A seasonal weather perspective on forest vitality, vapor pressure deficit, and Greenland melt in a warming climate

Abstract

A wide range of meteorological and climatological phenomena shape atmospheric variability on different timescales. For example, convection in thunderstorms leads to intense precipitation within minutes, while stationary anticyclones often cause multi-day heat waves. In contrast, only few atmospheric phenomena operate on the seasonal timescale, where climate variability has a very strong impact on many natural and socioeconomic systems. Moreover, seasonal climate variability is relatively more affected by global warming than shorter-term variability, making these systems particularly vulnerable to ongoing global warming. Therefore, the aim of this thesis is to improve our understanding of seasonal climate variability, which is crucial for the development of timely adaptation and preparedness strategies. To achieve this goal, our approach adopts a “weather perspective”, i.e., we investigate how shorter-term atmospheric variability aggregates to form seasonal anomalies and extremes. We also focus on three selected components of the climate system that show remarkable sensitivity to seasonal variability. First, we examine the meteorological precursors of low forest vitality events in Europe during summer (June-August; JJA) using satellite observations of forest greenness. Although these events are indicative of the observed drought-related forest dieback, they have not been systematically assessed from the weather perspective yet. Second, we examine extremely high vapor pressure deficit (VPD) in JJA, which is a major contributor to plant water stress and thus crop failure and wildfire risk. Extreme seasonal VPD can be caused by a combination of air temperature anomalies (T’) and humidity anomalies (q’), which we investigate for the first time over the northern mid-latitudes. Finally, we examine the increase in summer melt of the Greenland Ice Sheet (GrIS) over the 21st century, which is expected to contribute significantly to global sea-level rise. In particular, we quantify the role of melt expansion and intensification, and changes in synoptic circulation patterns. In Chapter 3, low-forest-greenness events in 2002-2022 are identified separately for the European temperate and Mediterranean forest biomes as widespread (on spatial scales of 50 × 50 km2) negative anomalies of the Normalized Difference Vegetation Index (NDVI) over most of JJA. First and foremost, according to these criteria, forest greenness was negatively affected in the summer of 2022 (the hottest on record) to an extent unprecedented in the study period. Low-NDVI events covered 37% of the forests in both biomes, exceeding previous records of around 24% coverage. Meteorological precursors to all events are then identified as 90-day meteorological signals that are statistically significantly different from climatology and shared among the events occurring in a biome. The summers of the low-NDVI events were unusually hot and dry. Negative precipitation anomalies (P’) reached statistically significant levels already in the winter preceding Mediterranean events. In temperate forests, the previous summer was also anomalously warm and dry, indicating potential negative legacy effects of drought. Notably, the persistence of dry periods (P’ < 0) was significantly increased for at least ~2 and ~3 years before low-NDVI events in the temperate and Mediterranean biomes, respectively. However, the persistence of hot periods (T’ > 0) was significantly increased only in the temperate biome, but not in the Mediterranean biome, over about the same 2-year period. Unusually large T’ and P’ were associated with significant changes in the frequency of weather systems, with less frequent cyclones relevant for dry periods in Mediterranean forests and more frequent anticyclones for dry periods in temperate forests. The analyses take into account the uneven distribution of events over the study period, and further shed light on spatial variations in the importance of the two weather systems for low-NDVI events. In Chapter 4, we apply a novel framework to identify spatial objects of extremely high VPD in JJA (VPDJJA+) from the exceedance of the local 40-year return level, and decompose their intensity (I) into contributions from seasonal mean T’ and q’. After detrending seasonal mean VPD, we identify about 100 VPDJJA+ with a center of mass in 30-60°N in ERA5 reanalysis covering the 1979-2020 period, and 2’500 such VPDJJA+ in 105 × 10 years of large-ensemble CESM1 simulations. The decomposition of I in the mid-latitudes shows a high agreement between the two datasets. On average, positive T’ makes a significant contribution of ~75% to the I ≈ 0.3 kPa. However, this warm anomaly also leads to a moistening resulting from the climatological co-variability of T and q, which reduces the I. Anomalous circulation dynamics causing negative q’ during VPDJJA+ are responsible for the remaining ~31% of I. Furthermore, we reveal robust spatial variations in these three contributions in the large set of CESM1 VPDJJA+. Chapter 4 also demonstrates the consequences of the nonlinear dependence of VPD on T. First, VPD is underestimated when calculated from seasonal mean T and q compared to instantaneous values. However, this underestimation does not affect the VPD anomaly and thus the I of VPDJJA+. Second, the contribution of T’ to I is expected to increase in a warmer climate, which is confirmed by ERA5 VPDJJA+. A final analysis of six observed high-impact VPDJJA+ illustrates how individual heat waves and dry spells aggregate over a JJA season, and how periods of high VPD thereby interact with partly extreme drought conditions. In Chapter 5, we attribute the summer melt increase of the GrIS from 2005-2015 to 2085-2095 to changes in melt area, melt intensity, and atmospheric circulation. The latter is quantified using the Self-Organizing Map method to identify specific synoptic circulation patterns and changes in their occurrence frequency. We consider the ∆m ̅_s = +501 Gt JJA-1 summer melt increase in the upper-elevation zone (≥ 1’200 m), where melt estimates from the surface energy balance of our CESM2 ensemble can be considered adequate. The ∆m ̅_s can fully be attributed to melt expansion (51%) and intensification (17%), their concurrent change (24%), and an increase in the daily covariance of melt area and intensity (8%). Changes in mid-tropospheric circulation occur mainly between synoptic flow patterns with a similar melt increase and thus do not contribute to the projected melt increase. Despite its irrelevance for mean melt changes, daily atmospheric variability will become more important for the inter-annual and intra-seasonal variability of melt and hence for the temporal evolution of mass loss from the GrIS. Taken together, the results presented in the three chapters demonstrate the diversity of processes that can shape the climate system on the seasonal timescale. We present novel tools that are useful for systematically studying the impact of short-term weather variability on seasonal climate anomalies. Moreover, the thesis highlights how terrestrial ecosystems and the cryosphere are currently undergoing profound changes due to global warming. The insights gained into the forest-meteorology interaction, on the characteristics of extreme VPD seasons, and on the processes leading to the projected increase in GrIS melt not only represent progress at the weather-climate interface, but it is hoped that this thesis will also stimulate new research that benefits from the methodological advances developed here.