Classification data of Mid-latitude Northwest Pacific

The clustering model used for the analysis in Sambe and Suga (2022) and the paper in prep, the training dataset used to create the model, and the dataset to which clustering was applied and labeled. The creation and classification of these models was done using pyXpcm(Maze, 2020). clutering_model.nc It contains the parameters of the Gaussian distribution representing each class and information on the principal components, in a format that can be used in pyXpcm. training_data.nc Profile information for 4544 points is stored. Dimensions are as follows. The names and coordinates in the data set are also noted. Pressure depth (DEPTH, -5~-1000, interval is 5) Number of profiles (N_PROF, 0~4543) The variables are as follows. The names in the data set and the dimensions on which they depend are also noted. Latitude (LATITUDE,(N_PROF)) Longitude (LONGITUDE,(N_PROF)) Observation date and time (TIME,(N_PROF)) Observation year (YEAR,(N_PROF)) Observation month (MONTH,(N_PROF)) Data name of profile (NAME,(N_PROF)) Temperature (TEMP,(N_PROF, DEPTH)) Salinity (PSAL,(N_PROF, DEPTH)) classified_data.nc Profile information for 57262 points is stored. Dimensions are as follows: Number of profiles (N_PROF, 0~57261) Class (pcm_class, 0~4) Pressure depth (DEPTH, -5~-1000, interval is 5) quantile (quantile, 0.05, 0.5, 0.95) The variables are as follows. The names in the data set and the dimensions on which they depend are also noted. Latitude (LATITUDE,(N_PROF)) Longitude (LONGITUDE,(N_PROF)) Observation date and time (TIME,(N_PROF)) Observation year (YEAR,(N_PROF)) Observation month (MONTH,(N_PROF)) Data name of profile (NAME,(N_PROF)) Temperature (TEMP,(N_PROF, DEPTH)) Salinity (PSAL,(N_PROF, DEPTH)) Assigned class (PCM_LABELS,(N_PROF)) Posterior probability for each class (PCM_POST,(pcm_class, N_PROF)) Temperature quantile for each class (TEMP_Q,(pcm_class, quantile, DEPTH)) Salinity quantile for each class (PSAL_Q,(pcm_class, quantile, DEPTH)) Labeling metric (robustness,(N_PROF)) Note that the class number starts with 0 in this data set. This means that the profiles belonging to Class 2 in the paper are those with PCM_LABELS=1 in this data set.