Seasonal-to-decadal predictions are inevitably uncertain, depending on the size of the predictable signal relative to unpredictable chaos. Uncertainties can be accounted for using ensemble techniques, permitting quantitative probabilistic forecasts. In a perfect system, each ensemble member would represent a potential realization of the true evolution of the climate system, and the predictable components in models and reality would be equal. However, we show that the predictable component is sometimes lower in models than observations, especially for seasonal forecasts of the North Atlantic Oscillation and multiyear forecasts of North Atlantic temperature and pressure. In these cases the forecasts are underconfident, with each ensemble member containing too much noise. Consequently, most deterministic and probabilistic measures underestimate potential skill and idealized model experiments underestimate predictability. However, skilful and reliable predictions may be achieved using a large ensemble to reduce noise and adjusting the forecast variance through a postprocessing technique proposed here.

No abstract

Abstract. Substantial amounts of secondary organic aerosol (SOA) can be formed from isoprene epoxydiols (IEPOX), which are oxidation products of isoprene mainly under low-NO conditions. Total IEPOX-SOA, which may include SOA formed from other parallel isoprene oxidation pathways, was quantified by applying positive matrix factorization (PMF) to aerosol mass spectrometer (AMS) measurements. The IEPOX-SOA fractions of organic aerosol (OA) in multiple field studies across several continents are summarized here and show consistent patterns with the concentration of gas-phase IEPOX simulated by the GEOS-Chem chemical transport model. During the Southern Oxidant and Aerosol Study (SOAS), 78 % of PMF-resolved IEPOX-SOA is accounted by the measured IEPOX-SOA molecular tracers (2-methyltetrols, C5-Triols, and IEPOX-derived organosulfate and its dimers), making it the highest level of molecular identification of an ambient SOA component to our knowledge. An enhanced signal at C5H6O+ (m/z 82) is found in PMF-resolved IEPOX-SOA spectra. To investigate the suitability of this ion as a tracer for IEPOX-SOA, we examine fC5H6O (fC5H6O= C5H6O+/OA) across multiple field, chamber, and source data sets. A background of ~ 1.7 ± 0.1 ‰ (‰ = parts per thousand) is observed in studies strongly influenced by urban, biomass-burning, and other anthropogenic primary organic aerosol (POA). Higher background values of 3.1 ± 0.6 ‰ are found in studies strongly influenced by monoterpene emissions. The average laboratory monoterpene SOA value (5.5 ± 2.0 ‰) is 4 times lower than the average for IEPOX-SOA (22 ± 7 ‰), which leaves some room to separate both contributions to OA. Locations strongly influenced by isoprene emissions under low-NO levels had higher fC5H6O (~ 6.5 ± 2.2 ‰ on average) than other sites, consistent with the expected IEPOX-SOA formation in those studies. fC5H6O in IEPOX-SOA is always elevated (12–40 ‰) but varies substantially between locations, which is shown to reflect large variations in its detailed molecular composition. The low fC5H6O (< 3 ‰) reported in non-IEPOX-derived isoprene-SOA from chamber studies indicates that this tracer ion is specifically enhanced from IEPOX-SOA, and is not a tracer for all SOA from isoprene. We introduce a graphical diagnostic to study the presence and aging of IEPOX-SOA as a triangle plot of fCO2 vs. fC5H6O. Finally, we develop a simplified method to estimate ambient IEPOX-SOA mass concentrations, which is shown to perform well compared to the full PMF method. The uncertainty of the tracer method is up to a factor of ~ 2, if the fC5H6O of the local IEPOX-SOA is not available. When only unit mass-resolution data are available, as with the aerosol chemical speciation monitor (ACSM), all methods may perform less well because of increased interferences from other ions at m/z 82. This study clarifies the strengths and limitations of the different AMS methods for detection of IEPOX-SOA and will enable improved characterization of this OA component.

Bioaerosols are relevant for public health and may play an important role in the climate system, but their atmospheric abundance, properties, and sources are not well understood. Here we show that the concentration of airborne biological particles in a North American forest ecosystem increases significantly during rain and that bioparticles are closely correlated with atmospheric ice nuclei (IN). The greatest increase of bioparticles and IN occurred in the size range of 2–6 μm, which is characteristic for bacterial aggregates and fungal spores. By DNA analysis we found high diversities of airborne bacteria and fungi, including groups containing human and plant pathogens (mildew, smut and rust fungi, molds, Enterobacteriaceae, Pseudomonadaceae). In addition to detecting known bacterial and fungal IN (Pseudomonas sp., Fusarium sporotrichioides), we discovered two species of IN-active fungi that were not previously known as biological ice nucleators (Isaria farinosa and Acremonium implicatum). Our findings suggest that atmospheric bioaerosols, IN, and rainfall are more tightly coupled than previously assumed

[1] Submicron atmospheric particles in the Amazon Basin were characterized by a high-resolution aerosol mass spectrometer during the wet season of 2008. Patterns in the mass spectra closely resembled those of secondary-organicaerosol (SOA) particles formed in environmental chambers from biogenic precursor gases. In contrast, mass spectral indicators of primary biological aerosol particles (PBAPs) were insignificant, suggesting that PBAPs contributed negligibly to the submicron fraction of particles during the period of study. For 40% of the measurement periods, the mass spectra indicate that in-Basin biogenic SOA production was the dominant source of the submicron mass fraction, contrasted to other periods (30%) during which out-of-Basin organic-carbon sources were significant on top of the baseline in-Basin processes. The in-Basin periods had an average organic-particle loading of 0.6 mg m À3 and an average elemental oxygen-to-carbon (O:C) ratio of 0.42, compared to 0.9 mg m À3 and 0.49, respectively, during periods of out-of-Basin influence. On the basis of the data, we conclude that most of the organic material composing submicron particles over the Basin derived from biogenic SOA production, a finding that is consistent with microscopy observations made in a concurrent study. This source was augmented during some periods by aged organic material delivered by long-range transport. Citation: Chen, Q., et al.

Precipitation nowcasting, the high-resolution forecasting of precipitation up to two hours ahead, supports the real-world socioeconomic needs of many sectors reliant on weather-dependent decision-making1,2. State-of-the-art operational nowcasting methods typically advect precipitation fields with radar-based wind estimates, and struggle to capture important non-linear events such as convective initiations3,4. Recently introduced deep learning methods use radar to directly predict future rain rates, free of physical constraints5,6. While they accurately predict low-intensity rainfall, their operational utility is limited because their lack of constraints produces blurry nowcasts at longer lead times, yielding poor performance on rarer medium-to-heavy rain events. Here we present a deep generative model for the probabilistic nowcasting of precipitation from radar that addresses these challenges. Using statistical, economic and cognitive measures, we show that our method provides improved forecast quality, forecast consistency and forecast value. Our model produces realistic and spatiotemporally consistent predictions over regions up to 1,536 km × 1,280 km and with lead times from 5–90 min ahead. Using a systematic evaluation by more than 50 expert meteorologists, we show that our generative model ranked first for its accuracy and usefulness in 89% of cases against two competitive methods. When verified quantitatively, these nowcasts are skillful without resorting to blurring. We show that generative nowcasting can provide probabilistic predictions that improve forecast value and support operational utility, and at resolutions and lead times where alternative methods struggle.

The global organic aerosol (OA) budget is highly uncertain and past studies suggest that models substantially underestimate observed concentrations. Few of these studies have examined the vertical distribution of OA. Furthermore, many model-measurement comparisons have been performed with different models for single field campaigns. We synthesize organic aerosol measurements from 17 aircraft campaigns from 2001–2009 and use these observations to consistently evaluate a GEOS-Chem model simulation. Remote, polluted and fire-influenced conditions are all represented in this extensive dataset. Mean observed OA concentrations range from 0.2–8.2 μg sm⁻³ and make up 15 to 70% of non-refractory aerosol. The standard GEOS-Chem simulation reproduces the observed vertical profile, although observations are underestimated in 13 of the 17 field campaigns (the median observed to simulated ratio ranges from 0.4 to 4.2), with the largest model bias in anthropogenic regions. However, the model is best able to capture the observed variability in these anthropogenically-influenced regions (<i>R</i>²=0.18&minus;0.57), but has little skill in remote or fire-influenced regions. The model bias increases as a function of relative humidity for 11 of the campaigns, possibly indicative of missing aqueous phase SOA production. However, model simulations of aqueous phase SOA suggest a pronounced signature in the mid-troposphere (2–6 km) which is not supported in the observations examined here. Spracklen et al. (2011) suggest adding ~100 Tg yr⁻¹ source of anthropogenically-controlled SOA to close the measurement-model gap, which we add as anthropogenic SOA. This eliminates the model underestimate near source, but leads to overestimates aloft in a few regions and in remote regions, suggesting either additional sinks of OA or higher volatility aerosol at colder temperatures. Sensitivity simulations indicate that fragmentation of organics upon either heterogeneous or gas-phase oxidation could be an important (missing) sink of OA in models, reducing the global SOA burden by 15% and 47% respectively. The best agreement with observations is obtained when the simulated anthropogenically-controlled SOA is increased to ~100 Tg yr⁻¹ accompanied by either a gas-phase fragmentation process or a reduction in the temperature dependence of the organic aerosol partitioning (by decreasing the enthalpy of vaporization from 42 kJ mol⁻¹ to 25 kJ mol^&minus;1). These results illustrate that models may require both additional sources and additional sinks to capture the observed concentrations of organic aerosol

Bioaerosols are relevant for public health and may play an important role in the climate system, but their atmospheric abundance, properties and sources are not well understood. Here we show that the concentration of airborne biological particles in a forest ecosystem increases dramatically during rain and that bioparticles are closely correlated with atmospheric ice nuclei (IN). The greatest increase of bioparticles and IN occurred in the size range of 2–6 μm, which is characteristic for bacterial aggregates and fungal spores. By DNA analysis we found high diversities of airborne bacteria and fungi, including human and plant pathogens (mildew, smut and rust fungi, molds, <i>Enterobacteraceae, Pseudomonadaceae</i>). In addition to known bacterial and fungal IN (<i>Pseudomonas</i> sp., <i>Fusarium sporotrichioides</i>), we discovered two species of IN-active fungi that were not previously known as biological ice nucleators (<i>Isaria farinosa</i> and <i>Acremonium implicatum</i>). Our findings suggest that atmospheric bioaerosols, IN and rainfall are more tightly coupled than previously assumed