Observation of Greenland Ice Velocities and Impact of Surface Melt

Abstract

Isbreer beveger seg og isdynamikken kontrolleres av mange prosesser. I dette arbeidet fokuserer vi spesielt på basalglidning, eller glidning mot underlaget. Basalglidning er interessant å studere fordi grønlandsisen mister masse til havet, og basalglidning står for opptil 90% av bevegelsen til isdekket. For å studere basalglidning, jobber vi med isbreer i sørvest-Grønland. De fleste isbreene her er landbaserte, slik at vi kan isolere glidningsprosesser og ikke trenger å tenke på prossesser mellom hav og is. En bedre forståelse av denne prosessen er viktig for å si noe om hvordan isbreer kommer til å respondere på fremtidig klima.

På sommeren smelter snø på toppen av isbreene, smeltevannet renner ned gjennom sprekker i isen og påvirker hvordan isen glir på underlaget. For å forstå hvordan smeltevann påvirker hvordan breene beveger seg over lange tidsskalaer, produserte vi isbevegelseshastighetskart for alle breene i sørvest-Grønland for perioden 2000-2019. Tidligere studier har visst at gjennomsnittsbevegelsen i disse isbreene har bremset opp mellom 2000 og 2012, og hastighetskartene våre bekrefter dette resultatet. For perioden 2012-2019 viser noen studier at breene akselererer igjen. Dette resultatet støttes ikke av våre hastighetskart. Årsaken er at tidligere studier har brukt hele regionens gjennomsnittshastighet (et datapunkt per år), mens vi har sporet ishastigheten på individuelle 150-kvadratmeters piksler (millioner av datapunkter per år). Vi viser at denne mer detaljerte fremgangsmåten er nødvendig fordi isbevegelsen ikke er konstant innenfor isdekket. Vår studie viser at det er viktig å sørge for at bildene, som brukes til hastighetskartene, er tatt i samme periode hvert år for å ha tall som er sammenlignbare fra år til år.

For å se på sammenhengen mellom produksjon av overflatesmeltevann og ishastigheter, kombinerer vi også hastighetskartene med klimareanalyser. Vi viser at ishastighet ikke varierer lineært med smeltevannsavrenning, men at ishastigheten først øker, så minker, og deretter øker igjen med økende avrenning. Det er ikke tidligere påvist at høy smeltevannsproduksjon kan føre til raskere ishastigheter.

I vår studie analyserte vi bare en liten del av kartdatasettet i detalj. Ishastighetskartene for årene 1984 til 2020 for sørvest-Grønland er åpent tilgjengelige.

Glaciers are flowing. While several processes take part in ice movement, this thesis focuses on one in particular: basal sliding. We are particularly interested in this last one as the Greenland Ice Sheet is losing mass through discharge of ice into ocean, and basal sliding accounts to up to 90% of the displacement of ice. In order to study this process, the study area was specifically chosen, Southwest Greenland, where most glaciers are land-terminating, so that ice-ocean interactions do not have to be taken into account. Recent studies also argued that, for the two other processes to consider (deformation of the ice, and deformation of the sediment underlying the ice), ice deformation is almost negligible, and hard-bed conditions were observed in our area. Being able to isolate basal sliding is an advantage to help understand how this single process works, and how it might be affected by climate. The first part of the scientific results reassesses the ice velocity trends over 2000-2019 using yearly optical feature-tracking, but goes beyond previous scientific work, as we derive here linear regressions for every single pixel in our area, while previous work derived velocity trends on area-wide yearly velocity averages. We used two time periods for our study, 2000-2012 and 2013-2019, and found a significant deceleration for the first period, but only find significant reacceleration for the second period in small part of the study area, namely low-lying glaciers tongues. We find inconsistent results for the rest of the study area, while another study did find a significant acceleration using area-wide yearly velocity averages. We display per-pixel linear regressions slopes maps and associated p-values maps in opposition to difference maps in previous papers, and proved that we cannot assume similar ice velocity trends for the entire area. Our second result in this first part is that the way data is averaged for yearly averages can affect the obtained velocity. Previous studies did use all data available, therefore aggregating yearly velocities derived from image pairs that start before, during and after summer. We show that doing so can affect yearly velocities by up to 15%, and argued for the selection of image pairs that start before summer, so that the period assessed is comparable for every year, encompassing a summer and the following winter.

For this first part, we derived a dataset of yearly velocities for the years 1984 to 2020 for Southwest Greenland. Constituting such dataset for the whole Southwest of Greenland nearly took two years, and we rendered all the data publicly available on an open repository in December 2022.

The second part of the scientific results is built from the same dataset containing yearly ice velocities in the same area, but combined with climate reanalysis. We focus on 5 basins in our study area, and for the period 2013-2021. The velocity variations in this land-terminating part of Greenland are relatively small, therefore we followed the recommendations from our first publication, and only selected image pairs starting before summer. This precise selection of data is only possible after the launch of Landsat 8, reducing our dataset to after 2013. We find that the impact of meltwater on velocities varies with altitude and location, and that positive melt anomalies do not necessarily lead to slower velocities, especially for the year 2019 were the melt was up to 40% above the 2013-2021 average, and where some basins show a faster than average velocities.

Both parts underline the importance of local ice behaviors: ice velocity trends vary with basin and altitude, and subsequently the relationship with runoff was also observed to vary in location and altitude. We therefore urge further studies to not consider similar ice velocity behaviors for different parts of the Greenland Ice Sheet, especially when performing long-term ice loss estimations.