Low Interannual Variability Research Articles

X-BASE: the first terrestrial carbon and water flux products from an extended data-driven scaling framework, FLUXCOM-X

Abstract. Mapping in situ eddy covariance measurements of terrestrial land–atmosphere fluxes to the globe is a key method for diagnosing the Earth system from a data-driven perspective. We describe the first global products (called X-BASE) from a newly implemented upscaling framework, FLUXCOM-X, representing an advancement from the previous generation of FLUXCOM products in terms of flexibility and technical capabilities. The X-BASE products are comprised of estimates of CO2 net ecosystem exchange (NEE), gross primary productivity (GPP), evapotranspiration (ET), and for the first time a novel, fully data-driven global transpiration product (ETT), at high spatial (0.05°) and temporal (hourly) resolution. X-BASE estimates the global NEE at −5.75 ± 0.33 Pg C yr−1 for the period 2001–2020, showing a much higher consistency with independent atmospheric carbon cycle constraints compared to the previous versions of FLUXCOM. The improvement of global NEE was likely only possible thanks to the international effort to increase the precision and consistency of eddy covariance collection and processing pipelines, as well as to the extension of the measurements to more site years resulting in a wider coverage of bioclimatic conditions. However, X-BASE global net ecosystem exchange shows a very low interannual variability, which is common to state-of-the-art data-driven flux products and remains a scientific challenge. With 125 ± 2.1 Pg C yr−1 for the same period, X-BASE GPP is slightly higher than previous FLUXCOM estimates, mostly in temperate and boreal areas. X-BASE evapotranspiration amounts to 74.7×103 ± 0.9×103 km3 globally for the years 2001–2020 but exceeds precipitation in many dry areas, likely indicating overestimation in these regions. On average 57 % of evapotranspiration is estimated to be transpiration, in good agreement with isotope-based approaches, but higher than estimates from many land surface models. Despite considerable improvements to the previous upscaling products, many further opportunities for development exist. Pathways of exploration include methodological choices in the selection and processing of eddy covariance and satellite observations, their ingestion into the framework, and the configuration of machine learning methods. For this, the new FLUXCOM-X framework was specifically designed to have the necessary flexibility to experiment, diagnose, and converge to more accurate global flux estimates.

Effect of Solar Irradiation Inter-Annual Variability on PV and CSP Power Plants Production Capacity: Portugal Case-Study

The sizing of solar energy power plants is usually made using typical meteorological years, which disregards the inter-annual variability of the solar resource. Nevertheless, such variability is crucial for the bankability of these projects because it impacts on the production goals set at the time of the supply agreement. For that reason, this study aims to fill the gap in the existing literature and analyse the impact that solar resource variability has on solar power plant production as applied to the case of Portugal (southern Europe). To that end, 17 years (2003–2019) of meteorological data from a network of 90 ground stations hosted by the Portuguese Meteorological Service is examined. Annual capacity factor regarding photovoltaic (PV) and concentrating solar power (CSP) plants is computed using the System Advisor Model, used here for solar power performance simulations. In terms of results, while a long-term trend for increase in annual irradiation is found for Global Horizontal Irradiance (GHI) and Direct Normal Irradiance (DNI), 0.4148 and 3.2711 kWh/m2/year, respectively, consistent with a solar brightening period, no corresponding trend is found for PV or CSP production. The latter is attributed to the long-term upward trend of 0.0231 °C/year in annual average ambient temperature, which contributes to PV and CSP efficiency reduction. Spatial analysis of inter-annual relative variability for GHI and DNI shows a reduction in variability from the north to the south of the country, as well as for the respective power plant productions. Particularly, for PV, inter-annual variability ranges between 2.45% and 12.07% in Faro and Santarém, respectively, while higher values are generally found for CSP, 3.71% in Faro and 16.04% in São Pedro de Moel. These results are a contribution to future instalments of PV and CSP systems in southern Portugal, a region with very favourable conditions for solar energy harvesting, due to the combination of high production capacity and low inter-annual variability.

Cross-continental comparison of plant reproductive phenology shows high intraspecific variation in temperature sensitivity

Abstract The success of plant species under climate change will be determined, in part, by their phenological responses to temperature. Despite the growing need to forecast such outcomes across entire species ranges, it remains unclear how phenological sensitivity to temperature might vary across individuals of the same species. In this study, we harnessed community science data to document intraspecific patterns in phenological temperature sensitivity across the multicontinental range of six herbaceous plant species. Using linear models, we correlated georeferenced temperature data with 23 220 plant phenological records from iNaturalist to generate spatially explicit estimates of phenological temperature sensitivity across the shared range of species. We additionally evaluated the geographic association between local historic climate conditions (i.e. mean annual temperature [MAT] and interannual variability in temperature) and the temperature sensitivity of plants. We found that plant temperature sensitivity varied substantially at both the interspecific and intraspecific levels, demonstrating that phenological responses to climate change have the potential to vary both within and among species. Additionally, we provide evidence for a strong geographic association between plant temperature sensitivity and local historic climate conditions. Plants were more sensitive to temperature in hotter climates (i.e. regions with high MAT), but only in regions with high interannual temperature variability. In regions with low interannual temperature variability, plants displayed universally weak sensitivity to temperature, regardless of baseline annual temperature. This evidence suggests that pheno-climatic forecasts may be improved by accounting for intraspecific variation in phenological temperature sensitivity. Broad climatic factors such as MAT and interannual temperature variability likely serve as useful predictors for estimating temperature sensitivity across species’ ranges.

Impact of groundwater representation on heat events in regional climate simulations over Europe

Abstract. The representation of groundwater is simplified in most regional climate models (RCMs), potentially leading to biases in the simulations. This study introduces a unique dataset from the regional Terrestrial Systems Modelling Platform (TSMP) driven by the Max Planck Institute Earth System Model at Low Resolution (MPI-ESM-LR) boundary conditions in the context of dynamical downscaling of global climate models (GCMs) for climate change studies. TSMP explicitly simulates full 3D soil and groundwater dynamics together with overland flow, including complete water and energy cycles from the bedrock to the top of the atmosphere. By comparing the statistics of heat events, i.e., a series of consecutive days with a near-surface temperature exceeding the 90th percentile of the reference period, from TSMP and those from GCM–RCM simulations with simplified groundwater dynamics from the COordinated Regional Climate Downscaling EXperiment (CORDEX) for the European domain, we aim to improve the understanding of how groundwater representation affects heat events in Europe. The analysis was carried out using RCM outputs for the summer seasons of 1976–2005 relative to the reference period of 1961–1990. While our results show that TSMP simulates heat events consistently with the CORDEX ensemble, there are some systematic differences that we attribute to the more realistic representation of groundwater in TSMP. Compared to the CORDEX ensemble, TSMP simulates fewer hot days (i.e., days with a near-surface temperature exceeding the 90th percentile of the reference period) and lower interannual variability and decadal change in the number of hot days on average over Europe. TSMP systematically simulates fewer heat waves (i.e., heat events lasting 6 d or more) compared to the CORDEX ensemble; moreover, they are shorter and less intense. The Iberian Peninsula is particularly sensitive with respect to groundwater. Therefore, incorporating an explicit 3D groundwater representation in RCMs may be a key in reducing biases in simulated duration, intensity, and frequency of heat waves in Europe. The results highlight the importance of hydrological processes for the long-term regional climate simulations and provide indications of possible potential implications for climate change projections.

Nonlinear responses in interannual variability of lake ice to climate change

AbstractClimate change is contributing to rapid changes in lake ice cover across the Northern Hemisphere, thereby impacting local communities and ecosystems. Using lake ice cover time‐series spanning over 87 yr for 43 lakes across the Northern Hemisphere, we found that the interannual variability in ice duration, measured as standard deviation, significantly increased in only half of our studied lakes. We observed that the interannual variability in ice duration peaked when lakes were, on average, covered by ice for about 1 month, while both longer and shorter long‐term mean ice cover duration resulted in lower interannual variability in ice duration. These results demonstrate that the ice cover duration can become so short that the interannual variability rapidly declines. The interannual variability in ice duration showed a strong dependency on global temperature anomalies and teleconnections, such as the North Atlantic Oscillation and El Niño–Southern Oscillation. We conclude that many lakes across the Northern Hemisphere will experience a decline in interannual ice cover variability and shift to open water during the winter under a continued global warming trend which will affect lake biological, cultural, and economic processes.

Analysis of the solar radiation/atmosphere interaction at a Mediterranean site: The role of clouds

This study is the last of a series of three research papers analysing the solar radiation and its interaction with the atmospheric components spanning an eleven-year period (2008–2018) at a mid-latitude urban site in the Mediterranean basin. During the previous works a detailed characterization of the photosynthetically active radiation (PAR; 400–700 nm), as well as analysis the aerosol and clouds effects on PAR was carried out. This research work addresses an analysis of solar radiation on the total solar irradiance spectrum (TSI; 280–3000 nm) in terms of the effects of different atmospheric conditions on it, considered by the clearness index and the cloud cover, bringing very valuable findings from the long-term comparative analysis of radiative effects of clouds on TSI and PAR wavelength ranges. The average values in the entire period of the study for the global solar radiation in the total solar irradiance (TSIGlobal) found to be 450 ± 100 Wm−2 and 530 ± 110 Wm−2 under all and clear- sky conditions, respectively. Meanwhile, the average values for the diffuse component (TSIDiffuse) are 141 ± 21 Wm−2 and 130 ± 21 Wm−2 (all and clear skies, respectively), with a relatively low interannual variation up to 11% for both global and diffuse TSI, as well as for both sky conditions. Analysis on the total cloud cover (TCC) shows that the clearness index is not a good parameter to discriminate between all and clear sky conditions, since there is a marked overlap in the ranges of the kt values for the different categories of TCC. Additionally, the cloud radiative forcing (CRF) are computed as the difference in solar radiation measured under all and clear sky conditions. A high seasonal variability is found for CRF, where CRFTSI,Global ranges between −37.6 Wm−2 and -137.4 Wm−2, while CRFTSI,Diffuse from 4.4 Wm−2 to 22.6 Wm−2. The positive sign implies increase in solar radiation at the surface, while the negative one implies the opposite, i.e., less availability of solar radiation on the surface. Finally, the analysis of the annual evolution of CRF reveals a downward trend on CRFTSI and CRFPAR, for both global and diffuse This relevant finding implies that clouds are exerting less cooling effects over time at this Mediterranean site.

Taxonomic and functional structure of understorey bird assemblages on Amazonian seasonally flooded river islands

Abstract The seasonal inundation of the floodplains of the Amazonian rivers is a natural and strong environmental filter for biota. Its effects on the taxonomic and functional attributes of assemblages, however, remain poorly understood. We investigated the taxonomic and trait diversity of understorey bird assemblages on fluvial islands subject to dramatic seasonal flooding in the Amazon. Birds were captured using mist-nets on ten fluvial islands and at 17 sites in unflooded terra firme forests. We captured 1552 individuals of 116 bird species in 4144 net-hours. Average capture rates were similar between the forest types, but species diversity was higher in terra firme forests. The bird assemblages of the islands showed higher species dominance, with only three species accounting for 47% of the individuals captured. Functional space was occupied similarly in both bird assemblages, as indicated by functional dispersion indices (Fdis). In contrast, functional evenness (Feve) was lower in the river island assemblages indicating that the dominant species share similar traits. Functional divergence indices (Fdiv) showed substantial niche differentiation between the bird assemblages of the two forest types. Low inter-annual variability in avian taxonomic and trait diversity was observed in river islands, probably associated with the predictability of the flooding regime.

Global climate-change trends detected in indicators of ocean ecology

Strong natural variability has been thought to mask possible climate-change-driven trends in phytoplankton populations from Earth-observing satellites. More than 30 years of continuous data were thought to be needed to detect a trend driven by climate change1. Here we show that climate-change trends emerge more rapidly in ocean colour (remote-sensing reflectance, Rrs), because Rrs is multivariate and some wavebands have low interannual variability. We analyse a 20-year Rrs time series from the Moderate Resolution Imaging Spectroradiometer (MODIS) aboard the Aqua satellite, and find significant trends in Rrs for 40% of the global surface ocean. The climate-change signal in Rrs emerges after 20 years in similar regions covering a similar fraction of the ocean in a state-of-the-art ecosystem model2, which suggests that our observed trends indicate shifts in ocean colour—and, by extension, in surface-ocean ecosystems—that are driven by climate change. On the whole, low-latitude oceans have become greener in the past 20 years.

Persistent decadal differences in plant communities assembled under contrasting climate conditions.

Plant community assembly outcomes can be contingent upon establishment year (year effects) due to variations in the environment. Stochastic events such as interannual variability in climate, particularly in the first year of community assembly, contribute to unpredictable community outcomes over the short term, but less is known about whether year effects produce transient or persistent states on a decadal timescale. To test for short-term (5-year) and persistent (decadal) effects of establishment year climate on community assembly outcomes, we restored prairie in an agricultural field using the same methods in four different years (2010, 2012, 2014, and 2016) that captured a wide range of initial (planting) year climate conditions. Species composition was measured for 5 years in all four restored prairies and for 9 and 11 years in the two oldest restored prairies established under average precipitation and extreme drought conditions. The composition of the four assembled communities showed large and significant differences in the first year of restoration, followed by dynamic change over time along a similar trajectory due to a temporary flush of annual volunteer species. Sown perennial species eventually came to dominate all communities, but communities remained distinct from each other in year five. Precipitation in June and July of the establishment year explained short-term coarse community metrics (i.e., species richness and grass/forb cover), with wet establishment years resulting in a higher cover of grasses and dry establishment years resulting in a higher cover of forbs in restored communities. Short-term differences in community composition, species richness, and grass/forb cover in restorations established under average precipitation and drought conditions persisted for 9-11 years, with low interannual variability in the composition of each prairie over the long term, indicating persistently different states on a decadal timescale. Thus, year effects resulting from stochastic variation in climate can have decadal effects on community assembly outcomes.

Long-term climatological and ecohydrological analysis of a paired catchment – flux tower observatory near Dresden (Germany). Is there evidence of climate change in local evapotranspiration?

Water budgets and climate are related in many ways and at all scales. Therefore, we expect climate change to trigger changes in all water budget components at any scale. For Central Europe observed and projected climate change indicates higher variability of precipitation, while evapotranspiration (ET) should increase due to higher temperatures, yielding lower and more variable infiltration and runoff. However, evidence in ET records is limited, as long-term measurements of ET are methodologically challenging and as factors other than climate are changing in parallel, like vegetation and land use.In this study, we take advantage of long-term hydro-meteorological data from the small research catchment Wernersbach (4.6 km2, dominated by Norway spruce) in operation since 1967 and from two eddy-covariance (EC) flux towers, all located in the Tharandt Forest, Germany. The tower DE-Tha is located a few kilometres east of the catchment, is spruce dominated and in operation since 1996. After a wind break of a spruce stand (situated inside the catchment) and planting of deciduous oaks, the tower DE-Hzd was set up in 2009. For the first time, we report systematically about observation, correction methods and metadata of the long data series of the observatory, represented by the Wernersbach catchment and the EC flux towers.Climate change signals in the region are mirrored in the Tharandt Forest records. They show rising air temperature with a breakpoint around 1988 and complex changes in solar radiation associated to a regional peak in air pollution around the same time. The catchment and both towers did not show any systematic differences in climate or meteorological data, allowing us to address observed changes in the water budget components as related to (i) climate change, (ii) change in vegetation, and (iii) different responses due to different soil and hydrogeological characteristics as well as methodological aspects. The catchment term ET plus storage, derived from precipitation minus runoff, showed the expected high variability with a significant increase over the more than 50 years of operation. The flux-tower DE-Tha showed much lower inter-annual variability in ET with an average annual total of 486 mm (1997 to 2019), but no significant trend. For the same period, average catchment ET was 734 mm/yr. The younger flux-tower DE-Hzd showed ET values in between, closer to catchment ET at the very dry end of the ten-year record (2010 to 2019). An analysis of decadal trends in a Budyko framework at catchment level revealed the dominating response of ET to land use or vegetation change until around 1990. The climate induced change of ET increased in the last decades, on the one hand directly due to an increased atmospheric demand. On the other hand, extreme weather events exerted harmful effects on vegetation, especially triggered by two dry years at the end of the record.Furthermore, we found that the mean annual tower ET was about 250 mm lower than catchment ET despite the careful correction for energy balance closure. We attribute this difference to soil and to a lesser extend to vegetation characteristics, but also to methodological uncertainties. There is evidence from interception and transpiration measurements at the flux tower as well as from water budget modelling that a major contribution of this difference is related to an insufficient EC closure correction during interception events. A careful consideration of rain events and evaporation from interception is recommended when addressing ET of similar evergreen forests in a humid climate, as EC records might be generally too low. This illustrates the necessity of redundant and complementary measurements when dealing with large system complexity.

Spatial and temporal variability of phytoplankton photophysiology in the Atlantic Southern Ocean

Active chlorophyll-a fluorescence was measured during five summer research cruises (2008 – 2016), spanning the Atlantic sector of the Southern Ocean. This unique data set provides information for assessing zonal, inter-annual and intra-seasonal variability (early versus late summer) of photosynthetic efficiency (Fv/Fm). The zonal variability of Fv/Fm showed a typical latitudinal decline from a maximum in the Polar Frontal Zone (PFZ) (0.24±0.03) to a minimum in the Southern Antarctic Circumpolar Current Zone (SACCZ) (0.18±0.07). The inter-annual variability in Fv/Fm (between each cruise) was the highest in the SACCZ, while the Antarctic Zone (AZ) exhibited low inter-annual variability. Intra-seasonal variability between the zones was limited to a significantly higher mean Fv/Fm in the PFZ and AZ in early summer compared to late summer. Intra-seasonal variability between the cruises was, however, inconsistent as higher mean Fv/Fm in early summer were seen during some years as opposed to others. Ancillary physical and biogeochemical parameters were also assessed to investigate potential direct and indirect drivers of co-variability with Fv/Fm through a series of statistical t-tests, where significant differences in Fv/Fm were used as focus points to interrogate the plausibility of co-variance. Inter-zonal variability of surface seawater temperature (SST) and Silicate:Phosphate (Si:P) ratios were highlighted as co-varying with Fv/Fm in all zones, whilst community structure played an indirect role in some instances. Similarly, inter-annual variability in Fv/Fm co-varied with SST, Nitrate:Phosphate (N:P) and Si:P ratios in the PFZ, AZ and SACCZ, while community structure influenced inter-annual variability in the PFZ and SACCZ. Intra-seasonal variability in Fv/Fm was linked to all the ancillary parameters, except community structure in the AZ, whilst different ancillary parameters dominated differences during each of the cruises. These results were further scrutinized with a Principal Component Analysis for a subset of co-located data points, where N:P and Si:P ratios emerged as the principal indirect drivers of Fv/Fm variability. This study highlights the scope for using Fv/Fm to reflect the net response of phytoplankton photophysiology to environmental adjustments and accentuates the complex interplay of different physical and biogeochemical parameters that act simultaneously and oftentimes antagonistically, influencing inter-zonal, inter-annual and intra-seasonal variability of Fv/Fm.

Cyanobacterial Bloom Phenology in Green Bay Using MERIS Satellite Data and Comparisons with Western Lake Erie and Saginaw Bay

Cyanobacteria blooms have been reported to be increasing worldwide. In addition to potentially causing major economic and ecological damage, these blooms can threaten human health. Furthermore, these blooms can be exacerbated by a warming climate. One approach to monitoring and modeling cyanobacterial biomass is to use processed satellite imagery to obtain long-term data sets. In this paper, an existing algorithm for estimating cyanobacterial biomass previously developed for MERIS is validated for Green Bay using cyanobacteria biovolume estimates obtained from field samples. Once the algorithm was validated, the existing MERIS imagery was used to determine the bloom phenology of the cyanobacterial biomass in Green Bay. Modeled datasets of heat flux (as a proxy for stratification), wind speed, water temperature, and gelbstoff absorption along with in situ river discharge data were used to separate bloom seasons in Green Bay from bloom seasons in nearby cyanobacteria bloom hotspots including western Lake Erie and Saginaw Bay. Of the ten-year MERIS dataset used here, the highest five years were considered “high bloom” years, and the lowest five years from biomass were considered “low bloom” years and these definitions were used to separate Green Bay. Green Bay had a strong relationship with gelbstoff absorption making it unique among the water bodies, while western Lake Erie responded strongly with river discharge as previously reported. Saginaw Bay, which has low interannual bloom variability, did not exhibit a largely influential single parameter.

Temporal and Spatial Variations in Black Carbon Aerosol in Different Atmospheric Background Stations in China from 2006 to 2020

Black carbon (BC) aerosol emissions are complex and have important environmental and meteorological effects. In China, the temporal and spatial variations in BC in different atmospheric environmental conditions need to be fully understood. Based on the long-term observational BC data in seven atmospheric background stations in China from 2006 to 2020, combined with meteorological data, emission source data, enhanced vegetation index (EVI) data, and aerosol optical depth (AOD) data, we comprehensively analyzed the characteristics of temporal and spatial variations, long-term evolution, and influencing factors of BC in China. The results showed that the BC and AOD values of different atmospheric environments in China were quite different, and BC positively contributed to AOD. The spatial distribution was high in the east and low in the west owing to the differences in emission sources and meteorological conditions. The ρ(BC) and AOD values were higher to the east of the "Hu Huanyong" line, such as at the Mt. Longfeng, Shangdianzi, Lin'an, and Jinsha stations, where the average values were (1699±2213)-(3392±2131) ng·m-3 and 0.36±0.32-0.72±0.37, respectively. These values were lower to the west of the "Hu Huanyong" line, such as at the Akedala, Mt. Waliguan, and Shangri-La stations, where the average values were (287±226)-(398±308) ng·m-3 and 0.20±0.13-0.22±0.19, respectively. The interannual variability in BC included differences between different atmospheric background stations, which could be divided into four categories:low interannual variability, such as at the Akedala station; an initial increase followed by a decrease and subsequent stabilization, such as at the Mt. Waliguan station; an initial decrease followed by stabilization, such as at the Shangri-La station; and an initial stabilization followed by a decrease, such as at the Mt. Longfeng, Shangdianzi, Jinsha, and Lin'an stations. Seasonal variations in BC included differences in different atmospheric background stations. The BC mass concentrations were lowest in autumn and higher in winter and spring west of the "Hu Huanyong" line and were highest in winter and lowest in summer east of the "Hu Huanyong" line. BC contributed to the AOD being larger in all stations in the spring and summer and contributed less at the stations west of the "Hu Huanyong" line in autumn and the stations east of the "Hu Huanyong" line in winter. The diurnal variations in BC were mainly bimodally distributed in the different atmospheric background stations, but the peak times varied in different stations and seasons.

Contrasting Sea‐Air CO2 Exchanges in the Western Tropical Atlantic Ocean

AbstractThe western Tropical Atlantic Ocean is a biogeochemically complex region due to the structure of the surface current system and the large freshwater input from the Amazon River coupled with the dynamics of precipitation. Such features make it difficult to understand the dynamics of the carbon cycle, leading to contrasting estimates in sea‐air CO2 exchanges in this region. Here, we demonstrate that these contrasting estimates occur because the western Tropical Atlantic Ocean can be split into three distinct sub‐regions in terms of the sea‐air CO2 exchanges. The sub‐region under the North Brazil Current domain acts as a weak annual CO2 source to the atmosphere, with low interannual variability. The sub‐region under the North Equatorial Current influence acts as an annual CO2 sink, with great temporal variability. The third sub‐region under the Amazon River plume influence shows greater interannual variability of CO2 exchanges, but it always acts as a net oceanic sink for CO2. Despite this large spatial variability, the entire region acts as a net annual CO2 sink of −1.6 ± 1.0 mmol m−2 day−1. Importantly, the Amazon River plume waters drive 87% of the CO2 uptake in the western Tropical Atlantic Ocean. In addition, we found a significant increasing trend in sea surface CO2 partial pressure in the North Brazil Current and North Equatorial Current waters. Such trends are more pronounced than the increase in atmospheric CO2 partial pressure, revealing the sensitivity of carbon dynamics in these sub‐regions to global climate change.

Seasonal forecasting of lake water quality and algal bloom risk using a continuous Gaussian Bayesian network

Abstract. Freshwater management is challenging, and advance warning that poor water quality was likely, a season ahead, could allow for preventative measures to be put in place. To this end, we developed a Bayesian network (BN) for seasonal lake water quality prediction. BNs have become popular in recent years, but the vast majority are discrete. Here, we developed a Gaussian Bayesian network (GBN), a simple class of continuous BN. The aim was to forecast, in spring, mean total phosphorus (TP) and chlorophyll a (chl a) concentration, mean water colour, and maximum cyanobacteria biovolume for the upcoming growing season (May–October) in Vansjø, a shallow nutrient-rich lake in southeastern Norway. To develop the model, we first identified controls on interannual variability in seasonally aggregated water quality. These variables were then included in a GBN, and conditional probability densities were fit using observations (≤39 years). GBN predictions had R2 values of 0.37 (chl a) to 0.75 (colour) and classification errors of 32 % (TP) to 17 % (cyanobacteria). For all but lake colour, including weather variables did not improve the predictive performance (assessed through cross-validation). Overall, we found the GBN approach to be well suited to seasonal water quality forecasting. It was straightforward to produce probabilistic predictions, including the probability of exceeding management-relevant thresholds. The GBN could be sensibly parameterised using only the observed data, despite the small dataset. Developing a comparable discrete BN was much more subjective and time-consuming. Although low interannual variability and high temporal autocorrelation in the study lake meant the GBN performed only slightly better than a seasonal naïve forecast (where the forecasted value is simply the value observed the previous growing season), we believe that the forecasting approach presented here could be particularly useful in areas with higher sensitivity to catchment nutrient delivery and seasonal climate and for forecasting at shorter (daily or monthly) timescales. Despite the parametric constraints of GBNs, their simplicity, together with the relative accessibility of BN software with GBN handling, means they are a good first choice for BN development with continuous variables.

Senescence in temperate broadleaf trees exhibits species-specific dependence on photoperiod versus thermal forcing

Incomplete understanding of the processes controlling senescence limits our ability to forecast how the timing of leaf senescence will change in coming decades. In this study, we use a hierarchical Bayesian model (HBM) in association with a 27+ year record of field observations for 12 temperate deciduous tree species collected at Harvard Forest in central Massachusetts to examine how variability in bioclimatic controls affects the timing of leaf senescence. To test how general and extensible our results are over a broader biogeographic range, we used a multi-year record of land surface phenology derived from remote sensing encompassing all forested lands in New England. Results from the HBM showed that while air temperature is an important factor that influences the timing of leaf senescence, photoperiod uniformly exerts the strongest control across all 12 species. Species exhibiting the strongest dependence on photoperiod, particularly Acer species, showed low inter-annual variation and no long-term trends in the timing of leaf senescence. In contrast, species with greater dependence on air temperature, particularly Quercus species, showed statistically significant trends toward later senescence dates in response to long-term warming. Results from analyses conducted at regional scale across all of New England using data derived from remote sensing corroborated results obtained at Harvard Forest. Specifically, relative to ecoregions dominated by Quercus species, the timing of leaf senescence in ecoregions dominated by Acer species exhibited lower interannual variability and lower correlation with year-to-year variation in pre-senescence period mean air temperatures. These results suggest that forecasting how the timing of leaf senescence in temperate forests will change in the future requires species-specific understanding of how bioclimatic forcing controls the timing of leaf senescence.

Deficiencies of Phenology Models in Simulating Spatial and Temporal Variations in Temperate Spring Leaf Phenology

AbstractSpring leaf phenology and its response to climate change have crucial effects on surface albedo, carbon balance, and the water cycle of terrestrial ecosystems. Based on long‐term (period 1963–2014) in situ observations of budburst date and leaf unfolding date of more than 300 deciduous woody species from 32 sites across the temperate zone in China, we conducted model‐data comparison of spatial and temporal variations for spring leaf phenology calculated using the phenology modules that were embed into 10 existing terrestrial ecosystem models. Our results suggested that ORganizing Carbon and Hydrology in Dynamic EcosystEms and Spatially Explicit Individual‐Based performed the best in reproducing the spatial patterns of spring leaf phenology, but tended to underestimate the temporal variations in responding to temperature warming, showing low interannual variability (IAV) and temperature sensitivity (ST). In contrast, the performances of Vegetation Integrated SImulator for Trace Gases were the best in modeling IAV and ST. BIOME3, Lund‐Potsdam‐Jena model, Joint UK Land Environment Simulator, BioGeochemical Cycles, Community Land Model, Integrated Biosphere Simulator, and Commonwealth Scientific and Industrial Research Organisation Atmosphere Biosphere Land Exchange Model failed to reproduce both the spatial and temporal patterns. Using temperature series (1960–2100) form Coupled Model Intercomparison Project Number 6 scenarios to force the 10 phenology modules, our results highlighted large uncertainties in predicting spring leaf phenology changes with the warming climate, and more work is required to deal with the deficiencies of phenology model parameters and algorithms.

Historical simulation of maize water footprints with a new global gridded crop model ACEA

Abstract. Crop water productivity is a key element of water and food security in the world and can be quantified by the water footprint (WF). Previous studies have looked at the spatially explicit distribution of crop WFs, but little is known about their temporal dynamics. Here, we present AquaCrop-Earth@lternatives (ACEA), a new process-based global gridded crop model that can simulate three consumptive WF components: green (WFg), blue from irrigation (WFbi), and blue from capillary rise (WFbc). The model is applied to analyse global maize production in 1986–2016 at 5×5 arcmin spatial resolution. Our results show that over the 2012–2016 period, the global average unit WF of maize is 728.0 m3 t−1 yr−1 (91.2 % WFg, 7.6 % WFbi, and 1.2 % WFbc), with values varying greatly around the world. Regions with high-input agriculture (e.g. Western Europe and Northern America) show small unit WFs and low interannual variability, while low-input regions show opposite outcomes (e.g. Middle and Eastern Africa). From 1986 to 2016, the global average unit WF reduced by a third, mainly due to the historical increase in maize yields. However, due to the rapid expansion of rainfed and irrigated areas, the global WF of maize production increased by half, peaking at 768.3×109 m3 yr−1 in 2016. As many regions still have a high potential in closing yield gaps, unit WFs are likely to reduce further. Simultaneously, humanity's rising demand for food and biofuels may further expand maize areas and hence increase WFs of production. Thus, it is important to address the sustainability and purpose of maize production, especially in those regions where it might endanger ecosystems and human livelihoods.

Landscape controls of surface-water/groundwater interactions on shallow outwash lakes: how the long-term groundwater signal overrides interannual variability due to evaporative effects

The spatial and temporal controls on variability of the relative contributions of groundwater within and between flow systems to shallow lakes in the low-relief glaciated Boreal Plains of Canada were evaluated. Eleven lakes located in a coarse glacial outwash, of varying topographic positions and potential groundwater contributing areas, were sampled annually for stable O and H isotope ratios over the course of 8 years. It was demonstrated that landscape position is the dominant control over relative groundwater contributions to these lakes and the spatial pattern of the long-term isotopic compositions attributed to groundwater overrides interannual variability due to evaporative effects. Lakes at low landscape positions with large potential groundwater capture areas have relatively higher and more consistent groundwater contributions and low interannual variability of isotopic composition. Isolated lakes high in the landscape experience high interannual variability as they have little to no groundwater input to buffer the volumetric or isotopic changes caused by evaporation and precipitation. An alternative explanation that lake morphometry (area and volume) control long-term isotopic compositions is tested and subsequently refuted. Landscape position within coarse outwash is a strong predictor for relative groundwater input; however, surface-water connections can short circuit groundwater pathways and confound the signal. A hydrogeological case study for three of the study lakes is used to contextualize and further demonstrate these results.

Modelling landfast sea ice and its influence on ocean–ice interactions in the area of the Totten Glacier, East Antarctica

The Totten Glacier in East Antarctica is of major climate interest because of the large fluctuation of its grounding line and of its potential vulnerability to climate change. The ocean above the continental shelf in front of the Totten ice shelf exhibits large extents of landfast sea ice with low interannual variability. Landfast sea ice is either crudely or not at all represented in current climate models. These models are potentially omitting or misrepresenting important effects related to this type of sea ice, such as its influence on coastal polynya locations. Yet, the impact of the landfast sea ice on the ocean–ice shelf interactions is poorly understood. Using a series of high-resolution, regional NEMO-LIM-based experiments, including an explicit treatment of ocean–ice shelf interactions, over the years 2001–2010, we simulate a realistic landfast sea ice extent in the area of Totten Glacier through a combination of a sea ice tensile strength parameterisation and a grounded iceberg representation. We show that the presence of landfast sea ice impacts seriously both the location of coastal polynyas and the ocean mixed layer depth along the coast, in addition to favouring the intrusion of mixed Circumpolar Deep Water into the ice shelf cavities. Depending on the local bathymetry and the landfast sea ice distribution, landfast sea ice affects ice shelf cavities differently. The Totten ice shelf melt rate is increased by 16% on average and its variance decreased by 38%, while the Moscow University ice shelf melt rate is increased by +54% in winter. This highlights the importance of including an accurate landfast sea ice representation in regional and eventually global climate models.