Search

Machine learning technology promises a more efficient and scalable approach to locating and aggregating data and information from the burgeoning scientific literature. Realizing this promise requires provision of applications, data resources, and the documentation of analytic workflows. GeoDeepDive provides a digital library comprising over 13 million peer-reviewed documents and the computing infrastructure upon which to build and deploy search and text-extraction capabilities using regular expressions and natural language processing. Here we present a model GeoDeepDive workflow and accompanying R package to show how GeoDeepDive can be employed to extract spatiotemporal information about site-level records in the geoscientific literature. We apply these capabilities to a proof-of-concept subset of papers in a case study to generate a preliminary distribution of ice-rafted debris (IRD) records in both space and time. We use regular expressions and natural language-processing utilities to extract and plot reliable latitude-longitude pairs from publications containing IRD, and also extract age estimates from those publications. This workflow and R package provides researchers from the geosciences and allied disciplines a general set of tools for querying spatiotemporal information from GeoDeepDive for their own science questions.

Climate variations in the North Atlantic region can substantially impact surrounding continents. Notably, the Younger Dryas chronozone was named for the ecosystem effects of abrupt changes in the region at circa (ca.) 12.9–11.7 ka (millennia before 1950 AD). Holocene variations since then, however, have been hard to diagnose, and the responsiveness of terrestrial ecosystems continues to be debated. Here, we show that Holocene climate variations had spatial patterns consistent with changes in Atlantic overturning and repeatedly steepened the temperature gradient between Nova Scotia and Greenland since >8 ka. The multicentury changes correlated with hydrologic and vegetation changes in the northeast United States, including when an enhanced temperature gradient coincided with subregional droughts indicated by water-level changes at multiple coastal lakes at 4.9–4.6, 4.2–3.9, 2.8–2.1, and 1.3–1.2 ka. We assessed the variability and its effects by replicating signals across sites, using converging evidence from multiple methods, and applying forward models of the systems involved. We evaluated forest responses in the northeast United States and found that they tracked the regional climate shifts including the smallest magnitude (∼5% or 50 mm) changes in effective precipitation. Although a long-term increase in effective precipitation of >45% (>400 mm) could have prevented ecological communities from equilibrating to the continuously changing conditions, our comparisons confirm stable vegetation–climate relationships and support the use of fossil pollen records for quantitative paleoclimate reconstruction. Overall, the network of records indicates that centennial climate variability has repeatedly affected the North Atlantic region with predictable consequences.

A comprehensive database of paleoclimate records is needed to place recent warming into the longer-term context of natural climate variability. We present a global compilation of quality-controlled, published, temperature-sensitive proxy records extending back 12,000 years through the Holocene. Data were compiled from 679 sites where time series cover at least 4000 years, are resolved at sub-millennial scale (median spacing of 400 years or finer) and have at least one age control point every 3000 years, with cut-off values slackened in data-sparse regions. The data derive from lake sediment (51%), marine sediment (31%), peat (11%), glacier ice (3%), and other natural archives. The database contains 1319 records, including 157 from the Southern Hemisphere. The multi-proxy database comprises paleotemperature time series based on ecological assemblages, as well as biophysical and geochemical indicators that reflect mean annual or seasonal temperatures, as encoded in the database. This database can be used to reconstruct the spatiotemporal evolution of Holocene temperature at global to regional scales, and is publicly available in Linked Paleo Data (LiPD) format., Measurement(s)climateTechnology Type(s)digital curationFactor Type(s)temporal interval • geographic location • proxy typeSample Characteristic - Environmentclimate systemSample Characteristic - LocationEarth (planet) Machine-accessible metadata file describing the reported data: https://www.ncdc.noaa.gov/paleo/study/27330

Cooling during most of the past two millennia has been widely recognized and has been inferred to be the dominant global temperature trend of the past 11,700 years (the Holocene epoch). However, long-term cooling has been difficult to reconcile with global forcing, and climate models consistently simulate long-term warming. The divergence between simulations and reconstructions emerges primarily for northern mid-latitudes, for which pronounced cooling has been inferred from marine and coastal records using multiple approaches. Here we show that temperatures reconstructed from sub-fossil pollen from 642 sites across North America and Europe closely match simulations, and that long-term warming, not cooling, defined the Holocene until around 2,000 years ago. The reconstructions indicate that evidence of long-term cooling was limited to North Atlantic records. Early Holocene temperatures on the continents were more than two degrees Celsius below those of the past two millennia, consistent with the simulated effects of remnant ice sheets in the climate model Community Climate System Model 3 (CCSM3). CCSM3 simulates increases in 'growing degree days'-a measure of the accumulated warmth above five degrees Celsius per year-of more than 300 kelvin days over the Holocene, consistent with inferences from the pollen data. It also simulates a decrease in mean summer temperatures of more than two degrees Celsius, which correlates with reconstructed marine trends and highlights the potential importance of the different subseasonal sensitivities of the records. Despite the differing trends, pollen- and marine-based reconstructions are correlated at millennial-to-centennial scales, probably in response to ice-sheet and meltwater dynamics, and to stochastic dynamics similar to the temperature variations produced by CCSM3. Although our results depend on a single source of palaeoclimatic data (pollen) and a single climate-model simulation, they reinforce the notion that climate models can adequately simulate climates for periods other than the present-day. They also demonstrate that amplified warming in recent decades increased temperatures above the mean of any century during the past 11,000 years.

We present a new oxygen isotope (δ18O) record from carbonate-rich lake sediments from central Vermont. The record from Twin Ponds spans from 13.5 cal ka BP (1950 AD) to present, but contains a 6 ka long hiatus starting shortly after 7.5 cal ka BP. We compare the record for ca. 13.5–7.5 cal ka BP with published δ18O data from the region after using a Bayesian approach to produce many possible chronologies for each site. Principal component analysis then identified chronologically-robust, multi-site oxygen isotope signals, including negative values during the Younger Dryas, but no significant deviations from the early Holocene mean of the regional records. However, differences among sites indicate significant trends that likely relate to interacting changes in the regional gradients of seasonal temperatures and precipitation as well as moisture sources, moisture pathways, and aridity that were controlled by large-scale climatic controls such as insolation, the progressive decline of the Laurentide Ice Sheet, and changes in oceanic circulation. Centennial shifts punctuate these trends at ca. 9.3 and 8.2 cal ka BP, and reveal that the local character of these short-lived features requires a detailed understanding of lake hydrology and regional isotopic gradients to yield reliable information for regional climate reconstructions.

A sequence of long-term and rapid changes during the Holocene appears in a network of 40 well-resolved paleoclimate datasets from mid-latitude North America, including records of pollen-inferred temperatures, alkenone-derived sea-surface temperatures (SSTs), lake-level changes, dust accumulation, and lake isotopes from Idaho to Maine. Statistical analyses reveal that changes in insolation and the Laurentide Ice Sheet explain 51.7% of the variance in the records, especially multi-millennial trends, but peak rates of change indicate additional rapid changes at ca. 10.8, 9.4, 8.3, 7.0, 5.5–5.2, 4.7, 2.1, and 0.9 ka. Step changes between 9.4 and 8.3 ka relate to ice sheet dynamics that warmed much of the region, and changes at 5.5 ka were the largest since the demise of the ice sheet. The shift at 5.5 ka initiated widespread cooling and increases in effective moisture, which culminated in the coolest, wettest millennia in most areas after 2.1 ka. Replicated evidence from multiple records also shows a spatially-varied set of multi-century fluctuations including 1) low temperatures and high effective moisture at 5.5–4.8 ka in the mid-continent and 2) repeated phases of low SSTs, cool summers, and drought superimposed upon long cooling, moistening trends in eastern North American since 5.5 ka.

Shifts in an ecosystem’s state can alter biogeochemical cycling and the extent of nutrient conservation within a terrestrial landscape on multiple time scales. Transient biogeochemical changes may follow disturbance and succession, although persistent long-term differences may exist under different climates and vegetation types. We evaluate the potential for such biogeochemical changes in the context of long-term ecological history by measuring the nitrogen isotope composition of organic matter in a lake sediment core. We targeted Little Windy Hill Pond (LWH) in the Medicine Bow Mountains, Wyoming because reconstructions of the lake level, fire, and vegetation histories from the lacustrine sediments indicated a century-scale transformation from an arid, shrub-dominated landscape to a sub-alpine, tree-dominated ecosystem with extensive woody cover and large, live biomass pools. We demonstrate that the afforestation at the beginning of the Holocene transformed the Artemisia-dominated ecosystem, which had persisted for millennia during the Pleistocene. The changes affected nitrogen cycling dynamics, especially through intensified nutrient conservation when live biomass pools increased with greater woody cover. The LWH sediments record a baseline δ15N shift from 2.2–3.0 to 0.3–2.0‰ as less 15N-enriched organic matter accumulated in the lake. We also observed a transient pattern of maximum nutrient conservation and minimum δ15N values as terrestrial biomass increased during the aggradation (~175 years) and transition phases of ecological succession. Our nitrogen isotope results support theoretical expectations of long-term biogeochemical dynamics as nutrient conservation increases during afforestation.

Mesic tree species such as Fagus grandifolia and Tsuga canadensis experienced multiple abundance declines in eastern North America during the last 8000 years, but the causes remain unclear. This paper presents a new sub-centennial record of Holocene vegetation, fire and sedimentological changes at Spicer Lake, IN, to test hypotheses about the role of fire and hydrological variations on shifts in vegetation composition. Four pollen zones are reported: Abies–Picea forests (15–11.8 ka BP), Pinus-dominated mixed forest (11.8–10.6 ka BP), transitional mixed forest (10.6–6.8 ka BP), and deciduous forest characterized by the expansion and high variability of F. grandifolia (after 6.8 ka BP). Macroscopic charcoal indicates five to seven fires between 6.1 and 4.4 ka BP and no fires between 4.4 and 2 ka BP, despite several large declines in F. grandifolia, and more fires after 1.8 ka BP likely linked to declining F. grandifolia abundances after 1.1 ka BP. Six peaks in mineralogenic sediments are suggestive of hydroclimate variability, but do not consistently correspond to shifts in F. grandifolia abundances. A Bayesian change-point analysis of 15 regional F. grandifolia pollen records identifies peak probabilities of events at 4.8 and 1.1 ka BP, similar in timing to variations in T. canadensis at other sites. Hence, fire can be ruled out as a driver of the mid-Holocene declines of F. grandifolia, but more work is needed to confidently establish the regional timing of F. grandifolia declines and to link them to past droughts and T. canadensis declines in eastern North America.

During the last deglaciation temperatures over midcontinental North America warmed dramatically through the Bølling-Allerød, underwent a cool period associated with the Younger-Dryas and then reverted to warmer, near modern temperatures during the early Holocene. However, paleo proxy records of the hydroclimate of this period have presented divergent evidence. We reconstruct summer relative humidity (RH) across the last deglacial period using a mechanistic model of cellulose and leaf water δ18O and δD combined with a pollen-based temperature proxy to interpret stable isotopes of sub-fossil wood. Midcontinental RH was similar to modern conditions during the Last Glacial Maximum, progressively increased during the Bølling-Allerød, peaked during the Younger-Dryas, and declined sharply during the early Holocene. This RH record suggests deglacial summers were cooler and characterized by greater advection of moisture-laden air-masses from the Gulf of Mexico and subsequent entrainment over the mid-continent by a high-pressure system over the Laurentide ice sheet. These patterns help explain the formation of dark-colored cumulic horizons in many Great Plains paleosol sequences and the development of no-analog vegetation types common to the Midwest during the last deglacial period. Likewise, reduced early Holocene RH and precipitation correspond with a diminished glacial high-pressure system during the latter stages of ice-sheet collapse.

Geophysical and sedimentary records from five lakes in Massachusetts reveal regionally coherent hydrologic variability during the Holocene. All of the lakes have risen since ~9.0 ka, but multicentury droughts after 5.6 ka repeatedly lowered their water levels. Quantified water level histories from the three best-studied lakes share >70% of their reconstructed variance. Four prominent low-water phases at 4.9–4.6, 4.2–3.9, 2.9–2.1, and 1.3–1.2 ka were synchronous across coastal lakes, even after accounting for age uncertainties. The droughts also affected sites up to ~200 km inland, but water level changes at 5.6–4.9 ka appear out of phase between inland and coastal lakes. During the enhanced multicentury variability after ~5.6 ka, droughts coincided with cooling in Greenland and may indicate circulation changes across the North Atlantic region. Overall, the records demonstrate that current water levels are exceptionally high and confirm the sensitivity of water resources in the northeast U.S. to climate change.

Many ecosystem processes that influence Earth system feedbacks, including vegetation growth, water and nutrient cycling, and disturbance regimes, are strongly influenced by multi-decadal to millennial-scale variations in climate that cannot be captured by instrumental climate observations. Paleoclimate information is therefore essential for understanding contemporary ecosystems and their potential trajectories under a variety of future climate conditions. With the exception of fossil pollen records, there are a limited number of northeastern US (NE US) paleoclimate archives that can provide constraints on its temperature and hydroclimate history. Moreover, the records that do exist have not been considered together. Tree-ring data indicate that the 20th century was one of the wettest of the past 500 years in the eastern US (Pederson et al., 2014), and lake-level records suggest it was one of the wettest in the Holocene (Newby et al., 2014); how such results compare with other available data remains unclear, however. Here we conduct a systematic review, assessment, and comparison of paleotemperature and paleohydrological proxies from the NE US for the last 3000 years. Regional temperature reconstructions are consistent with the long-term cooling trend (1000 BCE–1700 CE) evident in hemispheric-scale reconstructions, but hydroclimate reconstructions reveal new information, including an abrupt transition from wet to dry conditions around 550–750 CE. NE US paleo data suggest that conditions during the Medieval Climate Anomaly were warmer and drier than during the Little Ice Age, and drier than today. There is some evidence for an acceleration over the past century of a longer-term wetting trend in the NE US, and coupled with the abrupt shift from a cooling trend to a warming trend from increased greenhouse gases, may have wide-ranging implications for species distributions, ecosystem dynamics, and extreme weather events. More work is needed to gather paleoclimate data in the NE US, make inter-proxy comparisons, and improve estimates of uncertainty in the reconstructions.

A decline of hemlock (Tsuga) populations at ca 5.5 ka (thousands of calibrated radiocarbon years before 1950 AD) stands out as the most abrupt vegetation change of the Holocene in North America, but remains poorly understood after decades of study. Recent analyses of fossil pollen have revealed a concurrent, abrupt oak (Quercus) decline and increases in the abundance of beech (Fagus) and pine (Pinus) on Cape Cod in eastern Massachusetts, but the replacement of drought-tolerant oaks by moisture-sensitive beeches appears inconsistent with low lake levels in the region at the same time. The oak and beech changes are also limited to coastal areas, and the coastal-inland differences require an explanation. Here, we develop a new lake-level reconstruction from Deep Pond, Cape Cod by using a transect of sediment cores and ground-penetrating radar (GPR) profiles to constrain the past elevations of the sandy, littoral zone of the pond. The reconstruction shows that a series of multi-century episodes of low water coincide with the abrupt hemlock and oak declines, and interrupt subsequent phases of hemlock recovery. The lake-level variations equal precipitation deficits of ∼100 mm superimposed on a Holocene long moisture increase of >400 mm. However, because moisture deficits do not easily explain the oak and beech changes, we also evaluate how the climate preferences of the regional vegetation changed over time by matching the fossil pollen assemblages from Deep Pond with their modern equivalents. Reconstructions of the precipitation requirements of the vegetation correlate well even in detail with the lake-level record (r = 0.88 at Deep Pond), and indicate close tracking of effective moisture (precipitation minus evapotranspiration) by the vegetation despite the abrupt species declines, which could have decoupled climate and vegetation trends. Reconstructions of the temperature preferences of the vegetation indicate that coastal sites may have cooled by 0.5–2.5 °C after ca 5.5 ka, while inland sites warmed by 0.5–1 °C. The change in coastal temperature preferences agrees with sea surface cooling in the western Atlantic Ocean of >1 °C. Consequently, the persistence of low hemlock abundance after 5.5 ka in the northeast U.S. may have resulted from oceanic changes that produced multi-century droughts and thus delayed the post-decline recovery of hemlock populations.