Effectiveness of using representative subsets of global climate models in future crop yield projections.

Representative subsets of global climate models (GCMs) are often used in climate change impact studies to account for uncertainty in ensemble climate projections. However, the effectiveness of such subsets has seldom been assessed for the estimations of either the mean or the spread of the full ensembles. We assessed two different approaches that were employed to select 5 GCMs from a 20-member ensemble of GCMs from the CMIP5 ensemble for projecting canola and spring wheat yields across Canada under RCP 4.5 and 8.5 emission scenarios in the periods 2040-2069 and 2070-2099, based on crop simulation models. Averages and spreads of the simulated crop yields using the 5-GCM subsets selected by T&P and KKZ approaches were compared with the full 20-GCM ensemble. Our results showed that the 5-GCM subsets selected by the two approaches could produce full-ensemble means with a relative absolute error of 2.9-4.7% for canola and 1.5-2.2% for spring wheat, and covers 61.8-91.1% and 66.1-80.8% of the full-ensemble spread for canola and spring wheat, respectively. Our results also demonstrated that both approaches were very likely to outperform a subset of randomly selected 5 GCMs in terms of a smaller error and a larger range.

Wetter summers can intensify departures from natural variability in a warming climate.

Climate change can drive local climates outside the range of their historical year-to-year variability, straining the adaptive capacity of ecological and human communities. We demonstrate that dependencies between climate variables can produce larger and earlier departures from natural variability than is detectable in individual variables. Using the example of summer temperature (Tx) and precipitation (Pr), we show that this departure intensification effect occurs when the bivariate climate change trajectory is misaligned with the dominant mode of joint historical variability. Departure intensification is evident in all six CMIP5 models that we examined: 23% (9-34%) of the global land area of each model exhibits a pronounced increase in 2σ anomalies in the Tx-Pr regime relative to Tx or Pr alone. Observational data suggest that summer Tx-Pr correlations in distinct regions on all continents are sufficient to produce departure intensification. Precipitation can be an important driver of multivariate climate change signals relative to natural variability, despite typically having a much weaker univariate signal than temperature.

A closer look at novel climates: new methods and insights at continental to landscape scales.

Novel climates - emerging conditions with no analog in the observational record - are an open problem in ecological modeling. Detecting extrapolation into novel conditions is a critical step in evaluating bioclimatic projections of how species and ecosystems will respond to climate change. However, biologically informed novelty detection methods remain elusive for many modeling algorithms. To assist with bioclimatic model design and evaluation, we present a first-approximation assessment of general novelty based on a simple and consistent characterization of climate. We build on the seminal global analysis of Williams et al. (2007 PNAS, 104, 5738) by assessing of end-of-21st-century novelty for North America at high spatial resolution and by refining their standardized Euclidean distance into an intuitive Mahalanobian metric called sigma dissimilarity. Like this previous study, we found extensive novelty in end-of-21st-century projections for the warm southern margin of the continent as well as the western Arctic. In addition, we detected localized novelty in lower topographic positions at all latitudes: By the end of the 21st century, novel climates are projected to emerge at low elevations in 80% and 99% of ecoregions in the RCP4.5 and RCP8.5 emissions scenarios, respectively. Novel climates are limited to 7% of the continent's area in RCP4.5, but are much more extensive in RCP8.5 (40% of area). These three risk factors for novel climates - regional susceptibility, topographic position, and the magnitude of projected climate change - represent a priori evaluation criteria for the credibility of bioclimatic projections. Our findings indicate that novel climates can emerge in any landscape. Interpreting climatic novelty in the context of nonlinear biological responses to climate is an important challenge for future research.

Attributing extreme fire risk in Western Canada to human emissions.

Canada is expected to see an increase in fire risk under future climate projections. Large fires, such as that near Fort McMurray, Alberta in 2016, can be devastating to the communities affected. Understanding the role of human emissions in the occurrence of such extreme fire events can lend insight into how these events might change in the future. An event attribution framework is used to quantify the influence of anthropogenic forcings on extreme fire risk in the current climate of a western Canada region. Fourteen metrics from the Canadian Forest Fire Danger Rating System are used to define the extreme fire seasons. For the majority of these metrics and during the current decade, the combined effect of anthropogenic and natural forcing is estimated to have made extreme fire risk events in the region 1.5 to 6 times as likely compared to a climate that would have been with natural forcings alone.

Using a down-scaled bioclimate envelope model to determine long-term temporal connectivity of Garry oak (Quercus garryana) habitat in western North America: implications for protected area planning.

Under the Canadian Species at Risk Act (SARA), Garry oak (Quercus garryana) ecosystems are listed as "at-risk" and act as an umbrella for over one hundred species that are endangered to some degree. Understanding Garry oak responses to future climate scenarios at scales relevant to protected area managers is essential to effectively manage existing protected area networks and to guide the selection of temporally connected migration corridors, additional protected areas, and to maintain Garry oak populations over the next century. We present Garry oak distribution scenarios using two random forest models calibrated with down-scaled bioclimatic data for British Columbia, Washington, and Oregon based on 1961-1990 climate normals. The suitability models are calibrated using either both precipitation and temperature variables or using only temperature variables. We compare suitability predictions from four General Circulation Models (GCMs) and present CGCM2 model results under two emissions scenarios. For each GCM and emissions scenario we apply the two Garry oak suitability models and use the suitability models to determine the extent and temporal connectivity of climatically suitable Garry oak habitat within protected areas from 2010 to 2099. The suitability models indicate that while 164 km(2) of the total protected area network in the region (47,990 km(2)) contains recorded Garry oak presence, 1635 and 1680 km(2) of climatically suitable Garry oak habitat is currently under some form of protection. Of this suitable protected area, only between 6.6 and 7.3% will be "temporally connected" between 2010 and 2099 based on the CGCM2 model. These results highlight the need for public and private protected area organizations to work cooperatively in the development of corridors to maintain temporal connectivity in climatically suitable areas for the future of Garry oak ecosystems.

Nonlinear analog predictor analysis: a coupled neural network/analog model for climate downscaling.

Synoptic downscaling models are used in climatology to predict values of weather elements at one or more stations based on values of synoptic-scale atmospheric circulation variables. This paper presents a hybrid method for climate prediction and downscaling that couples an analog, i.e., k-nearest neighbor, model to an artificial neural network (ANN) model. In the proposed method, which is based on nonlinear principal predictor analysis (NLPPA), the analog model is embedded inside an ANN, forming its output layer. Nonlinear analog predictor analysis (NLAPA) is a flexible model that maintains the ability of the analog model to preserve inter-variable relationships and model non-normal and conditional variables (such as precipitation), while taking advantage of NLPPA's ability to define an optimal set of analog predictors that maximize predictive performance. Performance on both synthetic and real-world hydroclimatological benchmark tasks indicates that the NLAPA model is capable of outperforming other forms of analog models commonly used in synoptic downscaling.