Rapid Arctic Warming (RAW) dataset and results

Dataset_S01 is an Excel file that describes the RAW and PAGES2k datasets used in this analysis. It includes relevant metadata detailing the site location (latitude, longitude, and elevation), proxy type (archive, proxy type, and category used in this analysis), temporal characteristics (duration, resolution, and number of detected shifts), paleoclimate interpretation (seasonality and climate sensitivity), and references. These metadata are explained in detail in the ReadMe tab. The RAW dataset folder includes the LiPD files (RAW_LiPDwithAgeModels.zip), which contain the age ensembles generated for this analysis. A CSV file with additional metadata (RAW_metadata_revised.csv) is also included. RAWcompositesArctic.RDS is an R data file that contains the composite reconstructions shown in Figure 2. Each reconstruction includes: the composite ensemble (time × number of ensemble members), a matrix of the mean proxy values for each record (age × number of proxy records), a matrix of the proxies used at each time step (age × number of proxy records), a LiPD timeseries object with the paleoclimate data, and a dataframe listing the proxies used for the composite, along with their metadata. The cptResults folder contains the output of the change point detection analysis. The shiftsEventDetection files list each detected event, including its age and magnitude ("delta"), for each proxy record. The shiftsRAW and shiftsPAGES2k files summarize these results for each dataset, using the summary bin time step, and include significance testing of each record against the robust null hypothesis. The Fig3_Data folder summarizes the site-specific results from the cptResults folder into a single time series (shown in Figure 3), using the multisite significance test described in the methods section. These files are saved as CSVs for easier viewing compared to the R data files in the cptResults folder. Each file includes columns for time, p-value, and confidence intervals. The net columns indicate whether the event is significant relative to the probability of any event occurring under the robust null hypothesis. The negative and positive columns indicate whether the event is significant relative to the probability of observing only a negative (cool) or positive (warm) shift, respectively.