide. Landslides were mapped in the field onto base maps generated from LiDAR with 1 meter or better resolution and later transferred into GIS. Global Positioning System (GPS) coordinates were also collected for the head and toe of each landslide. All landslides greater than 19 m² (200 ft²) in aerial extent were field reviewed as part of this work.

Analysis of the data focused on several key aspects of management-related mass wasting. Landslide visibility and detection were assessed by comparing the ability to identify landslides using both stereo-paired aerial photographs and orthographically rectified aerial photography. Erosion rates were compared with evolving state regulations and industrial timberland management practices, by decade. Additionally, causal mechanisms were reviewed for both contemporary and historical management practices. Each of these aspects is discussed in the next section.

Figure 3b highlights landslides that were observed on the ground but were not able to be detected with aerial photography. As landslides become larger and/or more recent, they are easier to see and then would be detectable on aerial imagery. Therefore, it is logical that most of the landslides seen on the ground were concentrated in the smaller size classes. If the smaller class sizes are compared in Figure 3a and b, only 4 percent in the 0–150-m² class of the landslides could be identified, which is only one-third of the detection rate seen overall. However, in the next class size up (150–300 m²), 11 percent of the landslides were detectable in that class, which is consistent with the overall rate of 12 percent. Frequently when reviewing watersheds, aerial photo sets were not available at more frequent time intervals than 10 years. Historical landslides, roughly 10 years old or more, were likely to have significant vegetation growth and were more difficult to discern. Although landslides were identified down to 20 m², landslides 150–300 m² in size appeared to be the smallest mappable unit that was still reliably detectable for the study area.

The smallest landslides, in the 0–150-m² class, dominated the population of landslides in the sample set as shown in Figure 3c, but only accounted for a small portion of the total volume. This area class represents nearly two-thirds (65 percent) of the landslides observed but accounted for only 11 percent of the total volume. Conversely, the larger landslides dominated the total volume of sediment, especially those in the greater than 1,650-m² size class, which accounted for 27 percent of the total volume (Figure 3c) and accounted for only 2% of the landslides observed. These data allowed a reevaluation of the minimum map unit of landslides necessary to be reviewed in future studies. Omitting these smaller landslides had a negligible impact on the overall data, including erosion rates. In doing so one could conduct a similar study with less than half the effort and without compromising the results.

Detecting landslides using aerial photography is an essential element of the work because they are used to establish decadal erosion rates. Figure 3a shows that eliminating the review of landslides less than 150 m² would reduce the number of landslides observed in the photo record by 21 percent (77 of the 371 landslides) yet have a negligible impact on overall cumulative volume as those landslides account for 1 percent of the total volume. That portion of the sample set is well distributed over time and therefore would not likely have a significant impact when estimating average annual rates. Ultimately, setting a minimum map unit size of 150 m² will allow better use of time and make work more efficient by significantly reducing the amount of fieldwork and costs involved while still producing a robust sample set to work with.

A total of 2,995 landslides were reviewed and measured in the field and of those, 371 were also identified during the review of historical aerial imagery. Active to dormant historic landslides were detected in each of the decades reviewed with aerial imagery. Although the landslides observed in aerial imagery accounted for only 12 percent of the number of landslides that were reviewed, they accounted for 49 percent of the total volume of sediment. As a result, those identified with historical aerial imagery provided an opportunity to look at both long-term and decadal erosion rates in coastal northern California.

Erosion rates are typically evaluated from available aerial imagery, which is often during a brief period for the specified study area. The imagery in this study spans a much longer period (74 years) than many previous studies (e.g., Cafferata and Spittler, 1998 [38 years]; Brardinoni et al., 2003 [30 years]; and Imaizumi et al., 2008 [38 years]), and covers key periods encompassing the broad evolution of timberland management practices and regulations. The onset of aerial photography begins at a time (1942 and 1948) when the study area is largely characterized by old- and second-growth timber with virtually no forest regulations, allowing a unique opportunity to evaluate erosion rates over both historical and modern times of industrial timberland management.

Long-term erosion rates for the study area are 145 m³/km²/yr and cover the entire period of aerial photo sets reviewed. On average it was found that delivery rates were 52 percent of the erosion rates (48 percent of landslide debris remained on the hillside). Historical logging era erosion rates were 60 percent greater at 243 m³/km²/yr and were defined as the period from the mid-1950s through the late 1990s. This period is characterized by the largely unregulated era of the 1950s and 1960s, combined with a transitioning period of the mid-1970s through the 1990s that included significant regulatory changes in the industry. As noted earlier, there was a significant rise in forest practice regulations in the mid-1970s and the regulations continued to evolve throughout the following decades. Geologic considerations quickly became part of the process beginning in the late 1970s when the California Department of Forestry (CDF), now the Department of Forestry and Fire Protection (CAL FIRE), contracted the California Division of Mines and Geology (CDMG), now the California Geologic Survey (CGS), to map the geology and landslides in several sensitive watersheds along the north coast of California (Bedrossian, 2015). The 1940s were excluded from this period as most of the study area was characterized as old-growth or mature second-growth forests at that time and closely represented the conditions of a mature or virtually unharvested forest. The modern era is characterized by key influences from the regulatory aspect as well as advances in technology that began around the year 2000. Erosion rates in the modern logging era (post-2000) have declined significantly to 20 m³/km²/yr and are down more than 90 percent compared with peak rates in the 1970s. To better understand these trends, it is essential to look at external factors that have affected erosion rates, examined in the Discussion section.

Determining causal mechanisms for historical landslides can be difficult. There are rarely firsthand accounts of the landslide failure and establishing the timing of and correlation with contributing factors is difficult. Relative timing can be established using differences in vegetation type and age. However, it is often difficult for an estimate to be more accurate than a couple of decades. Aerial photographs are a key component in this analysis as they can allow the capture of anthropogenic influences before the event and can be constrained between photo sets. Table 2 shows a comparison of causal mechanisms of the historic logging era to the modern logging era using the study's data set of landslides that have been verified in aerial photographs. The table groups landslide causal mechanisms into three categories; harvesting, road, and naturally occurring (natural). Landslides characterized as related to harvesting are those having occurred in a harvested area within 20 years of operations. Road-related landslides are characterized as those that offset or truncate all or a portion of a haul road or skid trail prism or were determined to have been directly influenced by road drainage. While there are significant differences in the impacts on slope stability in legacy and contemporary roads, these have been lumped into one category for simplicity as differentiating the two was not part of the scope of this work. Naturally occurring landslides are characterized as those that have no observable connection with anthropogenic influences such as roads or harvesting as defined above. In Table 2, a reversal is seen in causal mechanisms of shallow landsliding between the historical and modern logging eras. Within the study area, it was observed that anthropogenic influences of landslides and related erosion rates were reduced to 23 percent in the modern era, whereas they accounted for 88 percent of historical erosion. To date, there has been no landslide sediment volume attributable to harvesting, as defined above, in the modern logging era. While realizing that the periods are not equal, landslides are not occurring as frequently as they used to and management-related landsliding has declined.

Table 2.

Causal mechanisms attributed to landslides observed in aerial photographs. Results are expressed as a percentage of erosion volumes. Note that there was no determination between legacy road-related influences and modern road-related influences. As a result, road-related causes identified in the Modern Logging Era may be a result of either or potentially both.

View largeDownload slide

View Large

This study evaluated the decadal erosion rates in comparison with the evolution of forest practice rules and private management practices, as well as regional climatic and seismic influences. In doing so, a strong correlation was found between erosion rates and evolving forest management practices and regulations. Be it intentionally or inadvertently, both management practices and regulations have been affecting the most sensitive areas on the landscape regarding slope stability and are doing so in positive ways. Seismic and climatic influences also appear to have been factors influencing rates as well. During the period of this study, there was significant seismic activity and elevated precipitation events within the region. Additionally, the role of geologic oversight and general knowledge on harvest activities has changed over time and may also be influencing landslide rates.

Before 1973, the timber industry was virtually unregulated with no limits to the size of harvest areas, and there were no protection measures for streams or wildlife or for unstable or potentially unstable slopes. Changes in forestry were observed after the approval of the Z'berg-Nejedley Forest Practice Act of 1973. The Act, administered by the State Board of Forestry, came with a declaration that “the forest resources and timberlands of the state furnish high-quality timber, recreational opportunities, and aesthetic enjoyment while providing watershed protection and maintaining fisheries and wildlife” (California, 1974, Chapter 8, Article 1, Section 4512 (b)). In response to the Act, the California Forest Practice Rules were revised and were regionally specific to three Forest Districts. These more stringent rules included limits to harvest unit sizes, riparian protection that included tree retention along streamside slopes, and new road building standards, all of which have continued to evolve and have had significant impacts on management-related mass wasting (California, 2022).

Key periods of time in changing the state of California's forest regulations that have impacted mass wasting are listed below:

• 1970s – The passing of the Z'berg Nejedley Forest Practice Act of 1973 (California, 1974) drives significant changes to the California Forest Practice Rules (CA FPR). Through a Timber Harvest Plan (THP) process, fish-bearing streams were protected by 30-meter-wide (100 ft) tree-retention buffers and 15-meter-wide (50 ft) buffers on some non–fish-bearing streams. Prior to the Act there were no protection measures for streams. Additionally, harvest blocks were limited to 32 hectares (80 acres) in size in the coast district, although there were exemptions that allowed many harvest blocks to be up to 48 hectares (120 acres).

• 1980s – The first significant revisions to stream protection areas within the CA FPRs established the Watercourse and Lake Protection Zone (WLPZ) rules in 1983 (Martin, 1989). This defined specific criteria for identifying types of watercourses and associated WLPZs in the field and expanded the widths of the zones, up to 61 meters (200 ft) on Class I, fish-bearing, streams and up to 46 meters (150 feet) on Class II, perennial-flowing non fish-bearing, streams (CDF, 1985).

• 1990s – Revisions to the California Forest Practice Rules WLPZ and roads and landings rules in 1991 resulted in restrictions on the placement of fill material on steep slopes (CDF, 1992). Additional changes to the WLPZ rules at that same time, which included elevated canopy retention, increased the overall level of protection of streamside slopes. In 1994, changes to CA FPR silviculture rules (silviculture is the theory and practice of controlling the establishment, composition, and growth of forests) and sustained-yield plans (the yield of commercial wood that an area of commercial timberland can produce continuously at a given intensity of management consistent with required environmental protection and which is professionally planned to achieve over time a balance between growth and removal) resulted in smaller harvest blocks and reduced harvest rates (CDF, 1994). Even-aged management (the goal of attaining or maintaining one age class of a stand of timber as opposed to many age classes under uneven-aged management) was now limited to a maximum of 16 hectares (40 acres). The silviculture used was mandated to maximize sustained production, which for industrial timberland owners was based on a sustained yield plan.

• 2000s – Increased protection of streamside slopes was mandated through modifications to WLPZ rules. Integration of the Threatened and Impaired (T&I) Watershed rules into the CA FPRs in 2001 increased the width of Class I watercourse zones (CDF, 2001). In 2010, the CA FPRs added the Anadromous Salmonid Protection (ASP) rule package, resulting in greater protection of streamside slopes in terms of area and elevated levels of canopy retention (CAL FIRE, 2010. Note that due to state rebranding of the Department of Forestry and Fire Protection, CDF became known as CAL FIRE in 2008.)

In the 1980s, new standards for planning, building, and maintaining roads were implemented, which required landowners to size culverts for specifically sized storm events, required new road drainage and design methods, and required maintenance of roads after completion of logging operations (Martin, 1989). In addition, erosion control rules were implemented that addressed watercourse crossings by tractors, brought extra precautions for winter period logging, and provided specific requirements on water-break construction. Each standard was significant as the changes simply did not exist before the 1973 Forest Practice Act and the forest practice rules that were derived from it. Driven in part by the Forest Practice Act, section 208 of the federal Clean Water Act also played a key role in changes that came about in the latter half of the 1980s. In 1985, the chairmen of the State Water Board and Board of Forestry (BOF), the directors of the California Department of Forestry (CDF) and Department of Fish and Game (DFG), and the executive director of the California Forest Protective Association signed an agreement to assess forest practices. This agreement established a multidisciplinary team that conducted a one-year qualitative field assessment of the impacts on water quality resulting from contemporary timber operations (Martin, 1989). The team was comprised of resource specialists from DFG, CDF, the State Water Quality Control Board, and the forest products industry, and was known as the 208 Assessment Team. The team examined 100 completed state-issued Timber Harvest Plans throughout the state and the final report was completed in 1987 (Martin, 1989). Known as the 208 Report, this report spawned many changes to regulations that affected slope stability.

As a direct result of the 208 Report, new roads and landings rules and WLPZ rules were implemented in 1991 (CDF, 1992). New rules for roads and landings covered all aspects of construction with an emphasis placed on construction techniques and activities that would aid in the reduction of excessive soil displacement, the avoidance of unstable areas, an overall reduction of erosion, and the potential for sediment deposition in watercourses. That same year, the WLPZ rules were amended for the first time since 1983, also as a direct result of the 208 Report. Among those changes was the recognition of torrent salamander habitat, which increased the recognition of Class II streams and associated protection zones. In the mid-1990s, these new rules broadened the review team agency's regulatory role by adding specific protection measures and operational limitations to protect or enhance water temperature, filter strip properties, upslope stability, fish and wildlife values, and sustained-yield rules.

These continued changes have contributed to a further reduction in erosion rates over time. However, despite these changes to regulations, an increase in erosion rates was observed in the 1990s compared with the 1980s (Figure 5). This may be explained by two factors working in conjunction with each other: strong to major earthquakes followed by several years with substantial precipitation, all of which occurred in the 1990s; and, more significantly, nearly all of this happened before the 1997 photo set. This is discussed later in the section Seismic and Climatic Influences.

Regulations have continued to progress in recent times and are more protective than previously. In 2001, the implementation of the T&I Watershed rules required mapping of habitat for anadromous salmonids and thereby increased the amount of Class I watercourses that were identified and then protected, resulting in additional protection of streamside slopes in those areas (CDF, 2001). More recently, in 2010, the ASP rule package was implemented as part of the updated CA FPRs at that time (CAL FIRE, 2010). This brought forth the largest and most complex changes to WLPZs to date, especially on lower-order non–fish-bearing streams. At that time, the WLPZs saw increases in overall width as well as elevated levels of canopy retention. The goal of these regulatory changes was to address wildlife habitats. However, these WLPZs were also some of the most sensitive areas potentially impacting slope stability. Additionally, significant improvements in road management were seen that led to a reduction in road-related landslides. The most recent CA FPR road rule package (CAL FIRE, 2015. Developed in 2013 and implemented in the 2015 CA FPRs) highlighted road surface drainage improvements that helped prevent road-related landslides. In the modern era, culverts are sized for 100-year storms, including sediment and debris, and ditch-relief culvert spacing, sizing, and placement are improved to avoid triggering shallow landslide and road-edge failures. Although these specifications are enforced via the CA FPRs, the specific design requirements are attributed to the work of Cafferata et al., 2004. Improved road management and increased protection of the WLPZs have certainly played a key role in the reduction of observed erosion rates. Others have noticed this correlation as well. For example, Klein and Anderson (2012) noted similar effects to these regulatory changes elsewhere in the region by assessing total sediment load.

Along with the continuing changes to the California Forest Practice Rules, timberland management practices have also evolved and improved over time. Such changes have been noted throughout the redwood region (Valachovic and Standiford, 2017). Among those changes are modified riparian buffers, preventative mass wasting zones, road-management plans, and low-impact harvest methods. Factors impacting erosion rates that have been associated with management practices include voluntary habitat conservation plans (HCPs), development and implementation of preventative landslide buffers, innovative riparian management zones (RMZs) that protect aquatic habitat, low-impact ground-based yarding methods, and improved road management planning. (Note: riparian management zones or RMZs are streamside habitat retention areas located along rivers and streams. These areas are analogous to the WLPZ that was established as part of the CA FPRs.) The advancement of these management practices over time has aided in the decline of erosion rates and may have had their most dramatic effect in the modern logging era when many of these factors were developed and implemented (Figure 5).

Habitat conservation plans have been under development in the study region since the early 1990s. The Simpson Timber Company established the first HCP in the industry for northern spotted owls in 1992, which increased tree retention levels in Class I and Class II streams (Simpson Timber, 1992). In 1999, the Pacific Lumber Company, now known as Humboldt Redwood Company, established an HCP for their ownership that elevated retention in RMZs when compared to the CA FPR (Humboldt Redwood Company, 2019). Their HCP also addressed slope stability issues by establishing preventative protection measures for areas defined as Mass Wasting Areas of Concern.

In 2007, an Aquatic Habitat Conservation Plan (AHCP) was implemented across the study area which included numerous measures that have influenced the observed decline in erosion rates (Green Diamond Resource Co., 2006). Among the most effective were revisions to the RMZs mentioned earlier, seen in Figure 6f. The RMZs varied in width and were characterized by two zones of canopy retention—an inner zone of 85 percent and an outer zone of 70 percent overstory canopy closure—that were applied to slopes adjacent to perennial flowing streams. At the time of implementation of this AHCP in 2007, the RMZs resulted in an increase in tree retention in these streamside areas relative to the CA FPR WLPZ. The widths of the areas were generally the same; however, the canopy retention of the WLPZ was less. By comparison, the WLPZ required the retention of only 50 percent of the overstory and understory canopy cover on perennial streams at that time. Although generally the same, in some circumstances, depending on stream classification and yarding methods, these RMZs also provided a wider buffered area in comparison to the CA FPR WLPZ.

Private landowners also address road building and management. Poor road building and management have been known to be significant contributors to landslide initiation and sediment input associated with timber harvesting (Swanson and Dyrness, 1975; Amaranthus et al., 1985). As part of the Green Diamond AHCP (Green Diamond Resource Co., 2006), a comprehensive road management plan was implemented, a three-part plan intended to address all roads across the property by the end of the plan design. The first part is a timber harvest plan assessment that addresses all appurtenant roads within the plan area by upgrading roads that are going to be used and decommissioning unnecessary roads. The second is a road-maintenance program that reviews all truck- and ATV-accessible roads every six years for maintenance and upkeep. The third part is a watershed-by-watershed complete assessment of all roads with an inventory of sediment sources and determination of imminent risk of failure that is to be completed by the end of the plan design.

Historical records indicate that the region has shown elevated levels of seismic activity (Youd and Hoose, 1978; McPherson and Dengler, 1992; and Dengler et al., 1995) that have resulted in increased landsliding (Youd and Hoose, 1978; McPherson and Dengler, 1992). Research regarding seismically induced landsliding has shown that earthquakes can generate long-term landsliding and subsequent slide debris (Keefer, 1994). Keefer (1994) also notes that the smallest earthquake likely to generate landsliding is around a magnitude (M) of 4 and that these earthquakes generally produce only a few landslides. The effects of larger earthquakes occurring in the region, M 6 and greater, have been evaluated during the study period.

Climate records demonstrated that both annual precipitation and storm events (months with greater than 25 cm of rainfall) were greater in the 1990s than in most decades within this study; the bulk occurred during four years from 1995 to 1998. Seismic records also showed that the 1990s saw both more frequent and higher magnitude earthquakes than in any other decade in the study. With the increases in annual precipitation, storm events, and seismicity, an increase would be expected in erosion rates, which is seen in the 1990s compared with the 1980s and 2000s. Figures 9 and 10 both illustrate this correlation. After the 1990s, another sharp drop in erosion rates is noted in the modern logging era. While there was elevated precipitation in the 2000s (Figure 10), seismicity was significantly less when compared with the 1990s (Figure 9) and, when coupled with improving management practices as discussed earlier, it may be part of the reason a drop in erosion rates was seen over this period (Figure 10).

Geologic input associated with timber harvesting began in the mid-1970s with the passing of the Z'berg-Nejedley Forest Practice Act of 1973. In 1978, under provisions of Section 208 of the Federal Water Pollution Control Act and with funding from the Environmental Protection Agency (EPA), the California Department of Forestry hired several geologists under Title II Geologic Data Compilation Project to map the geology and landslides in several sensitive watersheds in northern California (Bedrossian, 2015). The goal was to better understand non-point sources of sediment pollution from landslides within prospective THPs. It also made geologic and geomorphic mapping available to foresters for THP layout as well as for reviewing agencies. However, a review of local plans by California Division of Mines and Geology, was limited until the 1990s. With the addition of the T&I rules into the 2001 CA FPRs (CDF, 2001), the California Geologic Survey's involvement with THPs grew. At that time, CGS staffing in Humboldt and Del Norte Counties went from one employee to five employees. Licensed geologists from CGS reviewed all submitted THPs and plans with complex geologic issues and typically received on-site field evaluations known as Pre-Harvest Inspections. As a result of the increase in state review, more foresters began to seek private consulting geologists to review THPs during the layout phase. THPs with complex geologic issues typically included a geologic evaluation from a licensed geologist. Some industrial timber companies have geologists on staff to review harvest plans including Weyerhaeuser, Green Diamond Resource Co., and Humboldt and Mendocino Redwood Co., to name a few. Geologists typically review in-house LiDAR and geologic mapping, as well as published geologic mapping. At Green Diamond Resource Co., most plans receive some level of field review and 20 percent, on average, receive input in the form of a modified geologic and geomorphic map or a geologic report that is submitted with the THP. Additionally, the level of knowledge of geology, and more specifically slope stability, for a forester is likely at an all-time high. Various associations provide geologic seminars for foresters and some industrial companies provide ongoing geologic training for their forestry staff. The California Licensed Foresters Association (CLFA) has a guideline that helps foresters determine the need for input from a geologist (CLFA, 1999). This guideline, coupled with training, can help foresters during harvest plan layout to identify potential hazards and seek appropriate professional input when needed. The level of awareness regarding slope stability has increased over time and has likely contributed to a reduction in erosion rates.

Minimum map units can have a significant impact on the level of effort required for a mass wasting assessment. According to observations in northern California, one could increase the efficiency of future landslide inventories by setting a minimum map unit of 150 m² (1,615 ft²). Using this size threshold, 89 percent of the total landslide sediment volume would be recorded from only 35 percent of the landslides surveyed, a 65 percent reduction in fieldwork. A case could also be made to reduce field efforts even further by evaluating a minimum map unit of 300 m² (3,230 ft²), which would reduce the field evaluation efforts by more than 80 percent and still capture 78 percent of the total landslide sediment volume. In either case, for efficiency or economics, a minimum map unit should be carefully considered and designed to capture a balance that will accurately characterize sediment volumes with a practical number of data points.

The time span of this study provides a rare and insightful look at the effects of timberland management practices in northern California. With mostly mature forests occupying the study area during the 1940s, pre-management and post-management looks can be captured at these watersheds. As management activities increased in the decades following the 1940s, a compounding rise was clearly seen in landslide-related erosion. The Z'berg-Nejedley Forest Practice Act passed while erosion rates were at their peak and although it took several years to implement, there is no mistaking the dramatic effects it had on reducing erosion rates which were seen by the end of the 1980s. The continuing downward trend in decadal erosion rates correlates strongly with the evolution of regulations and management practices, especially those related to roads and streams.

The ability to detect and record landslides is at an all-time high thanks to the improved quality and the increased frequency of remotely sensed data and imagery. Today it is easier to track landslide-related erosion than it was previously. This study of historical landslide erosion shows that rates in the modern logging era in northern California have declined by more than 90 percent since their peak in the 1970s. Technological advances have contributed to this change and have been key in reducing ground disturbance associated with modern-day operations. However, evolving government regulations have been the catalyst in making these changes occur beginning with the establishment of the Z'berg-Nejedley Forest Practice Act of 1973. This in turn has led to an evolution of management practices and for more landowners, that includes self-imposed regulation-like habitat-conservation plans and road-management plans, which may be the most significant factors associated with the improvements seen in the modern logging era. Observations show that conscientious landowners can and are conducting timber harvesting without significant adverse impacts on watershed resources. Once a destructive process, managing industrial timberlands has evolved to become the responsibility of managing a healthy functioning forest.

This work would not have been possible without the dedication of the Simpson family which has maintained ownership of timberlands on the west coast for over 130 years. Their commitment to the safety of their employees and the stewardship of their lands is unprecedented. I also thank the geologists involved in collecting and reviewing the immense amount of data presented: Scott Matheson, Scott Kirkman, Evan Saint-Pierre, Nick Graehl, Dan Hadley, Esther Stokes, William Troxler, Ronna Bowers, Lyman Petersen, Michael Tanner, David Perry, Jason Brooks, Brian McMullen, Kyle Terry, Matt Kowalski, Nick Hawthorne, Annie Fehrenbach, Brian Cook, and Ross Hiatt. Thank you all for your hard work and dedication to combing the often steep and brushy terrain of northern coastal California. Thanks to my friend and colleague Mr. John M. Curless, CEG, who helped review and provide comments for this work. Lastly, a special thanks for the wisdom and support of my friend and coworker Matthew R. House (Aquatic Biologist) who provided many hours of discussion on all aspects of this work, including this paper and editorial reviews.