Documenting 20th and 21st century glacier change and landscape evolution with maps and land, aerial, and space-based geospatial imagery in Alaska’s Kenai Mountains

Data fusion and analysis of maps and remote sensing data collected from different spatial perspectives (ground, air, and space) at different times from the early 20 century to the present using different sensors were used to answer questions about glacier behavior and rapidly changing landscapes of Alaska’s southern Kenai Mountains. Expeditions to three fiords of the southern Kenai Mountains were conducted during the summers of 2004, 2005, 2006, and 2021. Each expedition used repeat photography to document glacier behavior and change, and landscape evolution at six Kenai Mountains glaciers, most located within Kenai Fjords National Park. Bear Glacier, Aialik Glacier, Pedersen Glacier, Holgate Glacier, Little Holgate Glacier, and Northwestern Glacier were studied and at a minimum, their terminus positions were determined for the following dates: 1909, 1950, 1961, 1973, 1990, 2004-2006, and 2021. Each glacier displayed unique asynchronous behavior. Since 1909, all displayed long-term terminus retreat. However, the timing for each glacier was unique. In 2021, Holgate Glacier was advancing, while the other five glaciers were retreating.


Introduction Introduction Introduction Introduction
Study Location. The location of this investigation includes several of the fiords of Alaska's southern Kenai Mountains. The Kenai Mountains, with maximum elevations approaching 2,000 m, are a ~195-kmlong by ~45-km-wide mountain range which extends south-southwest from near Anchorage, Alaska, to the southern end of the Kenai Peninsula and the Gulf of Alaska. Two icefields, the Harding Icefield and the Grewingk-Yalik Glacier Complex, straddle the higher elevations of the Kenai Mountains. The icefields receive over 10 m of snow each year. According to Loso et al. (2014), the Kenai Fjords National Park area of the Kenai Mountains contains 287 glaciers. These glaciers cover approximately 48.5% of the park. They range from glaciers that cover less than 1 km 2 to the largest, Bear Glacier, with an area of nearly 200 km 2 (Molnia, 2008).
More than 35 glaciers, including all the glaciers studied in this investigation originate from the eastern Harding Icefield, which has an area of >2,500 km 2 . Sixteen of these have lengths greater than 8 km. The Harding Icefield is the Kenai Mountain's largest icefield and the largest icefield located entirely in Alaska. More than a dozen of these glaciers flow from the icefield and terminate in fiords that connect to the Gulf of Alaska, including all in this study ( Figure 1). One or more glaciers was selected from each of the three fiords for analysis. At a minimum, their terminus positions were determined for the following dates : 1909, 1950, 1961, 1973, 1990, 2004-2006, and 2021. For several glaciers, crude 1849 terminus positions were discernible from the earliest known map that depicts the fiords (Figure 2 -Tebenkov, 1852). Each fiord hosts other small valley and mountain glaciers, several of which are named (Figure 3). Each of these fiords either terminates on the Gulf of Alaska Continental Shelf (Figure 4) or traverses the shelf and joins other fiords or sea valleys that were cut into the shelf during older glacial expansions (Molnia, 2008).         In addition to the imagery focused analysis of the selected glaciers, where possible, additional observations were made to document changes in ice thickness, presence and distribution of vegetation, and development of ice-marginal hydrological features. The investigated fiords are Resurrection Bay, the location of Bear Glacier; Aialik Bay, the location of Aialik, Pedersen, and Holgate and Little Holgate Glaciers; Harris Bay/Northwestern Fiord, the location of Northwestern Glacier.
The research presented here is focused on how data fusion and the analysis of maps and remote sensing data collected at different times from the early 20 th century to the present using several distinct types of sensors and from different spatial perspectives (ground, air, and space) can be used to answer questions about recent glacier behavior and the rapidly changing landscapes of Alaska's southern Kenai Mountains.
To achieve the goal of documenting sub-decadal-, to decadal-, to century-scale glacier change, and landscape evolution with maps, land-, aerial-, and space-based geospatial imagery, this paper (1) Briefly summarizes the history of landscape photography in Alaska; (2) Documents how photographs and satellite images can be used to construct an extensive baseline upon which a 'landscape change history' can be constructed for the Kenai Mountain fiords; and (3) Shows how fusion of photographs and imagery from multiple sources is useful in understanding the complex dynamics of landscapes, ecosystems, and glaciers in Alaska's Kenai Mountains, in particular, and Alaska, in general.
Literature Review. Literature Review. Literature Review. Literature Review. A detailed review of the scientific literature identified about a dozen articles that discussed various aspects of the glaciers of the greater Kenai Mountains area. Many deal with glacier mass balance and related changes in Kenai Mountain icefields and groups of glaciers. Several deal with identifying pre-20th century glacier behavior from analysis of glacier sediments or plant material deposited in glacier sediments. Much of this additional information is based on dendrochronology, analysis of vertical aerial mapping photography, space-based gravity measurements, and space-based laser altimetry. Some studies used remote sensing and Geographic Information Systems (GIS) to describe specific attributes of Kenai Mountain icefields or complexes of glaciers.
Only one study, Molnia et al. (2007) utilized repeat photography to document glacier change and landscape evolution at a number of Kenai Mountains glaciers, specifically glaciers located within or adjacent to Kenai Fjords National Park. This 2007 publication documents author Molnia's initial efforts to use repeat photography to qualitatively and quantitatively document and describe near-century-scale glacier change in the southern Kenai Mountains.
During the summers of 2004, 2005, and 2006, more than 40 sites were revisited where historical photographs were made in 1908 and 1909 by Grant and Higgins (1913) and in the early 20 th century (likely early 1920s) by unknown photographers. At each location, a new 'repeat photograph' was taken. Hundreds of additional photographs of the glaciers and associated landscapes of the fiords of the southern Kenai Mountains were also made. The fiords of the southern Kenai Mountains have glaciers that terminate in the ocean and sometimes calve icebergs (tidewater glaciers), glaciers that end on land (land-terminating glaciers), and glaciers that end in bodies of freshwater and sometimes calve icebergs (lake-terminating glaciers).
Of the other published studies, the earliest, by Rice (1987) examined changes in the Harding Icefield's areal extent and surface features and made planimetric measurements of all the icefield's glaciers, using 1950 and 1951, 1:63,360-scale, U.S. Geological Survey topographic maps. These are the same maps that serve as one data layer in this study. Rice compared them with 1984-85 aerial photography collected by the Alaska High Altitude Photography (AHAP) Program and found that the glacier covered area of the icefield had decreased by as much as 5%, with a net loss of ~123 km 2 of glacier during the intervening ~34 years. The greatest changes he observed were near sea level along the Gulf of Alaska coast and at the 300-to 600-m elevations on the northern and western sides of the Harding Icefield, the source area of all the glaciers studied. Many smaller glaciers located at elevations below 1,000 m had disappeared. Adalgeirsdóttir et al. (1998) obtained airborne surface elevation profiles of thirteen Harding Icefield glaciers and the upper accumulation area of the icefield in 1994 and 1996, using a technique developed by Echelmeyer et al. (1996). These profiles were compared with the same 1:63,360-scale topographic maps used by Rice, andthe 1950-1952 aerial photographs used by the USGS in their preparation and revision. They concluded that the Harding Icefield has been thinning and shrinking since the 1950s and estimated that it has lost about 34 km 3 of ice in the ~44-year period between 1950-1952 and 1994-1996. They calculated that this corresponds to an icefield-wide lowering of about 21 m, the equivalent of an average mass balance of -0.4 m/yr of water. They also concluded that the rate of change of surface elevation between 1994 and 1996 is significantly greater than the long-term average. Four of the thirteen glaciers that they profiled: Aialik, Bear, Holgate, and Northwestern Glaciers are included in this study.
Previously, Adalgeirsdóttir (1997), compared the location and aspect of Harding Icefield glaciers and found that glaciers on the icefield's south side thinned more than those on the north side. Comparing areaaveraged elevation change between tidewater-, lake-, and land-terminating glaciers, she found little difference. Tidewater glaciers thinned by an average of ~16 m, while the land-terminating glaciers thinned by ~17 m. She found no significant correlation between elevation change and glacier area, length, or surface slope. Wiles et al. (1999) and Barclay et al. (1999), performed dendrochronological studies at a number of Kenai Mountain tidewater and former tidewater glaciers. Their work involved both living trees, some more than 680-years-old and a 1,119-year tree-ring-width chronology derived from more than 100 logs, recovered from about a dozen glaciers in the western Prince William Sound area, ~ 100-220 km east of the Kenai Mountains. Each of the logs had been sheared or uprooted by a past glacier advance. Collectively, their work showed that glacier fluctuations during the Little Ice Age were strongly synchronous on decadal time scales at many glaciers. Studies at eight locations indicated that advances occurred during the late-12 th through 13 th centuries and from the middle-17 th to early-18 th centuries. Nine glaciers showed evidence of a late-19 th century advance. Hall et al. (2005) used Landsat imagery collected in 1973, 1986, and 2002 to explore terminus position changes of about 20 Harding Icefield and Grewingk-Yalik Glacier Complex glaciers, including Aialik, Bear, Holgate, Northwestern, and Pedersen Glaciers. Using GIS software, they calculated: the extent of terminus position advance and retreat; the areas for the two 'icefields' for the periods of 1973 to 1986 and 1986 to 2002; and the area for each glacier for the periods of 1973 to 1986 and 1986 to 2002. They note that their measurements were made at the part of the terminus that showed the greatest amount of change. With respect to area change, their results indicated that between 1986 and 2002 the area of the Harding Icefield decreased from 1,753 km 2 to 1,679 km 2 , a loss of ~78 km 2 . This is an area loss of 3.62%. For the Grewingk-Yalik Glacier Complex, between 1986 and 2002 the area decreased from 403 km 2 to 399 km 2 , a loss of ~4 km 2 . This is an area loss of ~1%. VonLooy et al. (2006) used remote sensing and digital elevation models (DEMs) to examine the accelerating thinning and contribution to sea level rise of Harding Icefield and Grewingk-Yalik Glacier Complex glaciers. They compared a 1950s USGS Digital Elevation Model (DEM) produced from the same USGS topographic maps used in several other studies and a 2000 Shuttle Radar Topographic Mission (SRTM) DEM with more recent DEMs produced from airborne Lidar profiles collected along glacier center-lines. Their results indicate that thinning rates from the mid-1990s to 1999 (-0.72 ± 0.13 m/yr) accelerated by a factor of 1.5 as compared with the 1950 to 1994-1996 period (-0.47 ± 0.01 m/yr) for the same 13 glaciers on the Harding Icefield that were investigated by Adalgeirsdóttir et al. (1998). Comparison of the USGS and SRTM DEMs indicate the Harding Icefield and Grewingk-Yalik Glacier Complex thinned an average of -0.61 ± 0.12 m/yr from . Between 1950, the volume of the 13 glaciers decreased by 72.1 ± 15.0 km 3 and the surface elevation decreased, with a thinning rate of 0.61 ± 0.12 m/yr. This is equivalent to an annual loss of 1.19 ± 0.24 km 3 /yr of melt water, approximately one percent of the estimated post-1950 mountain glacier contribution to sea level.
According to Arendt (2006), net balance is the total volumetric change of a glacier divided by the time interval between measurements. The average net balance rate is net balance divided by the average area of the glacier at the earlier and later times. Arendt computed net balance rates and average net balance rates for 10 Kenai Mountains glaciers including Aialik, Bear, and Holgate Glaciers. Rates were determined for two time-intervals: 1950 to 1994-1996 and 1994-1996 to 2001. For the 'early' interval, data were computed by comparing 1950s USGS topographic map data with airborne laser altimetry data collected May 28 or 29, 1994or May 19, 1996. For the 'recent' interval, data were computed by comparing airborne laser altimetry data collected May 28 or 29, 1994or May 19, 1996 with airborne laser altimetry data collected May 18, 2001.
Arendt found that the subset of land-terminating glaciers had an average change in net balance rate of ~0.070±0.30 m/yr, and 0.060±0.40 m/yr for all glaciers, indicating that there is no significant difference between the two groups. For the 'early' interval, Arendt found that Harding Icefield land-terminating glaciers had an average change rate of ~0.070±0.30 m/yr, and 0.060±0.40 m/yr for all glaciers, indicating that there is little difference between the two groups.
For Aialik Glacier, the Net Balance 'early' was 0.002±0.03 km 3 /yr water equivalent; the Net Balance 'recent' was -0.010±0.006 km 3 /yr water equivalent; the Average Net Balance Rate 'early' was 0.02±0.35 m/yr water equivalent; and the Average Net Balance Rate 'recent' was -0.11±0.07 m/yr water equivalent. For Bear Glacier, the Net Balance 'early' was -0.18±0.04 km 3 /yr water equivalent; the Net Balance 'recent' was -0.205±0.009 km 3 /yr water equivalent; the Average Net Balance Rate 'early' was -0.85±0.19 m/yr water equivalent; and the Average Net Balance Rate 'recent' was -1.02±0.04 m/yr water equivalent. For Holgate Glacier, the Net Balance 'early' was -0.021±0.011 km 3 /yr water equivalent; the Net Balance 'recent' was -0.007±0.002 km 3 /yr water equivalent; the Average Net Balance Rate 'early' was -0.31±0.16 m/yr water equivalent; and the Average Net Balance Rate 'recent' was -0.10±0.04 m/yr water equivalent. For McCarty Glacier, the Net Balance 'early' was 0.007±0.020 km 3 /yr water equivalent; the Net Balance 'recent' was 0.034±0.006 km 3 /yr water equivalent; the Average Net Balance Rate 'early' was 0.06±0.16 m/yr water equivalent; and the Average Net Balance Rate 'recent' was -0.29±0.05 m/yr water equivalent. Lanik et al. (2018), describes a suite of eight glacier management and monitoring activities that have been performed by the National Park Service (NPS) and affiliated researchers in Kenai Fjords National Park. Collectively, they include: (1) Mass balance; (2) Repeat photography; (3) Terminus mapping; (4) Surface elevation change; 5) Timelapse photography; (6) Thickness measuring; (7) Aerial extent measuring and; (8) Measuring glacier flow rates. Arendt et al. (2013) investigated the mass balance behavior of Gulf of Alaska glaciers using two complementary satellite sensors, one measuring changes in gravity (GRACE) and the other changes in glacier surface elevation (ICESat). GRACE (Gravity Recovery and Climate Experiment), a NASA experiment, maps Earth's gravity field by measuring the distance between two satellites, using GPS and a microwave ranging system. Observations of changes in gravity between data collections are related to changes in the mass of materials at the Earth's surface below. Causes of observed gravity variations include exchanges between ice sheets or glaciers and the ocean, such as accumulation of new snow or glacier ice melting (NASA, 2014).
ICESat (Ice, Cloud, and land Elevation Satellite) was the NASA Earth Observing System mission that measured glacier ice sheet mass balance, cloud and aerosol heights, as well as land topography and vegetation characteristics with space-based laser altimetry (NASA, 2017). A mass concentration (mascon) is a measure of the mass of glacier ice present in a predefined 3 o x 3 o grided cell of the Earth's surface.
The researchers collected, compared, and contrasted high-resolution GRACE mascon) data for 'Gulf of Alaska glaciers' with other in-situ glaciological, climate, and remote-sensing observations, including ICESat results, to compute a regional glacier mass balance, a measure of the volume of water present in a glacier or glacier area during the period of observation. They determined that for the 7-year period from December 2003 to December 2010, the mascon cells that included the Kenai Fiords glaciers lost 65 ± 11 Gt of ice per year. They concluded that summer mass balance was responsible for the maximum interannual variability. A Gt (gigaton) is equivalent to one cubic kilometer of water. Comparing GRACE and ICESat results, they found that in the ~6-year period between October/November 2003 and October 2009, GRACE data showed a mass balance loss of 61 ± 11 Gt/yr, while ICESat glacier elevation change data showed a loss of 65 ± 12 Gt/yr.
ICESat elevation changes show a strong elevation dependence, with 2-4 m/yr of thinning at low elevations tapering to near-zero changes at high elevations. Multiplying these elevation changes by the area distribution of 'Gulf of Alaska glaciers' they obtained an average of 0.79 ± 0.15 kg/m 2 /yr of water loss. Giffen et al. (2014) used Landsat Multispectral Scanner (MSS), Thematic Mapper (TM), and Enhanced Thematic Mapper Plus (ETM) imagery collected in 1973, 1986, and 2000 to examine all the glaciers in Kenai Fjords National Park. Image-processing software was used to create GIS shape files of glacier extent. They found that Pedersen, McCarty, and Dinglestadt Glaciers all retreated between 1951 and 2005, with little terminus change between 1986 and 2000. McCarty and Dinglestadt Glaciers are not being investigated in this study.
Recession rates were slightly higher for tidewater terminating glaciers (Aialik, Bear, Holgate, McCarty, and Northwestern Glaciers) compared to land and lake terminating glaciers (Chernof, Exit, Indian, Kachemak, Killey, Lowell, Nuka, Pedersen, Petrof, Skilak, Tustumena, and Yalik Glaciers) that flow to the west and north. Pedersen Glacier is the only glacier in this category included in this study.
Northwestern Glacier advanced from 2000 to 2005. Between 1951 and 2005, Bear Glacier became buoyant resulting in an increase in its rate of terminus retreat. Interior northward and westward flowing glaciers had a recession rate of ~29 m/yr, while coastal southward and eastward flowing glaciers averaged ~32 m/yr, a reduction of ~21 km 2 in total ice cover during the period from 1986 to 2000. Pelto (2017), summarized the behavior of Pedersen Glacier from 1951 to 2015. Citing Giffen et al. (2014) he stated that Pedersen Glacier slowly but steadily retreated 706 m between 1951 and 1986, averaging ~20 m/yr) and an additional 434 m (23 m/yr) from 1986 to 2005.
In 1994, part of the terminus was fronted by a small proglacial lake, with most of the terminus located on land. Eleven years later in 2005, the lake's length had grown to 1.1 km along its center axis.  Jakob et al. (2021), used CryoSat-2 interferometric-swath-processed data collected between 2010 to 2019 to generate new and independent mass balance estimates for the Gulf of Alaska region and High Mountain Asia. CryoSat-2 (European Space Agency, 2021) was designed and launched in 2010 to measure the thickness of polar sea ice and monitor changes in glaciers and ice sheets, especially those that blanket Greenland and Antarctica. Its mission is to provide a precise picture of how the Polar Regions are responding to changing climate. After breaking each area into multiple subregions, they extracted elevation-dependent thinning rates which revealed ongoing mass loss across the sub-regions. They also extracted monthly time series of elevation change, exploiting CryoSat's high temporal repeat capability to reveal seasonal and multiannual variation in rates of glaciers' thinning.
They found that between 2010 and 2019, the Gulf of Alaska region, which includes the Kenai Mountains study area, lost mass at a rate of 76.3 ± 5.7 Gt/yr (0.89 ± 0.07 m water equivalent per year), for a sea level contribution of 0.078 ± 0.008 mm/yr. Surprisingly, the High Mountain Asia region which is often referred to as the 'Third-Pole", only produced 37% of the Alaskan contribution, losing mass at a rate of 28.0 ± 3.0 Gt/yr (0.29 ± 0.03 m water equivalent per year). Both regions lost more than 4 % of their respective ice volume during the 9-year study period.

Methodology
In August 2021, Molnia led a new photographic expedition to the southern Kenai Mountains. The expedition revisited many of the sites that had been previously occupied and photographed between 2004 and 2006. The purpose of the 2021 expedition was to: 1) photographically document summer 2021 glacier positions and behavior; and 2) to rephotograph as many of the previous photo locations as possible to produce new 2021 photographs for comparison with three groups of older photos, so that glacier change could be examined on sub-decadal, decadal, and century scales. The 2021 expedition attempted to observe and photograph all the large glaciers in each fiord and more than 50 other smaller alpine and valley glaciers, and their associated landscapes from locations where each had been previously photographed at various known times in the past. Resulting 'pairs' or 'triplets' were then qualitatively and quantitatively analysed to determine what changes had occurred during the period between photographs, especially whether the targeted glaciers were thickening or thinning, growing, or shrinking, advancing, or retreating, and moving or stationary. Ancillary information including maps, satellite imagery, and aerial photographic data were also analysed so that the magnitude of the changes observed could also be determined. 2021 field activities consisted of: 1) finding the location from which a 'historical' baseline photograph had been made; 2) matching the field of view depicted in each photograph with its corresponding Kenai Mountains landscape; 3) rephotographing the original field of view from as close to the original site as possible; 4) recording the photo location's geographic coordinates to permit plotting of the location and to facilitate future revisits; and 5) photographing the GPS receiver's coordinates and the reproduced historical photograph to minimize future data confusion and to facilitate accurately identifying which collected data components correspond to each site when data processing begins weeks to months in the future. Cameras used also had internal GPS receivers and clocks to provide time and date, as well as latitude and longitude in the photograph's metadata.
Prior to the August 2021 expedition, Molnia prepared a photo book consisting of more than 150 baseline reference photographs. The book consisted of three groups of photographs. Chronologically, the three groups of photographs, described below, are: (Group 1) the early 20 th century 'historical' photographs; (Group 2) a group of photographs that are part of a 'Kenai Fjords National Park Special Photo Collection'; and (Group 3) photographs made by Molnia during a 2001 visit to the fiords, and during the 2004, 2005, and 2006 expeditions. Included in this group are more than 30 photographs taken from original Grant and Higgins photo locations to produce repeat photography pairs. The technique where a sequential photograph is made from a previously occupied photo location is termed 'repeat photography ' (Webb et al., 2010). Some of these resulting photo pairs, which are part of the National Snow and Ice Data Center (NSIDC) 'Glacier Photo Collection' (National Snow and Ice Data Center, 2015) can be downloaded from the NSIDC: https://nsidc.org/data/glacier_photo/search/.
Group 1, the 'historical' photographs consisting of approximately 50 photographs that were made more than 110 ten years ago in 1908 and 1909 by US Geological Survey geologists U.S. Grant III and D.F. Higgins. Many of these photographs are reproduced in USGS Bulletin 526 (Grant & Higgins, 1913). All can be downloaded from the USGS Photo Library site: https://library.usgs.gov/photo/#/. Four additional photographs, with little information about their provenance and little or no metadata, likely dating from the early 1920s are included in this first group. USGS Bulletin 526 also contains sketch maps of each fiord depicting 1908-1909 glacier positions and the locations from which the photographs were taken.
Group 2 includes more than 40 photographs taken more than 30 years ago in August 1990 by NPS Rangers Mike Tetreau and Bud Rice. These images are part of a larger 'Kenai Fjords National Park Special Photo Collection' that is housed both at the park and at the NSIDC. They can be downloaded from NSIDC at nsidc.org. A description of the collection is available at https://nsidc.org/the-drift/data-update/kenaifjords-national-park-special-photo-collection-added-to-nsidc-glacier-photo-collection/.
Group 3 includes about 100 photographs taken by Molnia between July 2001 and August 2006, during four the earlier visits (2001, 2004, 2005, and 2006) to the Kenai Mountains fiords. More than 30 of these photographs were taken in 2004, 2005, or 2006 from original Grant and Higgins photo locations in order to produce repeat photography pairs. The technique where a sequential photograph is made from a previously occupied photo location is termed 'repeat photography' (Webb et al., 2010). Some of these resulting photo pairs, which are part of the National Snow and Ice Data Center (NSIDC) 'Glacier Photo Collection' (National Snow and Ice Data Center, 2015) can be downloaded from the NSIDC: https://nsidc.org/data/glacier_photo/search/. Four additional photographs, with little information about their provenance and little or no metadata, likely dating from the early 1920s are included in this first group.
Together, the three groups of photographs serve as the baseline data set from which to assess the annual, to decadal, to century -scale glacier change and landscape evolution within the three fiords. The previous revisits had resulted in the production of more than 25 repeat photography pairs spanning 96-98 years. Some of these locations were again revisited in 2021 producing new image triplets. During the field survey, about ½ of the photo deck's locations were successfully revisited and photographs taken of more than 60 sites of the previously pictured glaciers. Several examples of new repeat photo pairs or triplets are presented below for each of the fiords visited.
To augment interpretation and to provide a historical context documenting how the area has changed over time, sketch maps made by Grant andHiggins in 1908 and1909, oblique aerial photographs taken by USGS hydrologist Austin Post in 1961, Landsat satellite imagery collected between 1973 and 2021, and oblique aerial photographs taken by the first author between 2000 and 2018 were used to provide additional views and perspectives and to document glacier terminus positions and changes.
Historical Photography and Imagery Historical Photography and Imagery Historical Photography and Imagery Historical Photography and Imagery Considering Alaska's remote location, its early landscape photographic record is extensive. Historical photographs depicting the land surface of southern Alaska has existed for more than 150 years. The earliest known photographs date from the late 1860s, all postdating the 1867 purchase of Alaska from Russia. The earliest known systematic vertical aerial photographs of Alaska date from the late 1920s. For the southern Kenai Mountains, they date from 1950.
The earliest authenticated photographs of Alaska were taken by Eadweard Muybridge, the 'father of motion pictures' in August 1868 (Molnia, 2010). Although many of Muybridge's photographs depict southeast Alaskan landscapes that were shaped by glacier erosion, they do not depict any glaciers. In fact, it would take an additional 15 years before the first documented photograph of an Alaskan glacier was taken.
In 1879, eleven years after Muybridge photographed the landscapes of the Fort Wrangel area, John Muir led an expedition to southeast Alaska to observe modern glacier behavior. Muir's written description of his discoveries, especially the magnificent iceberg-calving glaciers of Glacier Bay quickly spread throughout the world, focusing significant attention on the glaciers of Alaska. Unfortunately, a camera was not part of the field equipment Muir carried. Instead, Muir made sketches of the glaciers that he observed Four years later, in 1883, a US Army expedition led by Lt. Frederick Schwatka (Schwatka, 1885) explored areas adjacent to Glacier Bay and produced the first published photograph depicting a glacier, 'Saussure' Glacier, a small retreating valley glacier, adjacent to what is now the Chilkoot Trail ( Figure 5).
Also in 1883, the first photographs depicting large Alaskan tidewater glaciers and glacier-covered landscapes were taken in Glacier Bay by amateur and commercial photographers. In July 1883, Captain James Carroll piloted the steamship Idaho Idaho Idaho Idaho to within ~125 m of the terminus of Muir Glacier (Catton, 1995).
Passengers, including journalist Eliza Skidmore (Skidmore, 1884(Skidmore, & 1886 obtained photographs of glaciers within the bay. Her photographs and written descriptions, combined with Muir's writings, stimulated two decades of tourism resulting in hundreds of photographs depicting Glacier Bay's massive glaciers ( Figure 6). These early photographs provided the public, and policy and decision makers with the first visual confirmation of why the glacier-covered areas of southern Alaska were unique areas worthy of scientific study, protection, and preservation. This was certainly a factor in establishing the Glacier Bay National Monument in 1925, and the Kenai Fiords National Park, 55 years later.  Although it is labelled "Finger of Saussure Glaciers," this name was never adopted. The glacier, which is underfit in its valley, appears to have a retreating terminus. It is probably an unnamed glacier on the east side of Mount Hoffman, Coast Mountains. Lithograph from Schwatka (1885). Photograph by Charles A. Homan, U.S. Engineers. In 1909, every large glacier present in the fiords of the southern Kenai Mountains was photographed ( Figure 7) within a 20-day period by U.S. Grant and D.F. Higgins. They were engaged in a study of the ore deposits and the general geology of Prince William Sound and the southern part of Kenai Peninsula for the U.S. Geological Survey (Grant & Higgins, 1913). The previous year, 1908, Higgins took four photographs of cirque glaciers located in the Thumb Cove area of Resurrection Bay, ~12 km south of Seward. In all, 48 1908 and 1909 Grant and Higgins photographs of southern Kenai Mountain's glaciers digitized from the archive at the USGS Photo Library, located in Denver, Colorado (https://library.usgs.gov/photo/#/) were used in this study.  Grant and Higgins comment (1913), that in the course of their work all the tidewater glaciers and many other glaciers located near tidewater "were seen and some notes, photographs, and maps were made.". They continue that "it is thought worthwhile to put on record the information thus obtained regarding the glaciers, for it will afford a basis for future study of the fluctuations of these ice streams" (pg. 1). They add that it "is not expected to make many additions to the large amount of scientific material concerning the problems of glaciers and glaciation that is already available, but it is intended to supply some definite information regarding the present positions of the fronts of the glaciers and the more evident facts of their fluctuations. Moreover, it is hoped that this publication may attract attention to some of the most magnificent American scenery that is now accessible to the tourist and nature lover." It is these ~50 photographs that were the focus of the Molnialed 2004Molnialed , 2005Molnialed , and 2006 expeditions. During the 2021 expedition, the locations of many of these photographs were visited and a new image of the field of view was collected to upgrade previously made pairs to triplets.
Systematic space-based imaging of Alaska's remote landscapes, including the Kenai Mountains, began in 1972 with the launch of the first Landsat satellite, the Earth Resources Technology Satellite 1 (ERTS1), later renamed Landsat 1. Since 1972, when Landsat began collecting data, Alaska has been imaged thousands of times by Landsat and other space-based multispectral (MSS) instruments, as well as other types of sensors, such as space-based radar. An October 12, 2021 search of the USGS EarthExplorer website (earthexplorer.usgs.gov) for all Landsat scenes that included the three fiords, regardless of percentage of cloud cover, identified >900 individual Landsat images.
To assist in the understanding and interpretation of the glacier change and landscape evolution that has occurred in the fiords of the southern Kenai Mountains, a total of four Landsat images of the southern Kenai Mountains, one from 1973, 1991, 2006, and 2021, were downloaded from the USGS EarthExplorer website (Figure 8). Landsat has a footprint larger than most multispectral sensors, ~ 185 km by ~ 185 km. A time series of sequential images, or a combination of Landsat images with images from other types of sources can provide a long-term synoptic look at areas ranging from entire icefields to individual glaciers. Repeat Photography Repeat Photography Repeat Photography Repeat Photography -Repeat photography is a technique in which a historical and a modern photograph, both having similar fields of view, are compared, and contrasted to determine their similarities and differences quantitatively and qualitatively. In precision repeat photography, both photographs have the identical field of view, ideally being photographed from the identical location. The use of repeat photography to document temporal change in glaciers and mountainous landscapes is not new. It originated as a glaciermonitoring technique in the European Alps more than a century ago.
Since 2000, Molnia has systematically used ground-based, precision repeat photography at more than 75 glaciers from about 225 locations to document glacier and landscape change in Alaska's Coast Mountains, St. Elias Mountains, Chugach Mountains, Kenai Mountains, Prince William Sound, and Copper River Basin on time scales that range from inter-annual to multiple-decadal. More than 150 ground-based image pairs have been produced (Figure 9).
Through analysis and interpretation of these photo pairs and time series, both quantitative and qualitative information is extracted to document glacier dynamics and landscape evolution, especially landscape response to retreating glacier ice on time scales that range from inter-annual to multiple-decadal on local and regional scales. Airborne-platform-based repeat photography is also being used to augment the ground-based assessments and to monitor change at geographic scales ranging from individual glaciers to entire mountain ranges. Combining modern satellite data and historical and modern photographic imagery with historical reports, maps, and traditional knowledge provides significant information about Alaskan glacier behavior and history. It also provides unequivocal visual documentation of how Alaskan glaciers are changing. Combining and 'fusing' data from different sensors often enhances the information that can be extracted.
In addition to the activities described above, Kenai Fjords National Park has established its own glacier photography monitoring activity. In 2012, Kenai Fjords National Park Physical Scientist, Deb Kurtz compiled a 'Kenai Fjords National Park Glacier Photographs Collection' (National Park Service, 2021) to support an ongoing repeat photography project documenting coastal glacier change in Kenai Fjords National Park. The collection which is supported with metadata, has two parts: 1) the Grant/Higgins-Molnia dataset of repeat photography pairs (1909 'historical' photographs and 2004-2005-2006 'recent' photographs); and (2) a catalog of 43 glacier photographs with metadata taken between August 13 and August 15, 1990, by Park natural resources staff members, Mike Tetreau and Bud Rice. Most of these photos were taken from boat-based photo point locations. The original slides from this dataset are archived in the Park's collections in Seward, Alaska. Digital copies are available from NSIDC. Details about the collection are available at: https://www.nps.gov/kefj/learn/nature/glacier-repeat-photography.htm. The website presents the following description: "There are many challenges in accurately replicating and aligning photos taken from boats due to variable positioning influenced by currents, winds, tides, and the height of the boat itself. Despite the challenges, the resulting photo sets effectively document the changing landscape. Several glaciers exhibited remarkable change in the recent past, which inspired an effort to annually photo-document the more accessible glaciers, and to repeat all photos every few years when possible. As of 2020, most of the park's glaciers continue to shrink, and the repeat photo collection continues to grow. The collection currently consists of 77 sets of photos, including 265 photos of 40 individual glaciers or glacier groups.".

Results and and and and Discussion Discussion Discussion Discussion
Although the focus of this study is documenting the behaviour of individual glaciers in the three fiords, the study also summarizes the findings of a number of papers that focused on regional glacier behavior in the Kenai Mountains, the Harding Icefield, Kenai Fjords National Park, adjacent areas, or some combination of these individual geographic features.

Fiord Specific Glacier Studies
This section examines the behaviour of one or more glaciers in each of the three fiords studied in 2021. This is done in two ways. First, using glacier-specific published scientific journal data and the several types of imagery obtained for each glacier (described below), a summary describing glacier behavior and changes in glacier geometry is derived. Second, individual photographs or repeat photography pairs or triplets are presented for each of the studied glaciers. Each pair or triplet is accompanied by a narrative describing the changes observed.
'Imagery' assessed by the authors and used for determining the behaviour for each glacier examined consists of: (1) 1909 ground-or boat-based photographs taken by Grant and Higgins, and obtained from the USGS Photo Library.
(2) 1909 sketch maps made by grant and Higgins, and published in USGS Bulletin 526.
(3) Several 1920s ground-based photographs of Kenai fiord's glaciers, each lacking information about either the photographer, the location from which they were collected, the date they were collected, or some combination of data components. (10) One-of-a-kind maps, such as a historic Russian 19 th century navigational atlas. With respect to the last item, Figure 2 is a piece of an 1849 Russian chart of Cook Inlet and parts of the adjacent Kenai Peninsula (Tebenkov, 1852), that depicts the positions of the termini of several glaciers (Holgate, Northwestern, McCarty, and Yalik Glaciers), located at the heads of Kenai Mountains fiords. It is the earliest known map that shows glacier termini locations.

Resurrection Bay
Resurrection Bay is an ~40-km long fiord that extends from the city of Seward and the delta of the Resurrection River at the northern head of the fiord to the Harding Gateway, the ~10-km long passage located at the mouth of the fiord. The maximum water depth in the fiord is ~293 m. Bear Glacier, the largest glacier in the fiord, occupies a side valley that joins the fiord from the northwest, ~20 km south of Seward ( Figure 10). Figure 10. Figure 10. Figure 10. Figure 10. Part of the NPS (2020) Map of Kenai Fjords National Park, depicting Resurrection Bay and Bear Glacier. The full NPS map, made from Landsat imagery is part of the online digital brochure, that is available at the NPS website: https://www.nps.gov/kefj/index.htm Bear Glacier When mapped by Grant and Higgins (1913) on July 20 and 21 July 1909, Bear Glacier was ~26 km long and it ended on an outwash plain-sand flat, a maximum of about 400 m from the shore of Resurrection Bay, with the central part of the terminus much closer to the Bay. Its terminus consisted of a small piedmont lobe that was as much as 4.6 km wide (Figure 11 -left). Grant and Higgins (1913) state that "Along the center of the ice front high tides reach the glacier". They depict the glacier's terminus region as having two significant medial moraines that divide the terminus region into three nearly equal segments (northern, central, and southern). The southern moraine is the largest, being about two to three times the width of the northern moraine.
Using dendrochronology, Viereck (1967), determined that a trimline located seaward of the 1909 terminus position formed between 1835 and 1845. Tree-ring-counts performed on trees located beyond the trimline determined that these trees were as much as 350 years old, dating from the earliest 16th century. No evidence of a more extensive earlier Little Ice Age (LIA) terminus position was found.  Glacier's length experienced 'extensive recession' but the actual amount or retreat was 'difficult to measure'. Arendt (2006), calculated that Bear Glacier's net balance for the ~45-year period 1950 to 1994-1996 was 0.18±0.04 km 3 /yr water equivalent, and the net balance for the ~6-year period from 1994-1996 to 2001 was -0.205±0.009 km 3 /yr water equivalent. The average net balance rate for the period 1950 to 1994-1996 was 0.85±0.19 m/yr water equivalent, and the average net balance rate for the ~6-year period 1994-1996 to 2001 was -1.02±0.04 m/yr water equivalent. Error bars on the measured changes are large due to major differences in snowfall amounts during the two measurement periods.
Much of this rapid retreat was due to disarticulation (Molnia, 2012). Due to its buoyancy, parts of the terminus, which had rapidly thinned, decreased in thickness to the point of floatation. Rapid retreat through passive calving became the dominant retreat mechanism. This is characterized by the loss of large numbers of tabular icebergs, some up to 1 km in maximum dimension.      Nearly every image from 2000 onward shows tabular icebergs that have calved from Bear Glacier's retreating terminus. This is due to continued thinning of the glacier resulting in changes in the way Bear Glacier retreats. Bear Glacier's piedmont lobe lies completely within the ablation zone. Initially, terminus retreat was dominated by melting. By the early 1950s, part of the glacier began to actively calve into the eastern ice-marginal lake. As this part of the glacier retreated into deeper water, the calving rate increased, partly due to the absence of a protective end moraine. Near the end of the 20 th century, thinning of the glacier margin accelerated and the thickness of the lower piedmont lobe thinned to a point where its buoyancy began to cause parts of the terminus to float, resulting in an increase in active calving. In the 2000-2010 decade, with continued thinning, more of the terminus floated and disarticulated, resulting in more than 4 km of terminus retreat.
Bear Glacier's piedmont lobe occupies a deeply scoured basin, often referred to as Bear Glacier Lake or Bear Glacier Lagoon. Immediately offshore of Bear Glacier, seafloor depths approach 200 m. With respect to ice thickness and depths in Bear Glacier Lake, Truffer (2014) describes a radar system that could be deployed as both a ski-team ground-based survey instrument and as an airborne survey instrument. The ground-based survey proved more reliable for narrow valleys, while the airborne survey allowed for large area coverage that worked particularly well over open ice. The maximum Bear Glacier ice thickness measured exceeded 650 m indicating that the glacier's bed is grounded well below sea level. Ice thickness on the Harding icefield was up to 450 m.
Maximum water depths in Bear Glacier Lake are unknown. However, a preliminary survey conducted in 2006 by the Molnia revealed a number of locations where the depth to the floor of the basin exceeded 75 m, the limit of the depth sounder used. Hence, as Bear Glacier's piedmont lobe thinned, its terminus area could quickly change from grounded to floating as its thickness decreased below the neutral buoyancy thickness, or as it retreated into more deeply eroded parts of its basin. Between 2002 and 2007, a period of intense disarticulation, part of the terminus retreated nearly 3 km. Several times, 0.6-1.0 km-size icebergs separated from Bear Glacier's terminus.     In other studies, Adalgeirsdóttir et al. (1998) found that between the 1950's and the middle1990's, Bear Glacier with a 1950's area of 228.5 km 2 , had its terminus retreat a maximum of 1.55 km and had its area decrease by 8.75 km 2 . Its volume decreased by 9.7 km 3 , and its average elevation decreased by 38.4 m. They also found that its mean annual mass balance was -0.7 m. For the same period, on an annual basis, Bear Glacier thinned by 0.872 m, had its volume decrease by 0.195 km 3 , and shortened by 36 m/yr. Giffen et al. (2014) found that between 1951 and 2005, Bear Glacier became buoyant resulting in an increase in its terminus retreat.
Summarizing all the available information about Bear Glacier shows that during the early 16 th century, Bear Glacier extended several hundred meters beyond its 1909 terminus position. A trimline located seaward of the 1909 terminus position formed between 1835 and 1845. No evidence of a more extensive early LIA terminus position was found. In 1909, Bear Glacier was ~26 km long and it ended on an outwash plain-sand flat, ~400 m from the shore of Resurrection Bay, with the central part of the terminus much closer to the Bay. Its terminus consisted of a small piedmont lobe that was a maximum of 4.6 km wide.
By 1950, the northeastern part of Bear Glacier's terminus retreated ~400 m from its 1909 position and a small ice-marginal lake developed along the northern part of the margin. Little change occurred on the southern third of the terminus during this 41-year period. In 1961, the lake was about the same size as in 1950. By 1961, Bear Glacier's southern -southwestern terminus area showed little change.
By 1984, the length of the ice-marginal lake had nearly doubled, growing to ~4.5 km. However, much of the southwestern terminus still remained in contact with the fluvial plain that it had continuously occupied since prior to 1909. Between 1973 and 1991, the north side of Bear Glacier's terminus retreated a maximum of ~4.6 km, while the central and southern parts showed little change.
Between 1990 and 2002, there was little change on the north side of the glacier while the central and southern parts of the terminus began to retreat rapidly, each losing up to a kilometer. By 2005, the central and southern parts of the terminus retreated a maximum of ~3.0 km. By 2007, maximum retreat was ~4.1 km. Through 2021, the terminus retreated up to an additional 2.5 km. In summary, from ~1900 through 2021, Bear Glacier retreated a maximum of ~6.75 km, with most of the retreat of the northeastern part of the glacier occurring before 1984 and the retreat of the central and southern part of the terminus occurring after 2000.

Aialik Bay
Aialik Bay (Figure 19) is an ~40 km long fiord with maximum water depths of ~192 m. It extends from the terminus of Aialik Glacier at its head to the Chiswell Islands, located at its mouth. Pedersen Glacier (spelled 'Pederson' on many early maps and charts) is in a side valley on the west side of the fiord ~7 km south of Aialik Glacier. Immediately adjacent to Pedersen Glacier, Aialik Bay shoals to water depths as shallow as 5-9 m. This is the location of the submarine end moraine that marks Aialik Glacier's LIA Maximum position. The east and west margins of the moraine are exposed at low water. Holgate Glacier is located at the head of Holgate Arm, an 8 km long fiord that enters Aialik Bay from the west, ~15 km south of Aialik Glacier. Maximum water depths in Holgate arm are ~157 m. The arm shoals at the southeastern end, likely delineating the LIA maximum extent of Holgate Glacier. Figure 19. Figure 19. Figure 19. Figure 19. Two Landsat images of Upper Aialik Bay depict the northeast part of the Harding Icefield with its major outlet glaciers -Aialik, Pedersen, and Holgate Glaciers. In the nearly 48 years between August 16, 1973 (Landsat 1) and August 4, 2021 (Landsat 8), Aialik Glacier has retreated less than 1 km, Pedersen Glacier has experienced a major recent rapid retreat, and Holgate Glacier experienced a small recent advance. At this scale, the behavior of Little Holgate Glacier is difficult to discern. Aialik Glacier When mapped by Grant and Higgins (1913) between July 22 and 24, 1909, Aialik Glacier was an ~11 km long tidewater glacier located at the head of Aialik Bay (Figure 20). They observed the glacier from the top of Squab Island, a glacially-sculpted bedrock knob located in the bay, ~2 km southeast of the glacier's terminus. They write that: "On each side of the glacier is a marked bare zone, and in the bare zone on the south side is a lateral moraine. When the ice extended over this bare zone, possibly 10 years ago, the front was about a quarter of a mile (~400 m) in advance of its present position". They also note that: "Much more advanced positions of Aialik Glacier, occupied several centuries ago, are indicated by shoals, caused by morainic accumulations, stretching across the head of Aialik Bay opposite and a mile (1.6 km) north of the front of the Pederson Glacier" (Grant and Higgins, 1913). Between 1909 and 2021, Aialik Glacier's terminus position fluctuated, but never more than ~0.75 km. Field, 1975, commented that "On the whole, this terminus has been remarkedly stable over the last six decades" (1909 to 1975).
The 2021 position of the southwestern edge of the glacier is ~500 m to 600 m behind the 1909 location. In 1909, Aialik was actively calving and showed a large area of exposed bedrock at sea level along its southwest margin.
As depicted on the Blying Sound D-8 USGS Topographic Map Quadrangle, Scale: 1:63,360 ( Figure 21) compiled from 1950 aerial photography (USGS, 1950b), the northern part of Aialik Glacier's terminus retreated ~400 m from its 1909 position. Over the next 112 years, the glaciers terminus position fluctuated with a net result of less than a kilometer of retreat and a thinning of less than 30 m.  Figure 22 presents three oblique aerial images of the terminus of Aialik Glacier. An Austin Post August 12, 1961 oblique aerial photograph of the glacier shows that calving has produced a large, semi-circular alcove that extends several hundred meters into the face of the glacier's terminus. Large 'arms' of ice extend beyond the embayment along both walls of the glacier's valley, with the arm on the southwest (left) side extending three or more times further than the arm of the northeast (right) margin. A 2007 photograph by Molnia shows that the northeast margin of the terminus has advanced several hundred meters, while the southwest arm has retreated. Much of the central part of the terminus is at about the same location. The embayment is much smaller and the face much straighter. Figure 22. Figure 22. Figure 22. A 2015 photograph by Bruce Molnia shows that the terminus has retreated up to 400 m and a wall of ice has been stranded that marks a former post-2007 terminus location. A large mass of glacial-fluvial sediment, part of which is vegetated, now sits up to a kilometer forward of the glacier's southwest margin. All three images depict Squab Island, the location of the photo point for the repeat photography triplet displayed in Figure 23. Aialik Glacier has a width at its face of 0.1.6 km, and an accumulation area ratio (AAR) of 0.88. Adalgeirsdóttir et al. (1998), report that between the 1950's and mid-1990's, Aialik Glacier, with a 1950 area of 118 km 2 , had its terminus advance 540 m, but experienced no change in area. Between 1950 and the mid-1990s, its ice volume decreased by 2.6 km 3 , its average elevation decreased by 11 m, and its mean annual mass balance was -0.2 m. On an annual basis, Aialik Glacier thinned by 0.25 m, had its volume decrease by 0.03 km 3 , and lengthened by ~13 m/yr. Hall et al. (2005), found that with respect to terminus position change, between 1973 and 1986, Aialik Glacier's length decreased ~85 m, an average of ~7 m/yr, and between 1986 and 2002, its length decreased ~339 m, an average of ~7 m/yr. In the 19 years since 2002, the central part of the terminus retreated as much as 300-400 m, while the southwestern margin has retreated more than 500 m, exposing a large triangular shaped fluvial outwash plain. Figure 23. Figure 23. Figure 23. Figure 23. Aialik Glacier Repeat Photography Triplet -These three photographs, spanning more than 112 years, were taken from the same location on top of Squab Island in Aialik Bay. The first was taken by U.S. Grant on July 23, 1909. The other two were taken by Bruce Molnia, one on July 13, 2004 and the other on August 5, 2021. Together, they document the continuing thinning and retreat of the terminus of Aialik Glacier. Note the thinning of the glacier, the retreating southwest (left) margin, and the increase in the amount of bedrock exposed at the base and center of the glacier's face. Arendt (2006), calculated that Aialik Glacier's net balance for the ~45-year period from 1950 to 1994-1996 was 0.002±0.03 km 3 /yr water equivalent, and the net balance for the ~6-year period from 1994-1996 to 2001 was -0.010±0.006 km 3 /yr water equivalent. The average net balance rate for the ~45-year period from 1950 to 1994-1996 was 0.02±0.35 m/yr water equivalent, and the average net balance rate for the ~6-year period from 1994-1996 to 2001 was -0.11±0.07 m/yr water equivalent. Error bars on the measured changes are large due to significant differences in snowfall amounts during the two measurement periods.  Summarizing all the available information about Aialik Glacier shows that sometime during the LIA, Aialik Glacier extended at least 8 km beyond its present position, filling the northern fifth of Aialik Bay and depositing an end or recessional moraine adjacent to Pedersen Glacier. In 1909, when first mapped, it was an ~11 km long tidewater glacier located at the head of Aialik Bay with a recently exposed bare zone at its base. A decade earlier, Aialik Glacier's terminus was ~400 m in advance of its 1909 position. Since 1909, Aialik Glacier's terminus position has fluctuated, but never more than a kilometer. By 1950, the northern part of Aialik Glacier's terminus had retreated ~400 m from its 1909 position. Between 1950 and the mid-1990's, Aialik Glacier's terminus advance 540 m, despite about 300 m of retreat that occurred between and 1964. Between 1973and 1986, Aialik Glacier retreated ~85 m. Since 2002, the central part of the terminus retreated as much as 300-400 m, while the southwestern margin has retreated more than 500 m, exposing a large triangular shaped fluvial outwash plain. The 2021 position of the southwestern edge of the glacier is ~500 m to 600 m behind the 1909 position.
Pedersen Glacier Pedersen Glacier is located ~7 km south of Aialik Glacier (Figure 20). Tebenkov's 1852 Atlas depicts the presence of a glacier terminus ~ 800 m from the shoreline of Aialik Bay at Pedersen Glacier's general location. Unfortunately, the portion of Tebenkov's Kenai Peninsula chart shown in Figure 2, does not extend far enough south to accurately depict the location of Pederson Glacier.
Pedersen Glacier was retreating prior to first being mapped on July 22-24, 1909 by Grant and Higgins (1913). Its 1909 terminus position (Figure 25) was from 400 to 500 m behind its most recent post-Little Ice Age maximum position. Even then, part of the terminus was reached by high tide. Grant and Higgins speculated that the glacier may have been at its maximum position as recently as the early 1890s. Right Right Right ----August 12, 1961 oblique aerial photograph of Pedersen Glacier taken by Austin Post. Note the small ice-marginal lakes and the parts of the terminus in contact with its fronting outwash plain. Grant and Higgins (1913) describe the northern part of the terminus being "a perpendicular cliff of ice perhaps 100 feet (30 m) high", with a "well-marked bare zone on each side of the front", suggesting a recent greater extent. They write: "Along much of the front, a quarter to a third of a mile (~240 m to ~310 m) from the ice, are remnants of a low moraine, which has now been nearly cut away by the waves. On this moraine are herbaceous plants and some alders about 2 feet high. The moraine was probably deposited at the time when the glacier advanced to the edge of the bare zone mentioned above. This advance may have been made 15 years ago and apparently was the maximum advance of the glacier since the advent of the present forest...".
By 1950 (Figure 25), the glacier retreated an additional 400 m to 1,200 m (Field, 1975;Molnia, 2008). Then, a 600 m to 800 m-wide tidal embayment fronted much of the glacier. By 1964, another ~250 m of retreat occurred. As much as 1.5 km of retreat and the development of a large ice-marginal lake occurred between 1965 and 2000. Field (1975) commented that from 1950 to 1964, ice velocities in the lower 2 km of the glacier were 'about 70 m/yr. At least six studies (Grant and Higgins, 1913;Field, 1975;Hall et al., 2005;Molnia, 2008;Giffen et al. 2014;and Pelto, 2017) have described the behaviour of Pedersen Glacier since it was first observed. Pelto (2017) reported that Pedersen Glacier retreated 706 m between 1951 and 1986, averaging ~20 m/yr. Hall et al. (2005) found that between 1973 and 1986, the glacier's length decreased ~511 m, an average of ~39 m/yr. This would mean that between 1951 and 1973, the terminus retreated ~195 m, at a much slower average retreat rate of 8.86 m/yr. Post's 1961 oblique aerial photograph (Figure 27), shows several small ice-marginal lakes developing along the glacier's southern margin as well as parts of the terminus still connected to its southern outwash plain. Hall et al. (2005), report that between 1986 and 2002, the glacier's length decreased ~108 m, an average of ~7 m/yr. Giffen et al. (2014) found that Pedersen Glacier retreated between 1951 and 2005, with little terminus change between 1986 and 2000. Pelto (2017) found that Pedersen Glacier retreated 434 m (23 m/yr) from 1986 to 2005. This would suggest that the retreat rate exponentially increased during the final few years of the interval. Pelto states that for the period from 1994 to 2015 the glacier retreated 2.6 km, an average rate of 125 m/yr. He further describes the growth of the ice-marginal lake that fronts Pedersen Glacier: "In 1994 part of the terminus was fronted by a small proglacial lake, with most of the terminus terminating on land. Eleven years later in 2005, the lake's length had grown to 1.1 km long on its center axis". He further states that "A comparison of the 2013, 2015 and 2016 terminus indicated that the recession was continuing rapidly" (Figure 26). To try to both clarify and simplify the details of Pedersen's behaviour, we measured the position of Pedersen's terminus on several types of 'imagery' data (maps, photographs, and satellite images) that are being used to describe and quantify the behavior of the Glacier. The results (Table 1 and Figure 27), show that Pedersen Glacier retreated 4.86 km during the 121-years from the maximum advanced position (~1900) identified by Grant and Higgins. Therefore, the average retreat rate was 40.17 m/yr. Rates fluctuated over time, with peak rates of > 350 m/yr between 2006 and 2010, and minimum rates of <20 m/yr for much of the pre-1980s period. Figure 28 presents a five-image repeat photography montage showing the rapid retreat and thinning of the glacier from offshore in Aialik Bay during the nearly 31-year period from August 13, 1990to April 5, 2021 Accurately determining the amount of terminus change is exceedingly difficult. Although many images are available, most were not useful due to image resolution, cloud cover, or shadows. There were also several long intervals where no data were available. Two other issues encountered were that the shape of the terminus is unique each time that it is imaged; and with each terminus, the rate and amount of change is different for various parts of the terminus. As a result, measuring change from one terminus position to the next is subjective. The most accurate method would be to draw the entire perimeter of every terminus on a single map, not just the most advanced point, to correctly determine and show the varying amounts of change.  Figure 27. Figure 27. Figure 27. Figure 27. Google Earth view of Pedersen Glacier with fourteen terminus positions plotted. Table 1 presents data describing the incremental retreat rates associated with each pair of positions.   Summarizing all the available information about Pedersen Glacier shows that in the mid-19 th century, the terminus of Pedersen Glacier was ~ 800 m from the shoreline of Aialik Bay. It may have advanced for the next half century and it may have been at its maximum position as recently as the early 1890s. In 1909, its terminus position was from 400 to 500 m behind that recent maximum position. Even then, part of the terminus was reached by high tide. By 1950, the glacier retreated an additional 400 m to 1,200 m, resulting in a 600 m to 800 m-wide tidal embayment fronting much of the glacier. By 1964, another ~250 m of retreat occurred. Between 1965 and 2000, as much as 1.5 km of retreat occurred with the development of a large icemarginal tidally-influenced lake. Specifically, the glacier retreated ~195 m between 1951 and 1973, ~511 m between 1973 and 1986, ~108 m between 1986 and 2002, ~326 m between 1986 and 2005, ~1,500 m between 2005 and 2015, and ~680 m between 2015 and 2021. In the 121-years from ~1900 to 2021, Pedersen Glacier retreated 4.86 km.
Holgate Glacier Holgate Glacier is a tidewater glacier located at the head of Holgate Arm, the westernmost branch of Aialik Bay. Davidson (1904) describes Holgate as nearly reaching the beach, suggesting that in the 19 th century, its terminus position was 10s to 100s of meters up valley from where it sat when photographed and mapped by Grant and Higgins in 1909. It was shown as being present on the 1849 Tebenkov map (Tebenkov, 1852 -Figure 2). Grant and Higgins' photographs of the terminus ( Figure 7A) show that "Holgate Glacier reaches tidewater in two streams separated by a small mass of rock, which not many years ago was a nunatak in this glacier. The northern and larger stream (Holgate Glacier) is discharging rapidly, but the discharge from the southern stream (Little Holgate Glacier) is small. Near the south side of the larger stream is a small medial moraine, but the glacier as a whole is free from medial moraines" (p. 59).
Grant and Higgins continue: "There are no trees on the sides of Holgate Arm within 1 mile (1.6 km) of its head, and beyond this the forest is sparse. There are no bushes and very few herbaceous plants close to sea level from the glacier to a point a quarter of a mile (~400 m) east" … "The rock mass between the two parts of the glacial front has bushes only on its upper half on the front and on its upper fourth on the sides. In very recent years, possibly within the twentieth century, the front of the Holgate Glacier was about a mile in advance of its position in 1909." Grant and Higgins (1913), 1909 sketch map ( Figure 30) and photograph ( Figure 7A) of upper Holgate Arm and Holgate Glacier show both glaciers terminating in Holgate Arm. The Little Holgate Glacier is the name commonly applied to the smaller south-eastern arm.
The 1950 Blying Sound D-8 topographic map shows that the south side of Holgate Glacier's terminus had retreated ~400 m from its 1909 position, while the north side remained unchanged (Field, 1975). Comparing the 1950 topographic map to the 1961 oblique aerial photograph shows that significant retreat of the southern part of the glacier continued, exposing bedrock along about a third of the base of the terminus. Field (1975, p. 519) quantifies this as "a further recession of around 500 m". Hall et al. (2005), reported that between 1973 and 1986, the terminus retreated ~234 m, an average of ~18 m/yr. From 1986 to 2002, "no change was detected". Arendt (2006), calculated that Holgate Glacier's net balance for the ~45-year period from 1950 to 1994-1996 was -0.021±0.011 km 3 /yr water equivalent, and the net balance for the ~6-year period from 1994-1996 to 2001 was -0.007±0.002 km 3 /yr water equivalent. The average net balance rate for the ~45-year period from 1950 to 1994-1996 was 0.31±0.16 m/yr water equivalent. This equals the loss of a nearly 14 m vertical column of water, or a thinning of nearly 16 m. The average net balance rate for the ~6-year period 1994-1996 to 2001 was -0.10±0.04 m/yr water equivalent. This equals the loss of a nearly 60 cm vertical column of water, or a thinning of nearly 65 cm. Repeat photography of Holgate Glacier from the north side of Holgate Arm (Figure 31), shows that from 2004 to 2016, although much of the terminus of Holgate Glacier was not visible, part of the south margin of the glacier could be seen. It decreased in size from 2004 to 2011 to 2016. Sometime after 2016, it began to advance. By 2021, it had readvanced to a position where it spanned the full width of the channel visible from the photo point. During that same period, the tributary that joined the terminus from the south, on the left side, melted away. (Figure 31).
The north side of Holgate Glacier's terminus extends onto the beach located at the foot of the bedrock knob at the head of its fiord. On August 5 th , 2021, the terrestrial terminus was advancing and developing a 2-4 m-high push moraine along its entire margin (Figure 32).
Repeat photography of Little Holgate Glacier's terminus from the north side of Holgate Arm ( Figure  31), shows that after 2004 and before 2016, when it was close to sea level, but about 100 m from the shoreline, the terminus of Little Holgate Glacier retreated completely out of the field of view. By 2021, it no longer existed at or near sea level, having retreated off the valley floor to an elevation approaching 100 m ASL (Figure 31). Figure 33, a pair of images taken 15-years apart, depicts the rapid loss of the lower elevation area of Little Holgate glacier.      1973, 1991, 2006, 2011, 2017. Between 1991and 2011, Holgate retreated nearly 500 m. By 2021, it had advanced more than 700 m from its 2011 position.  Summarizing all the available information about Holgate Glacier shows that while currently a tidewater glacier located at the head of Holgate Arm, its 19 th century location may have been 10s to 100s of meters retracted from where it sat when photographed and mapped in 1909. Then, two separate arms of Holgate Glacier (Holgate and Little Holgate Glaciers), approximately 750 m apart, reached tidewater. By 1950, the south side of Holgate Glacier's terminus retreated ~400m from its 1909 position, while the north side remained unchanged. By 1961, as much as 500 m of retreat of the southern margin occurred, exposing bedrock along ~1/3 of the base of the terminus. Between 1973 and 1986, the terminus retreated about 230 m. Between 1986 and 2002, little change was observed, but analysis of Landsat images showed slow continuing retreat, especially on the southern side. From 2004 to 2011, the southern margin of the glacier decreased in size, retreating ~300 m. Sometime after August 2016, Holgate Glacier began to advance, and by 2021, advance totalled ~700 m.
Harris Bay and Northwestern Fiord Harris Bay (Figure 35) is an ~20 km long fiord with maximum water depths of ~221 m. It extends from the LIA recessional/end moraine deposited by Northwestern Glacier at the head of the bay, south to its mouth adjacent to Granite Island, where it joins the Gulf of Alaska. The recessional/end moraine is the boundary between Harris Bay and Northwestern Lagoon. Northwestern Glacier Fifteen-kilometer-long Northwestern Glacier extends from the southeastern side of the Harding Icefield to tidewater at the head of Northwestern Fiord. It was named by U. S. Grant for Northwestern University. The glacier's terminus position was mapped on 23 July 1909 ( Figure 35) by Grant and Higgins. It retreated more than 15 km since it reached its LIA maximum position in the latter half of the 19 th century. When first mapped, the glacier was one of the largest ice streams on the Kenai Peninsula. On Tebenkov's 1849 chart, the glacier is shown as almost reaching the Gulf of Alaska.
Comparing Grant and Higgins 1909 sketch map of Northwestern Glacier's terminus position with the terminus position displayed on the USGS (1951) Seldovia D-1, 1:63,360 scale topographic map, Field (1975) reports that ~10 km of retreat occurred in the ensuing 42-year period. This results in an average retreat rate of 238 m/yr. Based on oblique aerial photography taken by Austin Post (Figure 36), between 1950between and 1964between , Field (1975 documented that the glacier retreated another 3.5 km. This results in an average retreat rate of 233 m/yr. Figure 36. Figure 36. Figure 36. Figure 36. Three oblique aerial photographs showing the retreating terminus of Northwestern Glacier and the rapidly expanding separation of the terminus into two lobes. The August 12, 1961 photograph was taken by Austin Post. The two 2015 photographs were taken by Bruce Molnia. Hall et al. (2005), reported that between 1973 and 1986, Northwestern's length decreased ~67 m, an average of ~5 m/yr, and between 1986 and 2002, its length decreased ~2,184 m, an average of ~137 m/yr. Adalgeirsdóttir et al. (1998), reported that between the 1950's and middle 1990's, Northwestern Glacier, with a 1950's area of 66.25 km 2 , had its terminus retreat 4.2 km and had its area decrease by 8.0 km 2 . Its ice volume decreased by 5.0 km 3 , its average elevation decreased by 80.2 m, and its mean annual mass balance was -1.5 m. Between the 1950's and middle 1990's, on an annual basis, Northwestern Glacier thinned by 1.746 m, had its volume decrease by 0.109 km 3 , and shortened 92 m.
When photographed by Bruce Molnia in July 2000, a bedrock ridge separated the retreating terminus of Northwestern Glacier into two adjacent ice tongues. Much of the margin of the eastern tongue was located above tidewater. Since then, both tongues have been slowly retreating. When observed in August 2021, each ended above tidewater and contributed ice to the fiord by avalanching, rather than calving ( Figure 37). Barren zones along the lateral margins of both tongues documented recent thinning and narrowing of the glacier. Northwestern Glacier has an accumulation area of ~54 km 2 and an ablation area of about 4.5 km 2 , a width at its face of 0.9 km, and an AAR of 0.89. Giffen et al. (2014) found that Northwestern Glacier advanced from 2000 to 2005. Figure 37. Figure 37. Figure 37. Figure 37. August 6, 2021 photograph of the retreating terminus of Northwestern Glacier, located at the northern head of Northwestern Fiord. Note that the left lobe and the right lobe are now completely separated and are two unique glaciers. While the left lobe barely reaches tidewater, the right lobe terminates meters above sea level. Note the absence of icebergs in the fiord. The photograph was taken by Bruce Molnia. Figure 38 displays four Landsat images that document the changes in the position of Northwestern Glacier's terminus in the 48-year period between 1973 and 2021. They date from August 16, 1973, June 22, 1991, August 27, 2006, and March 13, 2021. Between 1973and 1990 Northwestern Glacier retreated nearly 800 m. Since 1990, the terminus position has fluctuated at the head of the fiord. In 2021, the distance between the two lobes continued to lengthen. Figure 38. Figure 38. Figure 38. Figure 38. Four Landsat images show changes in the position of Northwestern Glacier's terminus in the 48-year period between 1973 and 2021. They date from August 16, 1973, June 22, 1991, August 27, 2006, and March 13, 2021. Between 1973and 1990 Northwestern Glacier retreated nearly 800 m. Since 1990, the terminus position has fluctuated at the head of the fiord.
Summarizing all available information about Northwestern Glacier shows that the recessional/end moraine that separates Harris Bay from Northwestern Lagoon and Fjord, located nearly 16 km from the present terminus of Northwestern Glacier, marks the position of Northwestern's terminus at the start of the 20 th century. In the previous century, it may have reached into the Gulf of Alaska. In 1909, the glacier terminated just north of this moraine. From 1909 to 1950, ~10 km of retreat occurred. Between 1950 and 1964, the glacier retreated another 3.5 km. In the nine years between 1964 and 1973, a maximum retreat of ~1 km occurred. Between 1973 and 1986, the retreat rate decreased and Northwestern retreated, ~67 m. Between 1973 and 1990, Northwestern Glacier retreated ~800 m. Between 1986 and 2002, it retreated ~2.2 km. By 2000, a ridge of bedrock separated the terminus of Northwestern Glacier into two adjacent ice tongues. Since then, aside from a small advance through 2005, both tongues have been slowly retreating. In 2021, each ended above tidewater and contributed ice to the fiord by avalanching, rather than calving. Barren zones along the lateral margins of both tongues documented recent thinning and narrowing of the glacier.
Glacier retreat and shoreline change One of the most significant impacts of glacier retreat and thinning is the exposure of land areas that previously were located under the ice. In the southern Kenai Mountains, most of these areas are at or near sea level. The change from ice-filled fiord to newly exposed marine basin, with newly exposed coastline is apparent in each of the three fiords that were studied.
Concurrent with ice loss is the onset of vegetation, usually driven by the growth of plants growing from wind-transported seeds that were deposited on glacier till and related sediments. The resulting plants and trees attract pollinators, which in turn attract avian predators. As the vegetation expands, this newly established habitat attracts small mammals, which in turn attracts larger predators. A similar ecosystem development scenario occurs in the newly exposed aquatic environment.
Many tools are used to address landscape evolution and change in newly deglacierized areas. The NPS has invested significant resources to understand the rapidly changing coastal resources in Kenai Fjords National Park. Pendleton et al. (2004) developed a change-potential index (CPI) to map the relative coastal changepotential of the shoreline of the park to future sea-level changes. All or parts of the three fiords that are the focus of this study (the Bear Glacier and Bear Glacier Lake section of Resurrection Bay, Aialik Bay, and Harris Bay / Northwestern Fiord) are located within the park.
The CPI evaluates and ranks six parameters in terms of their physical contribution to coastal change. The parameters are: 1) geomorphology, 2) regional coastal slope, 3) rate of relative sea-level change, 4) historical shoreline change rates, 5) mean tidal range, and 6) mean significant wave height. Pendleton et al. (2004) state that "The CPI was developed from a Coastal Vulnerability Index (CVI) typically applied to coastlines undergoing long-term sea-level rise. The CPI is modified from the CVI and applied to the emergent coast of Kenai Fjords National Park to understand the limits of applying this type of assessment method in a variety of sea-level settings." Kenai Fjords National Park's shorelines are described as consisting of sand and gravel beaches, rock cliffs, calving tidewater glaciers, mudflats, and alluvial fans. The areas within the park that are likely to be most susceptible to coastal change due to sea-level change are tidewater glaciers and outer coast shorelines of unconsolidated sediment where both shoreline erosion potential and wave energy are high. Figure 39 depicts the Relative Coastal Change-Potential for Kenai Fjords National Park. Note that the coastline area adjacent to Bear, Aialik, Pedersen, and Holgate Glaciers, all correspond to the highest change category -the 'very high' ranking. The location of Northwestern Glacier falls into the 'low' category. This is due to the area being part of a large granitic pluton, a rock type and material that is very resistant to erosion. F F F Figure 39. igure 39. igure 39. igure 39. Relative coastal change-potential (CPI) for Kenai Fjords National Park -Shoreline color corresponds to the intensity of potential change determined from the six variables of the CPI index. The very high change-potential shoreline is located along sandy pocket beaches where shoreline changepotential and significant wave heights are highest. The low change-potential shoreline is located along rock cliffs which are usually within sheltered locations in the fiords. Image is Figure 13 from Pendleton et al. (2004).

Conclusions Conclusions Conclusions Conclusions
This study fused data from the analysis of maps and remotely-sensed imagery, collected from different spatial perspectives (ground, air, and space), at different times from the early 20 th century to the present, using different sensors to determine the behavior of six temperate glaciers located in Alaska's southern Kenai Mountains. The data about these glaciers (Bear Glacier, Aialik Glacier, Pedersen Glacier, Holgate Glacier, Little Holgate Glacier, and Northwestern Glacier) were studied and analysed, and at a minimum, their