This dataset consists of single band GeoTIFFs containing total annual counts of wet days for each year from 1980-2100 for one downscaled reanalysis (ERA-Interim, 1980-2015) and two downscaled CMIP5 global climate models driven under the RCP 8.5 baseline emissions scenario (NCAR-CCSM4 and GFDL-CM3, 2006-2100), all derived from the same dynamical downscaling effort using the Weather Research and Forecasting (WRF) model (Version 3.5). A day is counted as a "wet day" if the total precipitation for that day is 1 mm or greater.

This set of files includes downscaled historical estimates of monthly total precipitation (in millimeters) at 1 kilometer spatial resolution. Each file represents a single month in a given year. The original SNAP downscaled precipitation product at 2 kilometer spatial resolution was resampled to 1 kilometer spatial resolution via bilinear interpolation to create these data for input to the Integrated Ecosystem Model (IEM). Please note that this data is used to fill in a gap in available data for the IEM and does not constitute a complete or precise measurement of this variable in all locations.

This data includes quantile-mapped historical and projected model runs of AR5 daily mean near surface wind velocity (m/s) for each day of every year from 1958 - 2100 at 2.5 x 2.5 degree spatial resolution across 3 AR5 models. They are 365 multi-band geotiff files, one file per year, each band representing one day of the year, with no leap years.

This set of files includes downscaled historical estimates of monthly totals, and derived annual, seasonal, and decadal means of monthly total precipitation (in millimeters, no unit conversion necessary) from 1901 - 2006 (CRU TS 3.0) or 2009 (CRU TS 3.1) at 771 x 771 meter spatial resolution.

These GeoTIFFs include annual spatial representations of the following variables produced through summarization of ALFRESCO model outputs across 200 replicates: Flammability: likelihood of a pixel to burn across 200 replicates Modal vegetation type: statistical mode of vegetation type across 200 replicates Percent vegetation type: percent of each possible vegetation type across 200 replicates These outputs were derived from AR5/CMIP5 climate inputs, historical fire inputs from the Alaska Interagency Coordination Center (AICC), and several fire management options (FMO) inputs.

This dataset consists of spatial representations of relative vegetation change produced through summarization of ALFRESCO model outputs. These specific outputs are from the Integrated Ecosystem Model (IEM) project, and are from the linear coupled version using AR5/CMIP5 climate inputs (IEM Generation 2).

Atmospheric rivers (ARs) were detected from ERA5 6hr pressure level data, using a detection algorithm adapted from Guan & Waliser (2015). The algorithm uses a combination of vertically integrated water vapor transport (IVT), geometric shape, and directional criteria to define ARs. See the sources listed below and the GitHub repository for more detail and other references. The AR database is a zipped archive containing multiple attributed shapefiles. Polygon data includes individual timestep ARs, ARs making landfall in Alaska, and aggregated landfalling AR events. Point data includes coastal impact points landfalling AR events.

These wind speed and direction data are the underlying data displayed in the interactive webtool at http://snap.uaf.edu/tools/airport-winds. Original wind speed/direction observations were made by Automated Surface Observing System (ASOS) and the Automated Weather Observing System (AWOS) stations, and we accessed these data via the Iowa Environmental Mesonet (IEM). These observations were hourly in most cases, and we filtered data to routine measurements (nearest to clock hour) where measurements were more frequent than hourly to generate a true hourly dataset, save for periods of missing data. We used data from 166 weather stations located across Alaska, selected from a pool of 185 stations available in the IEM database for 1980-2019. For inclusion in the app and this dataset, a station must have a reasonably complete record, and must have begun measurements before June 6, 2010. We applied a spike-filtering algorithm to detect spurious spikes and dips, and a changepoint detection plus quantile mapping adjustment to statistically account for the possibility of sensors changing location, height, or surroundings such that the long term (month-scale) wind regimes were affected. **Methodology** --- All hourly ASOS/AWOS wind speed and direction data available via the Iowa Environmental Mesonet AK ASOS network were accessed and assessed for completeness (185 stations), and 166 of those stations were determined to be sufficiently complete for climatological analysis. Those data were cleaned to produce regular hourly data, and adjusted via a combination of changepoint analysis and quantile mapping to correct for potential changes in sensor location and height. **Attribute Description** --- ts: timestamp (YYYY-mm-dd HH:MM:SS) ws: wind speed (mph) wd: wind direction (degrees) Station identifiers used for locations is available at: https://www.faa.gov/air_traffic/weather/asos/?state=AK

These files include historical downscaled estimates of decadal average monthly snow-day fraction ("fs", units = percent probability from 1 – 100) for each month of the decades from 1900-1909 to 2000-2009 at 771 x 771 m spatial resolution. Each file represents a decadal average monthly mean. Version 1.0 was completed in 2015 using CMIP3. Version 2.0 was completed in 2018 using CMIP5. For more information on the methodology used to create this dataset, and guidelines for appropriate usage of the dataset, please see the data user's guide here: http://data.snap.uaf.edu/data/Base/AK_771m/historical/CRU_TS/snow_day_fraction/snow_fraction_data_users_guide.pdf

This set of files includes downscaled projections of decadal means of annual day of freeze or thaw (ordinal day of the year), and length of growing season (numbers of days, 0-365) for each decade from 2010 - 2100 at 2km x 2km meter spatial resolution. Each file represents a decadal mean of an annual mean calculated from mean monthly data. ---- The spatial extent includes Alaska, the Yukon Territory, British Columbia, Alberta, Saskatchewan, and Manitoba. Each set of files originates from one of five top ranked global circulation models from the CMIP5/AR5 models and RPCs, or is calculated as a 5 Model Average. Day of Freeze, Day of Thaw, Length of Growing Season calculations: Estimated ordinal days of freeze and thaw are calculated by assuming a linear change in temperature between consecutive months. Mean monthly temperatures are used to represent daily temperature on the 15th day of each month. When consecutive monthly midpoints have opposite sign temperatures, the day of transition (freeze or thaw) is the day between them on which temperature crosses zero degrees C. The length of growing season refers to the number of days between the days of thaw and freeze. This amounts to connecting temperature values (y-axis) for each month (x-axis) by line segments and solving for the x-intercepts. Calculating a day of freeze or thaw is simple. However, transitions may occur several times in a year, or not at all. The choice of transition points to use as the thaw and freeze dates which best represent realistic bounds on a growing season is more complex. Rather than iteratively looping over months one at a time, searching from January forward to determine thaw day and from December backward to determine freeze day, stopping as soon as a sign change between two months is identified, the algorithm looks at a snapshot of the signs of all twelve mean monthly temperatures at once, which enables identification of multiple discrete periods of positive and negative temperatures. As a result more realistic days of freeze and thaw and length of growing season can be calculated when there are idiosyncrasies in the data. Please note that these maps represent climatic estimates only. While we have based our work on scientifically accepted data and methods, uncertainty is always present . Uncertainty in model outputs tends to increase for more distant climatic estimates from present day for both historical summaries and future projections.