What is Time of Emergence?
- What is the "Time of Emergence” of climate change?
- Why is it important?
- How could results from Time of Emergence analyses be used?
- Determining Time of Emergence: The Signal Threshold Method
“Time of Emergence” refers to the time when conditions are projected to distinctively differ from the past due to climate change. In other words, it is when the effects of human-induced climate change become apparent.
In this tool, the Time of Emergence is defined as the time when the change in conditions becomes sufficiently substantial to matter for decision making or risk reduction activities.
Technically, the “Time of Emergence” is a concept indicating the time when the climate change “signal” is projected to emerge from the “noise” of past climate variability.
In this tool, the “noise” is defined in relation to management sensitivity (see Figure 1).
Figure 1. Illustrative figure showing the components of a Time of Emergence analysis.
Natural and human systems in any one place tend to be more or less adapted to the climate of that place. For instance, infrastructure design criteria and regulatory standards are often based on tolerances developed from knowledge of historical variability (e.g., drainage facilities may be designed to accommodate peak discharge resulting from a 24-hour, 100-year storm event). However, considerable changes in Pacific Northwest climate are projected in the coming decades as a result of greenhouse gas emissions from human activities. When climate change causes local conditions to move past those to which the system is adapted, ecological and societal disruptions could occur in the absence of adequate preparation for those changes.
This tool aims to support climate change decision-making (e.g., the identification of strategic and operational priorities for reducing climate risks) by profiling the projected location and timing of management-relevant effects of climate change. Because it is impossible to predict future climate precisely, this tool uses best available science to illustrate a range of plausible futures and allows the user to explore the sensitivity of results to different plausible underlying assumptions.
Most simply, Time of Emergence analyses indicate when climate change may result in significant changes in a variable of interest (see “Climate variables”). This relative indication, combined with information on local sensitivities, design standards or critical thresholds, and the time required for taking preparatory action, can be used to help prioritize decisions about when, where and for which climate change impacts specific response actions may be necessary.
Time of Emergence analyses could be used to identify which variables might be of more immediate concern – the effects of climate change may be more apparent sooner in some variables than others where the projected change is small compared to historic variability. This information could be used to prioritize climate change adaptation actions towards changes simulated to manifest earlier, rather than later. Prioritization of decisions should also be informed by information about the length of time required to implement necessary preparatory actions, management risk tolerance, and the relative consequences of negative climate impacts, in addition to other priority decision criteria.
The Time of Emergence results presented in this tool were developed using the signal threshold method, where:
Time of Emergence = The first year when the projected future state of a variable (determined by a linear fit to the time series) crosses a pre-defined threshold for emergence (see Maraun 2013).
Therefore, Time of emergence consists of two components:
- The climate change signal – extracted from the projected future conditions under the influence of climate change.
- The threshold at which climate change is said to emerge – defined based on assumptions about management's ability to cope with the “noise” of the baseline period.
Figure 2. Time of Emergence for the variable “Annual Number of Days Warmer than 90 degrees F” at a single grid cell in the Pacific Northwest based on projections from one global climate model. The Time of Emergence is defined as the year (in this case ~2021) when the climate change signal crosses the emergence threshold.
The Time of Emergence of management-relevant climate change is not a fixed value. It depends on how management-relevance is defined, since some systems will be affected by relatively minor climate changes, while others may be robust to all but the largest expected change. And because there is no single best estimate for how climate change will unfold, Time of Emergence also differs for different climate change scenarios. That is why this tool provides information about the plausible range of Time of Emergence, as well as opportunities for exploring how Time of Emergence varies under different assumptions.
Determining the Time of Emergence: Other Methods
Other methods of determining Time of Emergence include:
From the time series of a variable, the time varying signal, s(t), is estimated as the long term monotonic change in the variable. The noise, N, is based on the range of variability (e.g., the standard deviation) over some baseline period. The Time of Emergence is defined as the time t when s(t)/N exceeds some value, typically 1 or 2 (see Hawkins and Sutton, 2012). For example, a signal-to-noise-ratio of two implies the Time of Emergence when the signal – due to climate change – is twice as large as the noise due to natural variability.
2. Exceedance Threshold:
The upper limit for a variable is set based on some baseline period. The Time of Emergence is defined as the time when a selected number of consecutive years exceed this threshold, for example 3 years, 11 years, or all years (see Mora, 2013).
Deser, C., R. Knutti, S. Solomon, and A.S. Phillips, 2014: Communication of the role of natural variability in future North American climate, Nature Climate Change, 2, 775–779.
Diffenbaugh, N.S., and M. Scherer, 2011: Observational and model evidence of global emergence of permanent, unprecedented heat in the 20th and 21st centuries, Clim Change, 107(3-4): 615–624.
Hawkins, E., and R. Sutton, 2012: Time of emergence of climate signals, Geophys. Res. Lett., 39, L01702.
Maraun, D., 2013: When will trends in European mean and heavy daily precipitation emerge? Environ. Res. Lett., 8, 014004.
Mora, C. et al., 2013: The projected timing of climate departure from recent variability, Nature, 502, 183-187.
The signal, i.e., the change of a variable due to climate change, may be in the positive (increasing) or negative (decreasing) direction. The former “emerges” by crossing the upper emergence threshold – emergence of a positive trend. The latter “emerges” by crossing the lower threshold – emergence of a negative trend.
The signal itself is calculated using the slope of the linear best-fit to projected 21st century (2006-2100) conditions, and a starting year of 2001 (estimated as the mean of the climatological (1980-2010) baseline). To account for the uncertainty in estimating the climate change signal, we also calculate Time of Emergence using an upper and lower bound for the estimated slope (i.e., rate) of climate change. This range is designed to encompass, with 90% confidence, the true value of the signal.
Limited computational power limits the level of detail at which climate models can represent physical processes (e.g., formation of clouds, precipitation). Representing processes at finer scales than the model grid cells is called parameterization, and depends on the technique applied. Experiments based on perturbed physics ensembles (PPE) are often used to explore the effects of different choices of various model parameters, known as parameter uncertainties.
Changing the values of parameters slightly in the same climate model can produce different outcomes; i.e., using another simulation from the same global climate model could result in a linear signal with a different slope. Since only one simulation from each climate model is used to extract information about the climate change signal in this analysis, the uncertainty in this signal is estimated from the statistical error in computing the slope.
The 5th and 95th percentile confidence interval is computed as the standardized error from the LSR calculation, which is multiplied by the Student’s t-value at the 95% significance level, to obtain the 95% error term. Adding and subtracting this error term from the LSR calculated slope give an upper and lower bound to the signal. In other words, there is 95% confidence that the true signal lies between the upper and lower bounds.
In this tool, the estimated range of the (linear) climate change signal is presented as different user-selectable Rates of Climate Change:
Upper bound – Faster rate
Central estimate – Moderate rate
Lower bound – Slower rate
Time of Emergence calculations are done at each grid cell using projections from each global climate model. Time of Emergence is defined as the year at which the linear signal crosses the predefined thresholds. Thus, for each climate model and each grid cell, there are twelve ToE values (see Table XXX for the various combinations of threshold and slope).
Some variables show such significant future fluctuations that the future direction of change cannot be reliably estimated. For example, a variable may appear to emerge in the increasing direction, but have sufficiently large uncertainty in the computed slope that emergence in the decreasing direction is also plausible. In other words, the lower and upper bounds of the signal produce emergence in different directions. No Time of Emergence is calculated in cases with such a high degree of uncertainty in the climate change signal.
Emergence thresholds distinguish conditions to which the management system is presumed to be well adapted from those presumed to trigger impacts. In other words, the management system is presumed to be sensitive to conditions exceeding the emergence thresholds but to cope well with conditions lying between the emergence thresholds.
Recognizing that the specific thresholds beyond which management-relevant impacts occur are user- and context-specific, we provide Time of Emergence results for two different, user-selectable sets of thresholds. These thresholds represent low- and high levels of management sensitivity to changes in conditions (see Table 1):
- High management sensitivity: impacts triggered by the extreme 20% high or 20% low conditions experienced during the baseline period (1950-1999)
- Low management sensitivity: impacts triggered by the extreme 5% high or 5% low, conditions during 1950-1999
|Management Sensitivity to Changes in Conditions|
Emergence Thresholds (as percentile of baseline (1950-1999) conditions)
|Lower||20th percentile||5th percentile|
|Upper||80th percentile||95th percentile|
|Management coping range||
(adapted to the central 60% of historical conditions)
(adapted to the central 90% of historical conditions)
|Sensitivity to climate change||High||Low|
|Time of Emergence||Earlier||Later|
Spatially-aggregated ToE results are available for:
- Counties (in Washington, Oregon and Idaho)
- Watersheds – 4th-Level (8-digit) Hydrological Unit Codes (HUCs)
A spatial unit larger than a county or watershed may not provide useful information due to the topographic heterogeneity across the Pacific Northwest that would lead to the averaging of dissimilar results.
The models divide the Pacific Northwest into a grid, and ToE results are calculated separately at each grid cell. Within any selected spatial unit, some grid cells may show a positive trend (e.g., conditions becoming wetter) and some may show a negative trend (e.g., drying). Results for each global climate model are only aggregated across a county or watershed if 60% or more of the grid cells within that spatial unit show changes in the same direction. Then, an average is done for results in these grid cells.
Results from a single global climate model (GCM) only suggest a single trajectory of how future climate might unfold, which overlooks the full range of possible future conditions and could therefore be misleading. Hence, multi-model – known as ensemble – results are presented here:
- Central estimates are represented by the median values for the group of global climate models examined.
- Projection range is characterized by the 25th and 75th percentile values, which capture the middle 50% of the inter-model results.
The global climate model results are not weighted in any way; each model is considered equally credible, with each projection equally plausible (note: but not equally likely). This is because:
- A model that does well reproducing the past climate does not necessarily mean it would do well for the future, especially when long-term climate projections cannot be validated directly through observations; and,
- Whether or not a GCM does “well” at reproducing the past climate can vary with the aspect of past climate you are evaluating its performance on. Therefore, model rankings can change depending on which metrics/criteria are used to evaluate them.
Global Climate Model Agreement
Projections from different global climate models may give different direction of change and/or Time of Emergence. GCM agreement provides an indication of the strength of evidence based on the GCMs applied in this analysis, as represented by the consensus in the model ensemble.
GCM agreement is expressed as the percentage of GCMs examined projecting emergence in the same direction, and by the indicated time period. For example, 75% GCM agreement denotes that at least 15 out of the 21 GCMs examined simulate emergence in the same direction. If 12 of the 21 models show emergence in the negative direction, the ToE for model agreement of 25% and 50% can be calculated, but the ToE for 75% model agreement cannot be calculated. For hydrologic results where only 6 GCMs were available, 75% GCM agreement implies at least 4 out of the 6 GCMs examined project emergence in the same direction.
Results are presented for 25%, 50% and 75% GCM agreement. For the spatially-aggregated results, GCM agreement was calculated using the spatially-averaged ToE values, for each variable, emission scenario, management sensitivity, and rate of climate change at each spatial unit.
Note: Users should not be over-optimistic about consensus estimates, because it gives no information on the likelihood, or probability, of something occurring in future. For example: A 75% GCM agreement is not indicating that there is a 75% chance that future conditions will deviate, from the 1950–1999 period, by the indicated date.
There are a variety of reasons why a variable may be indicated as “non-emergent”, based on the Time of Emergence methodology used, including lack of agreement among projections for that variable, infrequent occurrence during the baseline period or extremely slow emergence. Specifically, “No emergence” is indicated for an individual variable in a particular location in cases with:
- No occurrence of the conditions of interest (i.e., exceedance of the threshold) during the baseline period (1950-1999) for a specific input model. If the threshold is never exceeded in the baseline period, the Time of Emergence would be defined as the first occurrence in the future. In this case, another approach, such defining emergence as the third occurrence of this exceedence (Mora, 2013), might be required to ensure that a robust emergence date is provided.
- No emergence by 2100 for a particular combination of threshold and slope for a specific input model.
“No emergence” is indicated for spatially aggregated Time of Emergence results if:
- The grid cells in the selected spatial unit that show emergence by 2100 do not agree on the direction of the climate change signal, or
- Less than 60% of the grid cells in the selected spatial unit show emergence by 2100.
"No emergence" is indicated for multi-model spatially-aggregated ToE results (e.g., ToE at 25%, 50% (multi-model median), and 75% model agreement)
- Less than 25%, 50% or 75% agreement on the direction of the climate change signal within the original ensemble of climate models considered for the spatially-aggregated ToEs
- Fewer than 25%, 50% or 75% of the spatially-aggregated ToEs show emergence by 2100.
A variable may show emergence by 2001. This could be due to:
- A very fast rate of climate change and very low emergence threshold (i.e., relatively small fluctuations during the baseline period).
- For temperature-related variables, conditions over the 1980-2010 period may already diverging from those during the baseline (1950-1999) period.
- For precipitation-related variables, this is likely caused instead by a few large events during the 1980-2010 period, rather than an overall change in conditions.
The Time of Emergence project was conceived and funded by U.S. Army Corps of Engineers Climate Preparedness & Resilience programs & U.S. Environmental Protection Agency-Region 10. Methodologies and stakeholder engagement were developed and implemented by the University of Washington's Climate Impacts Group. The Time of Emergence online tool was developed with support from the Center for Data Science, University of Washington-Tacoma.