Vegetation browning is the decline in plant biomass and productivity arising from climate change, biotic interactions and disturbance. It is now considered one of the major disruptions in a rapidly changing Arctic landscape. Damaged Arctic vegetation due to extreme winter weather events such as warming events and frost drought conditions, has been shown to change from a sink to a net CO2 source at the peak of the growing season. It is crucial to understand the satellite-based signature of browning events due to the challenging nature of field work in the Arctic and the sporadic nature of such events. It is important to understand how browning events can unfold in the future in response to projections of increased frequency, magnitude and severity of extreme winter weather events in the Arctic. This research is the first to provide a remote sensing and climate modelling based framework to examine Arctic browning. Northern Norway was selected as the study area for this PhD research. The first research objective of this PhD thesis was to understand the satellite-based signature of browning events caused by extreme winter weather conditions. This was achieved through examining the effectiveness of two different MODIS vegetation indices at quantifying the on-record ground observations of vegetation decline in the Norwegian Arctic and sub-Arctic areas. The indices included the Chlorophyll Carotenoid Index (CCI) and the Normalized Difference Vegetation Index (NDVI). The CCI and NDVI were extracted for early, peak and end of the growing season (July-September). Moreover, the average growing season CCI and NDVI were calculated as well. These calculations were conducted for three case study sites in northern Norway. The NDVI presented a more robust signal compared to CCI for detecting decreases in the Gross Primary Productivity (GPP) of dwarf shrub vegetation across different Arctic landscapes. This was concluded to be mainly due to the higher spatial resolution of NDVI (0.25 km) compared to that of CCI (1 km). The second research objective of this work was to determine the main meteorological drivers of satellite-based observations of vegetation decline in the Norwegian Arctic and sub-Arctic. Currently there is a substantial research gap with regards to the understanding of relationships between the variability of individual meteorological variables in winter and the summer NDVI. For this, a regional climate model, the Weather Research and Forecasting Model (WRF), was used to produce high-resolution (1 – 10 km) simulations for the winter months November – April, over the time period 2000 – 2020, for northern Norway. The driving dataset for WRF here was ERA5. WRF’s skill at reproducing the extreme winter weather conditions, which lead to recorded browning events at the three case study sites was examined, considering variables including 2m near-surface temperature, snow depth and precipitation. WRF was able to simulate extreme winter warming and low snow depth conditions at the case study sites after bias-corrections were applied. Following this, correlations between the different winter month-based meteorological variables and mean summer NDVI were examined. The correlations identified the most important winter meteorological variables with regards to summer NDVI, for the study area. These variables were used in multivariate regression analysis against summer NDVI to develop statistical models for projecting summer NDVI at the end of this century under different emission scenarios.
This leads to the third research objective of this thesis, which was to assess the changes in frequency and intensity of climatic drivers of Arctic browning at the end of this century in the Norwegian Arctic and sub-Arctic. Therefore, WRF was forced with the Community Earth System Model (CESM1) under three Representative Concentration Pathways (RCPs) 4.5, 6.0 and 8.5, for 2090 – 2100. The future simulations were compared with a historical baseline, 1990 – 2000, to assess the changes in the frequency and spatial extent of the different winter meteorological drivers of NDVI. The findings of this work can be viewed in a threefold-perspective; spatial context, seasonal winter meteorology and climate change scenario based. In the spatial context the main findings included; the vegetation most at risk of damage is projected to be in Trøndelag County, based on the strongest increases in frequency and intensity of winter warming events, low snow depth conditions and ROS occurrence. This research’s projections about increased exposure of Norway’s coastal areas to higher intensity warming events (duration-based), as compared to the inland regions, agrees with previous studies. Large spatial variability was found across the study domain with regards to the meteorological parameters and extreme weather indices of different winter months affecting the summer NDVI. The projections of browning frequency at one of the case study sites (Storfjord), located well inside the Arctic Circle, are reflective of the pronounced negative impacts arising from multiple extreme winter weather events and conditions. At this site the maximum duration of winter warming events index (MDW) in December and the mean January temperature best explained the variance in the NDVI. In context of the three RCPs studied here, major findings with regards to overall impacts on vegetation included projections of mean December, January and March temperatures staying above 0℃ for most of the study area. These temperature projections imply an increased probability of ROS in these peak winter months as precipitation would likely fall as rain rather than snowfall. Moreover, as vegetation can get damaged under low-snow conditions, it is concerning that under RCP 8.5, the average number of days with snow depth < 20 cm (SC20), per winter season, is projected to increase by 80-100 days, in Trøndelag County, compared with the 1990 – 2000 time period. In general, this study predicts large scale vegetation disturbance in response to changes in the overall winter meteorology in northern Norway.
