Thermokarst and precipitation drive changes in the area of lakes and ponds in the National Parks of northwestern Alaska, 1984–2018

Figures & data

Figure 1. Study-area location. The ecological subsections analyzed for water surface area are displayed, with their lake and pond surface area in percent. The excluded areas are subsections that had less than 1 percent lake and pond surface area. National Park Service unit abbreviations are: Bering Land Bridge National Preserve (BELA), Cape Krusenstern National Monument (CAKR), Gates of the Arctic National Park and Preserve (GAAR), Kobuk Valley National Park (KOVA), and the Noatak National Preserve (NOAT).

Figure 2. Mean annual air temperature for Kotzebue, Alaska, 1980–2018. National Weather Service data from NOAA Regional Climate Centers (2018). Points for years mentioned in the text are labeled. Means were computed for water years (October–September).

Figure 3. Annual precipitation and estimated annual potential evapotranspiration at Kotzebue (top) and Bettles (below). Precipitation and temperature data were obtained from NOAA Regional Climate Centers (2018). Evapotranspiration was estimated by the Hargreaves-Samani method, which is the most accurate at high latitudes of the simple temperature-based methods (Hargreaves and Allen 2003; Almorox, Quej, and Pau 2015).

Figure 4. Examples of generalized lake landscapes: ice-rich thaw lake plain (left), ice-rich deep silt (center), and glacial deposits with low to moderate ice content (right). The former two landscapes have lakes of mainly thermokarst origin and outlines of drained former lakes are visible on the images.

Figure 5. Histograms of Landsat SWIR reflectance (times 10³) of random points in water bodies and on land in Landsat ARD tiles h02v04 and h04v03 for years 1984–2017.

Figure 6. (a) Water surface area versus year for the Noatak Glaciated Lowlands subsection (NOAT). This subsection showed a strong minimum at approximately 2010, with no significant long-term change. (b) Water surface area versus year for the Bering Straits Lower Coastal Plain subsection (BELA). This subsection showed a significant long-term decline, with a minor minimum at approximately 2008–2010. The estimated 2018 value was determined by subtracting from the 2017 value the area of lakes that drained abruptly in 2018. (c) Water surface area versus year for the Nigu River Valley (GAAR). This subsection showed an increase in water surface area from 2000 to 2017. Note the different scales of the three figures: the range of variation in (b) is large, more than 1 percent, compared to 0.3 percent and 0.25 percent in (a) and (b).

Table 1. Lake area change 2000–2017 by ecological subsection^a.

Subsection	NPS Unit	Permafrost	Subsection Area (km^2)	Lake Area (%)	Slope (ha yr)^a	p Value	Ratio of r^2 (linear/segmented)	Past Ten Years vs. 1980–1990
Subsections with Decreasing Water Area 2000–2017
BSLCP	BELA	Thaw-lake plain	1,630	9.0	−71.6	<0.01	0.93	-
BSUCP	BELA	Ice-rich	1,327	6.2	−19.4	<0.01	0.73
NIP	KOVA	Ice-rich	750	1.3	−7.3	<0.01	0.71	-
IL	BELA	Ice-rich	339	23.5	−7.3	<0.01	0.59
AKP	KOVA	Ice-rich	381	1.5	−5.3	<0.01	0.72
WUL	CAKR	Ice-rich	351	0.9	−2.6	<0.01	0.52	-
LNM	CAKR	Ice-rich	97	4.2	−0.7	0.01	0.44
Subsections with Fluctuating Water Area 2000–2017
DU	BELA	Ice-rich	826	9.8	−2.9	0.54	0.07	+
NGL	NOAT	Ice-poor	655	4.6	−2.9	0.40	0.00
GBLCP	BELA	Thaw-lake plain	116	5.8	−1.3	0.20	0.07	+
ACP	CAKR	Thaw-lake plain	112	2.8	0.1	0.76	0.02	+
IMP	CAKR	Thaw-lake plain	18	9.1	0.2	0.52	0.26	+
KOY	GAAR	Ice-rich	230	2.5	0.3	0.79	0.00
AGU	NOAT	Ice-poor	231	11.8	0.7	0.40	0.08
KOB	GAAR	Ice-rich	240	1.3	0.7	0.10	0.18
MNU	NOAT	Ice-poor	544	4.6	0.8	0.50	0.04
NOA	GAAR	Ice-poor	177	6.8	0.9	0.32	0.43
AHW	KOVA	Ice-poor	142	12.1	2.4	0.26	0.19
GBUCP	BELA	Ice-rich	1,320	3.1	2.6	0.40	0.05	+
IGU	NOAT	Ice-poor	984	3.1	3.4	0.40	0.35
LNL	NOAT	Thaw-lake plain	404	10.5	6.4	0.60	0.31
UNB	NOAT	Ice-poor	2,475	2.0	8.1	0.21	0.30
NGU	GAAR	Ice-poor	207	3.9	1.5	0.09	0.60
KGU	NOAT	Ice-poor	584	3.0	2.6	0.07	0.43
Subsections with Increasing Surface Water 2000–2017
KZL	BELA	Ice-rich	226	4.7	1.7	0.04	0.38	+
ITK	GAAR	Ice-poor	31	17.5	0.4	0.02	0.57
KOP	CAKR	Thaw-lake plain	118	2.4	1.1	0.02	0.62	+
KIL	GAAR	Ice-poor	92	5.7	2.8	0.01	0.84
NIG	GAAR	Ice-poor	78	4.7	1.1	<0.01	0.96
ALA	GAAR	Ice-poor	23	22.3	1.6	<0.01	0.98

^aNPS Unit; see Figure 1.

Slope. Rate of change in water surface area, ha yr^a. Theil-Sen non-parametric slope.

p value. Probability value for the Mann-Kendall trend test.

Ratio of r²: linear/segmented. Ratio of r² of linear regression to r² of a three-part segmented regression with break points at the years 2010 and 2013.

Past ten years vs. 1980–1990. A dash (-) indicates that water area in all of the past ten years (2008–2017) was less than both the 1980s and 1990s values. A plus (+) indicates that water area was greater in all of the past ten years than in the 1980s or 1990s.

Figure 7. (a) Lake-area change in the western part of the study area (BELA), (b) the central part (CAKR, KOVA, and NOAT), and (c) the eastern part (GAAR). Subsection colors give the overall trend in water area 2000–2017: red for decreasing and blue for increasing, with bright colors indicating significance at p < .05. Circles mark the location of lakes that drained, with circle size proportional to the area lost and shading keyed to the decade of drainage: white (1985–2000), gray (2001–2010), and black (2011–2017). Drained lakes are defined as those that had half or less as much surface area in 2017 as 2000. The year they first lost 25 percent or more of their area was identified as the year of drainage.

Figure 8. (a) Loss of water area by year of the 108 lakes that were identified as “drained” during 2000–2017. These were lakes that had half or less as much surface area in 2017 as 2000, and had a maximum surface area of 5 ha or more. (b) Count of new lake-drainage events by year, 2000–2017. The final values (2018) on both plots were estimated from partial coverage of the study area in September 2018.

Figure 9. Lake loss in northeastern BELA. Areas that lost water between 2017 and 2018 are shaded in yellow, and those more than 5 ha in area are labeled in yellow with area lost, in hectares. Pre-2017 lake losses are outlined and labeled with the year in orange. The white lines outline ecological subsections (Jorgenson 2001) used as analysis areas: BSLCP, Bering Straits Lower Coastal Plain; BSUCP, Bering Straits Upper Coastal Plain; DU, Devil Uplands; GBLCP, Goodhope Bay Lower Coastal Plain; GBUCP, Goodhope Bay Upper Coastal Plain. The background image is from 2010 (Alaska Geospatial Council 2011).

NOAA Regional Climate Centers. 2018. XmACIS Version 1.0.38. XmACIS Version 1.0.38. 2018. http://xmacis.rcc-acis.org/.

 Google Scholar

Hargreaves, G. H., and R. G. Allen. 2003. History and evaluation of Hargreaves evapotranspiration equation. Journal of Irrigation and Drainage Engineering 129 (1):53–63. doi:10.1061/(ASCE)0733-9437(2003)129:1(53).

 Web of Science ®Google Scholar

Almorox, J., V. H. Quej, and M. Pau. 2015. Global performance ranking of temperature-based approaches for evapotranspiration estimation considering Köppen climate classes. Journal of Hydrology 528:514–22. doi:10.1016/j.jhydrol.2015.06.057.

 Web of Science ®Google Scholar

Jorgenson, M. T. 2001. Ecological subsections of Bering Land Bridge National Preserve. Anchorage: Inventory and Monitoring Program, National Park Service, Alaska Region. https://irma.nps.gov/App/Reference/DownloadDigitalFile?code=418816&file=BELA_EcologicalSubsections.pdf.

 Google Scholar

Alaska Geospatial Council. 2011. SDMI SPOT 5 2.5-meter statewide imagery. Anchorage, Alaska. http://gis.dnr.alaska.gov/terrapixel/cubeserv/ortho?DATASTORE=SPOT5_SDMI_ORTHO_CIR&SERVICE=WCS&REQUEST=GetDescription.

 Google Scholar