Info for files associated with the main analysis These files correspond to the output from the BARCAST code package (available at ftp://ftp.ncdc.noaa.gov/pub/data/paleo/softlib/barcast), as well as some manipulations to the posterior ensemble. Inputs to the code are discussed in the Inputs folder. The plotting routines require two additional Matlab packages: The m_map package is available at http://www.eos.ubc.ca/~rich/map.html. The Colormap and Colorbar packages are available on the Matlab File Exchange at: http://www.mathworks.com/matlabcentral/fileexchange/24371 Parameters_MCMC_Samples: 4000 samples of the 14 scalar parameters in BARCAST. The order is 1 = alpha, the AR(1) coefficient %2 = mu, the constant par in the linear mean of the AR(1) process %3 = sigma2, the partial sill in the spatial covariance matrix %4 = phi, the range parameter in the spatial covariance matrix %5 = tau2_I, the instrumental measurement error %6 = tau2_P, the measurement error, in the MXD observation equation. %7 = Beta_1, the scaling parameter for the MXD. %8 = Beta_0, the additive parameter for the MXD. %9 = tau2_P, the measurement error for the ice cores %10 = Beta_1, the scaling parameter for the ice cores.. %11 = Beta_0, the additive parameter for the ice cores. %12 = tau2_P, the measurement error for the log-varves. %13 = Beta_1, the scaling parameter for the log-varves. %14 = Beta_0, the additive parameter for the log-varves. Parameters_pers3: The 5th, 50th, and 95th percentiles of the scalar parameters. Temperature_Array_3pers: A 428 by 613 by 3 array. First index gives locations, second time (again, first year is the year prior to the first observation) and third the 5th, 50th, and 95the percentiles of the posterior ensemble members. Temperature_Array_XXpercentile are ASCII files for each of the 5th, 50th, and 95th percentiles. SM_Temperature_Array_3pers: As above, but the time series at all locations and for all ensemble members have been smoothed by nine point Hanning windows. BlockAve_All_MCMC_Samples: 4000 samples of the spatial average time series. BlockAve_pers3: 5th, 50th, and 95th percentiles of the spatially averaged temperature time series. Note that the timeline starts one year prior to the first observations, so in 1399. BlockAve_Smooth_pers3: As above, but each sample of the block average is first smoothed with a nine point Hanning window. PathWise_BlockAve_CI: 90% Pathwise uncertainty intervals for the block average time series, formed by inflating the 90% point-wise uncertainty intervals such that 90% of the posterior draws are entirely within the envelope. PathWise_BlockAve_SM_CI: As above, but for the smoothed block average. The following two files are used to construct the maps and histogram in Fig. 2: MaxYears_1400_2011, MinYears_1400_2011: Each is a 399 by 4000 matrix, where the first index gives the locations (only those corresponding to the target grid) and the second the ensemble member. Each entry is the year for which the ensemble member was warmest/coldest at each location. MaxYears_1400_2003, MinYears_1400_2003, MaxYears_1400_1994, MinYears_1400_1994, MaxYears_1400_1959, MinYears_1400_1959: As above, but curtailing the reconstruction in the indicated year. MinYears_1400_1994_ICE_VARVE, MaxYears_1400_1994_ICE_VARVE: As above, but for a reconstruction using only the ice core and varve time series. The following files contain information on centennial slopes: BlockAve_Slopes_win100: 4000 by 513 matrix of samples of the slopes of linear fits to 100-year sliding windows for the block average time series. The second dimensions indexes the year, with the first 100-year window beginning in 1400. Slopes_100win_IndsCentral_Subset_pers3.mat is a 399 by 2 by 3 array. The first dimension gives the locations (only those on the target grid), the second gives the 100-year window (1547-1646 then 1912-2011), and the third the 5th, 50th and 95the percentiles. The following provide the same information, but as 2D arrays saved as ascii files: Slopes_100win_IndsCentral_1547_05th Slopes_100win_IndsCentral_1547_50th Slopes_100win_IndsCentral_1547_95th Slopes_100win_IndsCentral_1912_05th Slopes_100win_IndsCentral_1912_50th Slopes_100win_IndsCentral_1912_95th There are also several Matlab .m files that produce a number of plots. Plot_TimeSeries.m : Plots the posterior median and 90% point-wise uncertainty intervals for the time series at two chosen locations, along with the process-level 90% uncertainty. FieldPlotter_4years.m plots the posterior median of the field for four selected years, along with the widths of the 90% point wise uncertainty intervals and the data availability. Plot_MaxMin_Field_and_Hist.m: For selected years and decades, plots the percentage of ensemble members for which those years/decades featured the warmest/coldest value. Also plots histograms of the occurrence of the maximum and minimum, binning across space and decades. Compare with Fig.2 from the main text. Slopes_plotter.m: Plots spatial maps of the centennial linear trends for the intervals featuring the largest rates of cooling and warming. The construction of Fig. 3 is detailed in the Fig3_Files folder.