Background

Who we are

SNAP is a collaborative network of the University of Alaska, state, federal, and local agencies, and NGOs, whose mission is to provide timely access to management-relevant scenarios of future conditions in Alaska. The primary products of the network are (1) datasets and maps projecting future conditions for selected variables, and (2) rules and models that develop these projections, based on historical conditions and trends.

Projections are always accompanied by metadata that clearly describe the methods and assumptions underlying the projections. Network participants have the responsibility of interpreting these scenarios objectively to potential users. These products also serve to communicate information and assumptions among researchers and to identify knowledge and data gaps requiring further research.

The SNAP integration team consists of a Network Director who assumes responsibility for determining priorities; other UA-funded SNAP personnel to compile data, prepare the scenarios, and deliver products to the users who have requested it; and UA faculty members and collaborators who have assumed responsibility for preparing particular scenarios.

SNAP accepts funds to conduct specific projects. Depending on the scenario and the nature of the stakeholder group, a SNAP collaborators serves in one or more of the following capacities: requesting a scenario, providing the input data necessary for the scenario, providing funding for scenario development, serving as a co-PI in proposals to acquire funding, and using scenario products.

The Need for a Scenarios Network in Alaska

Currently most policy and management planning for Alaska and elsewhere assumes that future conditions will be similar to those of our recent past experience. However, there is reasonable consensus within the scientific community that future climatic, ecological, and economic conditions will likely be quite different from those of the past. We now know enough about current and likely future trajectories of climate and other variables to develop credible projections. We can also make projections for other variables that are closely correlated, such as frequency of intense storms, risk of wildfire or flooding, and habitat and wildlife changes associated with these events.

Scenarios of future climate and associated environmental hazards have already been developed in Scandinavia, British Columbia, England, Wisconsin lakes district, and elsewhere to guide planning for climate change. Thus, conceptual validity and technical feasibility have already been demonstrated.

The idea of developing a scenario planning process for Alaska emerged in 2006 and 2007 from discussions by an interdisciplinary group of about a dozen University of Alaska faculty. The consensus of that group was that such a process would be feasible and would be one of the most useful ways that University of Alaska researchers could convey the societal significance of their research to Alaskan decision-makers and other stakeholders. Indeed, individual researchers had already completed some of the basic future scenarios for Alaska.

Scenario Development

Types of problems addressed

A wide range of problems is amenable to the scenario approach. We place highest priority on problems or issues where management decisions can affect the outcome (e.g., decisions affecting fire suppression), or where long-term societal impacts may be unanticipated (e.g., permafrost instability near communities).

We are currently addressing issues matching the research expertise of university faculty in the SNAP network, which centers on the consequences of climate change in Alaska. We are addressing each problem in at least two stages: we first examine the effects of climate change on biophysical processes, and then examine the effects of these physical changes on society. These include (1) loss of sea ice and coastal erosion, which affect community infrastructure, marine mammals, and oil development; (2) land cover change (due to development or wildfire), which affect fire risk, habitat and subsistence; (3) loss of permafrost integrity, which affects infrastructure and habitat; and (4) changes in hydrology, which affects flooding, water supply, snow cover, winter exploration, and agriculture. Other important issues that might be developed later include energy use and invasive species.

Steps required to develop scenarios

There are three steps involved in the development of a scenario:

Step 1. Estimate the current spatial pattern for Alaska and/or its surrounding seas of key variables (e.g., climate, moose, coastal erosion, median income, population size) based on existing data. Develop statistical or dynamic rules or simple models that describe the spatial distributions based on potential predictor variables. Because all datasets are spatially incomplete, use these spatial rules to interpolate spatial patterns to parts of Alaska and adjacent seas, where information is sufficient for valid interpolation. Check maps and the interpolations against other datasets and experienced observers to improve over time the rules and confidence in spatial patterns.

Step 2. Use historical time series to reconstruct maps of these variables for the time periods and locations for which data are available. Develop temporal rules based on potential predictor variables to predict the response variable for times and locations where information is sufficient for valid interpolation. As with spatial patterns, we can improve the rules for interpolation and our confidence in past patterns based on comparisons with new datasets and discussions with experienced observers. The process of attempting interpolation of spatial and temporal patterns will identify gaps in data and in understanding.

Step 3. Use these temporal prediction rules to project into the future those variables that will likely be controlled by the same processes in the future as in the past (e.g., climate). The products of this analysis are time series of maps of climate, probability of extreme events (e.g., risk of wildfire or flooding), and ecosystem services from the historical past to some date in the future.

Step 4. Scenarios can also be developed to explore the consequences of alternative management or policy decisions that might influence future conditions—for example, the impact on future fire risk of suppressing all fires near communities, as contrasted with allowing some types of fires to burn.

Assuring scientific rigor of scenarios

There are several critical science issues addressed in providing rigorous scenarios, given that there is uncertainty in both the data and the model structure used to develop scenarios. For example, the plausible range of values in the predictor variables (i.e., the driving dataset) and the assumptions embedded in statistical or dynamic models are always identified and clearly stated in the metadata provided with resulting scenarios. The temporal and spatial scale and resolution differ among scenario projections, depending on the problem to be addressed. The default scale of resolution for our scenarios is 2 km2 with annual time steps. Where feasible, scenario results are archived for the largest area and time period for which they are valid, so as to be available to other potential users. For variables strongly influenced by human actions, plausible time frames are roughly 30-50 years. For ecological processes (e.g., wildfire) whose probability depends on previous infrequent events, the most useful time frames are sometimes >100 years of historical data, and projections to 2100. This framework provides products that can be used at scales from individual communities to pan-arctic or global models.

Derivation of SNAP Climate Projections

Use of GCMs to model future climate

General Circulation Models (GCMs) are the most widely used tools for projections of global climate change over the timescale of a century. Periodic assessments by the Intergovernmental Panel on Climate Change (IPCC) have relied heavily on global model simulations of future climate driven by various emission scenarios.

The IPCC uses complex coupled atmospheric and oceanic GCMs. These models integrate multiple equations, typically including surface pressure; horizontal layered components of fluid velocity and temperature; solar short wave radiation and terrestrial infra-red and long wave radiation; convection; land surface processes; albedo; hydrology; cloud cover; and sea ice dynamics.

GCMs include equations that are iterated over a series of discrete time steps as well as equations that are evaluated simultaneously. Anthropogenic inputs such as changes in atmospheric greenhouse gases can be incorporated into stepped equations. Thus, GCMs can be used to simulate the changes that may occur over long time frames due to the release of excess greenhouse gases into the atmosphere.

Greenhouse-driven climate change represents a response to the radiative forcing associated with increases of carbon dioxide, methane, water vapor and other gases, as well as associated changes in cloudiness. The response varies widely among models because it is strongly modified by feedbacks involving clouds, the cryosphere, water vapor and other processes whose effects are not well understood. Changes in the radiative forcing associated with increasing greenhouse gases have thus far been small relative to existing seasonal cycles. Thus, the ability of a model to accurately replicate seasonal radiative forcing is a good test of its ability to predict anthropogenic radiative forcing.

Model Selection

Different coupled GCMs have different strengths and weaknesses, and some can be expected to perform better than others for northern regions of the globe.

John Walsh et al. evaluated the performance of a set of fifteen global climate models used in the Coupled Model Intercomparison Project. Using the outputs for the A1B (intermediate) climate change scenario, they calculated the degree to which each model’s output concurred with actual climate data for the years 1958-2000 for each of three climatic variables (surface air temperature, air pressure at sea level, and precipitation) for three overlapping regions (Alaska only, 60-90 degrees north latitude, and 20-90 degrees north latitude.)

The core statistic of the validation was a root-mean-square error (RMSE) evaluation of the differences between mean model output for each grid point and calendar month, and data from the European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis, ERA-40. The ERA-40 directly assimilates observed air temperature and sea level pressure observations into a product spanning 1958-2000. Precipitation is computed by the model used in the data assimilation. The ERA-40 is one of the most consistent and accurate gridded representations of these variables available.

To facilitate GCM intercomparison and validation against the ERA-40 data, all monthly fields of GCM temperature, precipitation and sea level pressure were interpolated to the common 2.5° × 2.5° latitude–longitude ERA-40 grid. For each model, Walsh et al. calculated RMSEs for each month, each climatic feature, and each region, then added the 108 resulting values (12 months x 3 features x 3 regions) to create a composite score for each model. A lower score indicated better model performance.

The specific models that performed best over the larger domains tended to be the ones that performed best over Alaska. Although biases in the annual mean of each model typically accounted for about half of the models’ RMSEs, the systematic errors differed considerably among the models. There was a tendency for the models with the smaller errors to simulate a larger greenhouse warming over the Arctic, as well as larger increases of Arctic precipitation and decreases of Arctic sea level pressure when greenhouse gas concentrations are increased.

Since several models had substantially smaller systematic errors than the other models, the differences in greenhouse projections implied that the choice of a subset of models might offer a viable approach to narrowing the uncertainty and obtaining more robust estimates of future climate change in regions such as Alaska. Thus, SNAP selected the five best-performing models out of the fifteen: MPI_ECHAM5, GFDL_CM2_1, MIROC3_2_MEDRES, UKMO_HADCM3, and CCCMA_CGCM3_1 These five models are used to generate climate projections independently, as well as in combination, in order to further reduce the error associated with dependence on a single model.

Downscaling model outputs

Because of the enormous mathematical complexity of GCMs, they generally provide only large-scale output, with grid cells typically 1°-5° latitude and longitude. For example, the standard resolution of HadOM3 is 1.25 degrees in latitude and longitude, with 20 vertical levels, leading to approximately 1,500,000 variables.

Finer scale projections of future conditions are not directly available. However, local topography can have profound effects on climate at much finer scales, and almost all land management decisions are made at much finer scales. Thus, some form of downscaling is necessary in order to make GCMs useful tools for regional climate change planning.

Historical climate data estimates at 2km resolution are available from PRISM (Parameter-elevation Regressions on Independent Slopes Model), which was originally developed to address the lack of climate observations in mountainous regions or rural areas. PRISM uses point measurements of climate data and a digital elevation model to generate estimates of annual, monthly and event-based climatic elements. Each grid cell is estimated via multiple regression using data from many nearby climate stations. Stations are weighted based on distance, elevation, vertical layer, topographic facet, and coastal proximity.

PRISM offers data at a fine scale useful to land managers and communities, but it does not offer climate projections. Thus, SNAP needed to link PRISM to GCM outputs. This work was also done by John Walsh, Bill Chapman, et al. They first calculated mean monthly precipitation and mean monthly surface air temperature for PRISM grid cells for 1961-1990, creating PRISM baseline values. Next, they calculated GCM baseline values for each of the five selected models using mean monthly outputs for 1961-1990. They then calculated differences between projected GCM values and baseline GCM values for each year out to 2099 and created “anomaly grids” representing these differences. Finally, they added these anomaly grids to PRISM baseline values, thus creating fine-scale (2 km) grids for monthly mean temperature and precipitation for every year out to 2099. This method effectively removed model biases while scaling down the GCM projections.

Based on this small-scale grid, SNAP now has statewide maps (available as static maps or GIS layers) of mean monthly temperature and precipitation based on each of the five selected models and the means of all five, as well as projections for 353 communities. As described above, these data and maps can be linked to biophysical processes and thereby to social change and management and policy choices.