The Loss of Ice Worm Glacier, North Cascade Range, Washington USA

Abstract

A forty-year record (1984–2023) of glacier mass balance and areal extent measurement documented the decline and loss of the Ice Worm Glacier in the North Cascade Range, Washington. After a period of minor variations from 1944 to 1986, the glacier lost 83% of its area from 1986 to 2023 and had a cumulative mass loss of −31.5 m w.e. In 2023, the area at 32,000 m² and the majority of the ice thickness at 2–10 m was insufficient to generate movement. The bottom of the glacier was observed in all existing crevasse features, and stream channels in 2023 at depths of 2–10 m. An ice cave extended the length of the glacier in 2024 illustrating an ice thickness of less than 8 m. This glacier area loss has led to declining glacier runoff into Hyas Creek and the Cle Elum River.

1. Introduction

The United Nations has declared 2025 as the Internal Year of Glaciers’ Preservation [1]. Globally, the recent increase in mass loss rates despite significant ongoing terminus retreat illustrates an expanding number of glaciers undergoing a disequilibrium response to climate change [2,3,4]. Glaciers respond to the persistent negative mass balances by retreating, if this retreat does not re-establish equilibrium, the glacier will not survive the climate that led to this disequilibrium [5]. The increasing rate of loss of alpine glaciers has led to the addition of a map layer for extinct glaciers in GLIMS Glacier Viewer (Global Land Ice Measuring from Space) [6]. A glacier is defined “as a perennial mass of ice, firn and snow that flow” [7]. Movement is what separates a glacier from other perennial snow and ice features and is usually identified by the presence of crevasses [7,8]. Most studies rely on satellite and aerial images, hence they rely on a minimal size threshold, commonly somewhere between 0.05 km² and 0.01 km², to distinguish a glacier from perennial snow patches [8,9]. Below 0.05 km² movement is difficult to sustain as ice thickness is limited, with ice flow not being possible below 20 m [10]. A simple minimum size threshold will not identify the lack of movement that characterizes most of these small ice features. Current satellite image resolution allows the mapping of snow and ice features that are smaller than this, which is useful, even when they are not glaciers [11,12]. It can be difficult to discern when a glacier is no longer moving from satellite imagery; however, field verification provides a straightforward means to assess this [5,13].

The North Cascade Glacier Climate Project (NCGCP) began in 1984 to identify the response of North Cascade glaciers across the mountain range to regional climate change [5,14]. In 1984, the North Cascade Glacier Climate Project began the periodic monitoring of 107 glaciers across the range. By 1995, all 107 glaciers, where we had measured terminus change in the 1980s and again in the 1990s, had retreated [5]. By 2005, a disequilibrium response of North Cascade glaciers was identified with ongoing retreat accompanying increased mass balance loss [5,14]. This included the loss of several glaciers such as Lewis, Milk Lake and Spider Glacier [5]. In the 21st Century, the pace of glacier change in the North Cascade Range has accelerated, matching the record of glaciers around the globe experiencing rapid melt [3,14].

Here, we examine the loss of Ice Worm Glacier, which after 40 consecutive years of annual field monitoring was determined no longer to be a glacier in 2023. Annual observations of mass balance and periodic ice margin mapping were completed from 1984 to 2023. This glacier area loss has led to a decline in late summer glacier runoff to Hyas Creek and the Cle Elum River. This is likely the longest detailed annual record of observations of a glacier leading up to loss, providing a unique record of a glacier disappearing, that will become a common story [1].

2. Study Area

Ice Worm Glacier is an east-facing cirque glacier on the east flank of Mount Daniel, WA. Mount Daniel is on the crest of the North Cascade Range of Washington, the crest separates the dry east side that drains to the Columbia River and the wet west side that drains to Puget Sound (Figure 1). The glacier is at the headwaters of Hyas Creek, which drains into the Cle Elum River and the Cle Elum Reservoir. The Cle Elum Reservoir has a storage volume of 538,900,000 m³ and is primarily used for flood control in spring, and agricultural irrigation in summer. This reservoir is the largest in the Yakima River Basin and provides irrigation to 180,000 hectares of agricultural land [15]. The glacier is located in the Alpine Lakes Wilderness area which prohibits the use of mechanized equipment. The Ice Worm Glacier cirque floor is at 1940 m and the headwall at 2050 m. There is a bench at 2150–2300 m that held a glacier/perennial icefield prior to 2015. The glacier is accessed by backpacking 8 km from the Cathedral Rock trailhead with the same base camp used each year at 1700 m on the bank of Hyas Creek.

The glacier was fed by wind drift accumulation along the ridge on the south side of the glacier that was just above the top of the glacier during the 1944–2000 period. Avalanching from the slopes below the East Peak of Mount Daniel and from the ridges extending along the north and south flank of the glacier has also been significant.

3. Methods and Data Sources

3.1. Glacier Mass Balance Measurement

The North Cascade Glacier Climate Project (NCGCP) has monitored the annual mass balance during the period 1984–2024 on Ice Worm Glacier [14,16]. The measurement network is a grid covering the entire glacier with measurements at a spacing of 30 m, yielding a minimum of 40 measurements of accumulation and ablation. The resulting density is over 500 points km⁻², much denser than typical mass balance networks. Snow depth is assessed via probing of the snowpack to the impenetrable previous summer surface. Ablation is measured using ablation stakes, changes in snow depth from repeat probing measurements and from snowline migration [14,16]. These data, including the specific glacier area, number of measurements and annual mass balance are reported annually to the WGMS [17]. The reported error to the WGMS for Ice Worm Glacier because of the high density of measurements that reduces extrapolation is ±0.15 m w.e.

During the course of mass balance observations, the depth of crevasses, moulins and surface streams is measured. using a rope marked in 0.25 m intervals. This is important when bedrock is observed at the base, and ice thickness is determined.

3.2. Glacier Area Mapping

The two primary global glacier inventories where each glacier has a delineated margin and resulting area are GLIMS (Ice Worm = G238839E47559N) and the Randolph Glacier Inventory (Ice Worm = RGI160—02-18346) [6,18]. The GLIMS and RGI in this region relied on aerial photographs. For small glaciers like this high-resolution satellite imagery, such as Landsat 8–9, or Sentinel 2 with 10 m to~30 m resolution is sufficient to accurately map glacier margins, but was not available before 2013 [11,12]. Aerial imagery utilized by RGI had a resolution of 6 m [8]. In the field, we mapped the glacier area with a GPS dynamic point mapping technique marking the perimeter of the glacier in 1986, 1992, 2005, 2015, 2021, 2022, 2023 and 2024, starting and finishing at Ground Control Points. The 1986 and 1992 observations pre-dated sufficiently accurate GPS positioning technology; however, perimeter stakes were emplaced and left in until 2003 when they were resurveyed using GPS and removed. The importance of field inspection to distinguish glacier ice, from ice-cored moraine and detached areas of ice was illustrated by [13]. The GPS points observed can be readily compared with satellite imagery from the same month for validation and visualization [11,13]. Sentinel 2 imagery has the best resolution and is readily available and applicable to identifying small glacier margins [11,12]. Field surveys were conducted between August 13 and 16th each year. Sentinel 2 imagery was available from 2021 to 2024 in late August or Early September each year. The GPS observations are imported into the Sentinel Hub browser for direct comparison with satellite imagery, matching Ground Control Points [19]. The accuracy of GPS in this setting is 3–5 m [11,20].

The error in spatial area determination using GPS field observations for areas between 10,000 m² and 100,000 m² was determined to be ±1–3% with specific shapes having little impact [13,19,20].

4. Results

4.1. Glacier Area Change

In 1986, William (Bill) Prater, who had made many first ascents in the area between 1944 and 1960, joined us in the field. Comparing images from these early visits with the current margin of Ice Worm Glacier indicated that there was little change in this glacier from 1944 to 1986. The mapped area in 1958 was 0.19 km² [6]. In 1986, we mapped the area at 0.18 km². From 1984 to 1992, the glacier extended to within 15 m of the ridge on the south side of the basin. By 2006, the recession from this ridge was greater than the recession at the terminus (Figure 2). The glacier perimeter was surveyed in the low snow years of 2005 and 2015, identifying the glacier area to be 0.15 km² and 0.110 km², respectively. In 2015 [8], the inventory for RGI noted an area of 0.106 km². Annually, from 2021 to 2024, low snow cover allowed for mapping of the glacier perimeter during a rapid decline from 0.09 km² to 0.03 km². The area loss from 1986 to 2015 was 0.07 km², which is less than the area loss from 2015 to 2023 of 0.08 km². Figure 3 provides a photo comparison of the change from 1986 to 2023, with the people standing in the same location in each case. Figure 4 provides a view from the above in 2024 from the north ridge and south ridge, illustrating the change in glacier area from 1986 to 2024. The observed area determined in the field from the GPS position locations closely matches the RGI inventory area in 2015 and the area derived from overlaying the points on Sentinel 2 imagery from August 2021 to 2024, which have a 10–30 m resolution (

Table 1. The observed area of Ice Worm Glacier from field measurements. The area of Ice Worm Glacier from previous inventories and Sentinel 2 imagery.

Year	Field Mapped Area (m2)	Validating Aerial/Satellite Image Area (m2)
1958		190,000 (GLIMS)
1986	180,000 (±10,000)	Field Observation only
1992	170,000 (±10,000)	Field Observation only
2005	150,000 (±5000)	Field Observation only
2015	110,000 (±5000)	106,000 (RGI)
2021	88,000 (±2000)	90,000 (Sentinel)
2022	68,000 (±2000)	Field Observation only
2023	41,000 (±2000)	40,000 (Sentinel)
2024	32,000 (±2000)	30,000 (Sentinel)

).

4.2. Glacier Mass Balance Observations

The forty-year mass balance record of Ice Worm Glacier illustrates a negative balance for each of the four decades (

Table 2. The decadal mean annual balance of Ice Worm Glacier and that of all nine North Cascade glaciers including Ice Worm Glacier reported annually to the WGMS (Columbia, Daniels, Easton, Ice Worm, Lower Curtis, Lynch, Rainbow, Sholes and South Cascade).

Decade	Mean Annual Balance (m w.e.)	All North Cascade Glaciers (m w.e.)
1984–1993	−0.62	−0.47
1994–2003	−0.28	−0.32
2004–2013	−0.81	−0.63
2014–2023	−1.45	−1.38

). With a substantial increase in the last decade to −1.45 m w.e. a⁻¹, compared to −0.57 m w.e. a⁻¹ during the first three decades; Figure 5. The decadal mass balance matches that of all nine North Cascade glaciers with long-term mass balance records reported to the WGMS, including eight from NCGCP and the South Cascade Glacier observed by the USGS. The annual mass balance correlation coefficients with the other glaciers range from 0.85 for Rainbow Glacier to 0.95 for the adjacent Daniels Glacier. This high degree of correlation is evident in the chart of the annual mass balance of Ice Worm Glacier compared to the mean of all the assessed North Cascade glaciers (Figure 5). This is an illustration that the climate change experienced by the glacier is sufficient to overwhelm the physiographic differences of the glaciers [14] Each year from 2021 to 2024, the annual mass balance was below −1 m w.e., and no significant accumulation zone remained at the end of summer. The 2021–2024 cumulative mass balance was −8.5 m w.e., which is approximately 10 m of thinning, representing more than 40% of the glacier volume in 2021.

There has been a substantial increase in the ablation rate observed in the last decade, with the average daily ablation of glacier snowpack observed in August, which has increased from 0.041 m w.e. d⁻¹ to 0.054 m w.e. d⁻¹. In 2005, an area of 150,000 m² with a melt rate of 0.041 m. w.e. d⁻¹ would yield 6150 m³ of runoff daily. In 2023, with an area of 40,000 m² and a melt rate of 0.054 m w.e. d⁻¹, glacier runoff would be 2160 m³ daily, a 65% decline in glacier melt.

The main change in accumulation has been a reduction in avalanches onto the glacier from the slopes below the East Peak of Mount Daniel, see Figure 3 and Figure 4. This slope and any depressions on it were largely filled by perennial snow and glacier ice until 2015. After 2015 this slope area has lost its perennial snowpack, revealing several shallow basins. Each winter, these basins must now be filled with new snow before avalanching can occur. In 2023, a complete loss of snow cover by early August led to a rapid 3.5 m decline of the glacier surface. The rate of loss was enhanced by dirtier ice surfaces after the two previous years when snow cover had also been lost.

4.3. Glacier Base Observations

Beginning in 2015, we routinely assessed how deep each moulin, crevasse or supraglacial stream channel was on the glacier. One measure of a glacier no longer being a glacier is when crevasse features and stream channels consistently reach the bedrock below the glacier. In 2023, we examined 24 of these features and each reached bedrock at depths of 2–10 m (Figure 6). These features were distributed widely across the glacier. There are undoubtedly limited areas of thicker ice. In 2024, we explored an ice cave that extended 250 m from the top of the glacier to the end of the glacier. The cave roof was 1–4 m above bedrock, and the roof was consistently less than 2 m thick allowing light to penetrate from the glacier surface into the cave (Figure 7). An extensive ice cave transecting the entire glacier such as this cannot exist in a current glacier because ice movement would lead to ice cave closure.

5. Discussion

Glacier inventories compiled from aerial and satellite imagery rely on glacier size as the primary criterion to identify a glacier [8,9,11]. To determine if there is movement [7], which defines a glacier, field identification is an ideal practice when possible [13]. The Ice Worm Glacier represents a unique case with detailed field measurements spanning 40 years. Beginning in the 2003–2006 period, the glacier began to experience consecutive years without an accumulation zone. This is indicative that the glacier cannot survive [5]. From 2019 to 2023 no accumulation existed, except in 2020. The rapid loss of glacier area and ice thickness from 2021 to 2024 has led to the Ice Worm Glacier no longer meeting the criteria to be a glacier, there is no longer movement or ice thick enough to allow movement [10]. This joins several other glaciers that we have worked on in the North Cascades that we have observed disappear such as Milk Lake, Lewis and Spider Glacier [5].

The annual mass balance record of Ice Worm Glacier is highly correlated with both melt temperatures (June–September) and the April 1 snowpack snow water equivalent [14]. Long-term climate trends for the region are evident in the long-term climate records from NOAA’s National Center for Environmental Information (NCEI) Washington Division 5 [21] and April 1 SWE at USDA SNOTEL stations [22]. Division 5 is the Cascade Mountain West region and contains complete data for the 1896–2023 period [21]. The United States Department of Agriculture-SNOTEL (snow telemetry) program, has six long-term April 1 SWE records from six North Cascade SNOTEL stations, observed annually since at least 1946 (Fish Lake, Lyman Lake, Park Creek, Rainy Pass, Stampede Pass, Stevens Pass). This includes one station in the basin at Fish Lake at 1045 m [22].

Trends in summer temperature at Western Cascade weather stations for the 1896 to 2023 period indicate that seven of the ten warmest melt seasons (June–September) have occurred since 2013. Melt season temperatures from 2014 to 2023 were 1.3 °C above the 1896 to 2022 mean. The long-term winter temperature trend from 1896 to 2023 has been 1.2 °C. From 2014 to 2023 winter temperatures were 0.8 °C above the long-term average, the warmest decadal period of the record.

During the 1896 to 2023 period from June to September, precipitation exhibited no significant trend. For the November–April (winter season) there was a 3% increase in precipitation from 1896 to 2023. From 2014 to 2023, winter precipitation averaged 1.68 m vs. the long-term average of 1.65 m. April 1 SWE from six long-term SNOTEL stations where April 1 SWE has a declining trend of 30% from 1946 to 2023, with a 10% decline since 1984. The April 1 SWE loss reflects increased melting of the snowpack or rain events during the winter season [14].

Each year in mid-August, stream discharge has been observed immediately below the 1986 terminus position of Ice Worm Glacier at 11 a.m., 2 p.m. and 5 p.m. during the field visit. This stream is also fed by perennial snowfields and, before 2015, a small glacier. These observations are insufficient to quantify daily runoff, but because of the consistent timing and methods, do allow for comparison. The glacier is located on the dry side of the range and has not experienced rainfall during any of our field observation periods, which would contribute to stream discharge. From 1985 to 2002, streamflow was observed on 36 days, with the average discharge being 0.12 m³ s⁻¹. From 2021 to 2024, discharge was observed on 6 days with the average discharge being 0.03 m³ s⁻¹. This roughly 75% decline in August runoff is similar to the 60% loss in runoff modeled for the loss of small glaciers in basins in the Alps and observed when the Lewis Glacier, North Cascade Range was lost [10,23]. The change in summer streamflow in Hyas Creek near our base camp is apparent, as it had been a challenging stream crossing to keep your feet dry until 2013, and now is a simple step across. There has also been a marked increase in algae coating the substrate of the stream with the lower flow, clearer and warmer water, which is expected [24].

Accelerated glacier area loss and glacier disappearance have been noted at Mount Hood, Oregon [13], Olympic Mountains, Washington [25] and Vancouver Island, British Columbia [12]. On Mount Hood, one glacier was lost between 2003 and 2023 and the glacier area has declined by 40% [13]. In the Olympic Mountains, WA 35 glaciers have disappeared since 1980 [25]. In 1984 there were 55 glaciers on Vancouver Island and by 2020 only 38 remain [12].

6. Conclusions

Identification of glacier loss utilizing cessation of movement is not practical on most glaciers, but on glaciers where other field measurements are being made, can be readily included. The cumulative mass balance loss of Ice Worm Glacier from 1984 to 2023 was −31.5 m, markedly increasing in the 2014–2023 decade. For a glacier that had an estimated average thickness of 30–40 m in 1986, this resulted in the loss of more than 80% of the glacier’s area and volume. The glacier area declined from 0.18 km² in 1986 to 0.032 km² in 2024, with particularly rapid loss after 2020. As the glacier continues to decline and the basal topography is revealed, the prior overall volume of the glacier can be quantified. The 83% decline in glacier area despite a 20% increase in melt rate, is resulting in less glacier runoff late in the summer.