By Ken Winterberger and Jim Labau
Between the two of us, we have spent over 50 years being involved with the forest inventory in Alaska. In 1960, Jim joined the Alaska inventory program as a freshly minted forester from Colorado State, and his first job was on the Chugach National Forest inventory. Ken arrived in Alaska for a summer inventory job in 1975. Thoroughly hooked by the state, he spent the rest of his career inventorying the state of Alaska.
Inventorying a state that is 663,267 square miles presented a number of logistical and statistical challenges, and here we provide an overview of what we worked through.
The early years of inventory
Since its formation, the US Geological Survey has gathered forest data and mapped forests in the US, including Alaska, as early as the late 1880s and early 1900s in preparation for the set-aside of the first Forest Reserves. The earliest Forest Reserves in the Pacific Northwest were aimed at timber and watershed management. Alaska’s Afognak Forest and Fish Culture Reserve became the first Forest Reserve in America and was established by executive proclamation on December 24, 1892. In A History of the Forest Survey in the United States: 1830–2004, President Benjamin Harrison is said to have stressed the importance of the island’s forests, vegetation, and as a site for fish hatcheries in establishing the Forest Reserve. This multi-resource view, especially the obvious importance of fish habitat, was an indication of how inventory and monitoring of forestland would take shape in Alaska.
The first forest inventory conducted by the federal Forest Survey program was in the southern part of Southeast Alaska using a two-phase sample design. Phase 1 involved the photo interpretation of existing aerial photography, much of which was acquired by the US Navy for early Alaskan topographic mapping efforts. The photography was black and white, and at a very small scale to cover large areas with one frame. Some of the photographs were even in an unusual format, e.g., Trimetrogon photos (See https://tinyurl.com/y6aademp). None of these aerial photo characteristics made the job of interpreting forest characteristics easy.
The ground plots established in Phase 2 covered an area of about one acre. Each ground plot consisted of a series of three nested rectangular subplots, with each subplot being a 1/5 acre (2 chains x 1 chain) spaced two chains apart and placed up and down slopes within strata delineated on the aerial photograph. All sawtimber was measured on each of the 1/5-acre plots. Log-grades were determined for each sawtimber tree on the first half chain of each 1/5-acre plot (1/20 acre). Pole-size trees were measured on the right-half of the log-grade plot (1/40 acre) and sapling-size trees were measured in a 33 by 6.6-foot-wide strip to the right of the centerline of the log-grade plot (1/200 acre). The results of this initial inventory were used to support the planned establishment of Ketchikan Pulp Company (KPC) mills in Ketchikan and Wrangell. Once the inventory was concluded in the southern region, it moved north along the coast to Yakutat.
Concurrently with these inventory efforts, the Forest Survey moved into the Chugach National Forest. The forest was divided into three units: Prince William Sound, Afognak Island, and the Kenai Peninsula. The same twophase sampling method and groundplot design used in Southeast Alaska was repeated on the Chugach National Forest. The Chugach inventory of 199 ground-plots was completed in 1963. The Forest Survey’s focus now shifted north to interior Alaska. This time, the sample design broke interior Alaska into 10 units and acquired 1/5000 black and white infrared photography in strips 25 miles apart over each of the ten units. The photo-strips were flown perpendicular to the main drainage.
During Phase 1, a plot was delineated at the center of each photo and the vegetation on the plot was interpreted. Any plot determined to contain “commercial forest” would have tree height and crown diameters measured photogrammetrically. A systematic subset of these plots was “air-checked” in Phase 2, and finally in Phase 3, a subsample was visited on the ground. The ground sample was a rectangular 1 ⁄2-acre plot 330-feet long and 66-feet wide. Sawtimber was measured over the entire plot and pole timber was measured on the right side of the plot. Seedlings and saplings were measured on 10 equally spaced 4.16-foot radius plots (1/800 acre each). Mortality was also estimated. In all, 100 ground samples were established on commercial forest in Phase 3 of the original inventory of interior Alaska.
The reason for these sampling procedures was to use the tree height and crown diameter measurements from Phase 1 to estimate tree-stand volumes. However, there were serious problems with developing statistical estimators for this three-phase design. Ultimately the final estimates for the original inventory fell back to a twophase design with only Phase 1 and Phase 3 measurements being used.
Changing inventory methods yet again
During the inventory of interior Alaska, testing of a new sampling method, “variable plot sampling” or “prism cruising,” was underway. In fact, some testing of the variable plot sampling methodology took place on the 1 ⁄2-acre original inventory plots with the expectation of moving to the new ground sample design in the future.
This transition to a sampling methodology that hinges on the probability of a tree being sampled based on its basal area is Probability Proportional to Size (PPS). The interpretation and stratification of what was deemed commercial forest, which required an aircheck and ground samples, is an indication of what was important in the sample. All early sampling was aimed at ascertaining the existence and quantity of a commodity, essentially timber and forest products in general.
As the primary investigator of the new variable radius plot sampling methodology, Jim helped establish this new methodology but was then given the task of establishing and coordinating the new variable radius plot sampling design in Southeast Alaska based on his 1967 Master’s thesis “An analysis of design and technique of cluster point sampling in coastal Alaska old-growth forests.” By the mid-1960s, a new ground plot sampling methodology using the PPS sampling strategy was deployed in both interior and coastal Alaska.
Rather than fixed rectangular or circular plots, the new PPS sample design called for tree samples being selected on points established on a triangular network placed on an area of about one acre. This regular network was designed so the grid could be modified, and the points could be systematically moved to keep the plot within a defined forest type.
By the mid-1970s, more non-tree information began to be collected, especially detailed data on understory vegetation. The Forest Survey experimented with new names to describe their interest in collecting data on more than just timber. For a short time, the inventory in Alaska was referred to as the Alaska Renewable Resource Evaluation Program.
The 1980s were a time of experimentation with several sample designs. The inventory program long realized that interior Alaska would always be difficult to inventory because the vast landscape had very limited access. While much of the landscape is forested, most of this forestland is considered scrubland. One proposed inventory method was a four-phase system.
• Phase 1 would be a wall-to-wall Landsat image classification.
• Phase 2 was a photo interpretation of sample locations on 1:60,000 high altitude color-infrared photography at a 10 km interval.
• Phase 3 consisted of a photo interpretation, vegetation mapping, and photogrammetric measurement of vegetation on 1:5,000 low-altitude photos acquired at a 20 km interval.
• Phase 4 would be an 18-point triangular network of points spread over eight hectares measured on the ground at a 40-km interval.
This sampling system was used on the Tanana River Basin, the Copper River Basin and in Southeast Alaska. However, the sampling method had many challenges. The Landsat sensor had a 60 m resolution. The large-scale photography and ground sampling was done prior to the availability of GPS and plot colocation was difficult. It was also very difficult to acquire complete cloud-free sample coverage for Phase 3.
In the early-1990s, the Forest Survey was renamed to the Forest Inventory and Analysis (FIA) program and began moving to a standard sample grid still in use. For most of the country, one ground sample is allocated to every 2,400 hectares (6,000 acres). Each of the locations is interpreted to be forest or non-forest using remote sensing techniques, a ground sample consisting of four circular fixed plots is established. Each year a 1/10 of this sample is remeasured, with all plots remeasured every 10 years. A full-intensity sample grid (1 plot per 2400 hectares) has been established for Southeast Alaska. To implement the 10-year cycle of inventory over all of southeast and southcentral coastal Alaska, FIA used a research vessel named the Maritime Maid that served as field-crew quarters, office, and helicopter base.
While FIA established the grid inventory in coastal Alaska, interior Alaska was being neglected. A former Alaska State Forester even suggested that FIA actually meant “Forget Interior Alaska.” In 2013, the FIA finally decided they needed to implement a variation of the national grid system in interior Alaska. The one plot per 2,400-hectare design would simply not work so the FIA decided to dramatically de-intensify the grid and divide Alaska into “lower-48” statesized units. A less intense grid of plots would be established in each of these units. Rather than the annual inventory methodology used in the contiguous United States, a modified periodic inventory approach would be used. To maintain sample precision while dramatically reducing ground sample size a collaboration between FIA and NASA began.
In 2014, NASA began collecting GLiHT (Goddard’s LiDAR, Hyperspectral and Thermal Imager) data in strips over the interior Alaska units (similar to the units created for the original Interior Alaska inventory). The G-LiHT data has been modified to include high-resolution digital image data. A 1/5 ground-plot sample intensity is established within the G-LiHT sample strips creating a ground sample intensity of one ground-plot per 12,000 hectares (30,000 acres). The groundplot measurement is being accomplished with much of the work being done by Alaska Division of Forestry field crews.
Several goals are hoped to be accomplished with this shared responsibility: the state will have “ownership” in the plots, there might be more of an opportunity for local hire, and there is a better chance of getting buy-in from local residents. The new ground plots are also designed to measure the entire forest, whether it might be considered commercial or not. The ultimate goal is to measure forest condition changes over time. The Alaska State Forester said that it looked like FIA actually meant “Finally Interior Alaska.”
Even with this ground-plot intensity reduction and use of G-LiHT, the vastness and remoteness of Alaska makes the inventory process difficult and expensive. There is currently a move afoot to undertake yet another redesign of the inventory methodology being used in coastal and interior Alaska, perhaps combining the two methodologies, annual and periodic, into one Alaska-wide system.
Over time, interest has continued to shift from an almost purely commodity-based inventory to a more holistic measurement of Alaska’s forest condition as a carbon source or sink, a potential indicator of climate change, shifts in forest vegetation composition, forest pest damage, forest as wildlife habitat, and latitudinal and elevational forest migration. The key continues to be the standardization of ground plot measurements so that the measurements can be compared over time, because continuity of actual measurement is vital.
Ken Winterberger was a forester who worked with the Juneau and Anchorage Forestry Sciences Laboratory inventory program from 1975 until his retirement in 2015. An SAF member, he can be reached at firstname.lastname@example.org. Jim Labau, an SAF member, came to work for the Juneau Forestry Sciences Laboratory inventory program in 1960 and later served as the Anchorage Forestry Sciences Laboratory Project Leader until 1990. He can be reached at email@example.com.