NeuralGCM simulation datasets

Before running NeuralGCM yourself, consider if you can use one of our pre-computed simulation datasets. All are available in the Zarr format on Google Cloud Storage the under the Creative Commons CC-BY 4.0 license.

We currently have two types of datasets available:

• Medium-range weather forecasts for 2020, distributed as part of WeatherBench2:

  • 15 day weather forecasts for 2020 from a deterministic model at 0.7° resolution (NeuralGCM-0.7). Forecasts regridded to different resolutions are available at gs://weatherbench2/datasets/neuralgcm_deterministic: 2020-512x256.zarr , 2020-240x121_equiangular_with_poles_conservative.zarr, 2020-64x32_equiangular_conservative.zarr

  • 15 day weather forecasts for 2020 from a stochastic model at 1.4° resolution (NeuralGCM-ENS). Forecasts regridded to different resolutions are available at gs://weatherbench2/datasets/neuralgcm_ens : 2020-256x128.zarr, 2020-240x121_equiangular_with_poles_conservative.zarr, 2020-240x121_equiangular_with_poles_conservative_mean.zarr, 2020-64x32_equiangular_conservative.zarr, 2020-64x32_equiangular_conservative_mean.zarr

• Yearly to decadal atmosphere-only climate simulations:

  • 2-year runs from a deterministic model at 1.4° resolution (NeuralGCM-1.4). This dataset includes snapshots every 12 hours and is used in the NeuralGCM paper to evaluate the seasonal cycle (e.g., Figure 4a,i): gs://neuralgcm/amip_runs/v1_deterministic_1_4_deg/2019-to-2021_256x128_gauss_37-level_stride12h.zarr

  • 40-year simulations from a deterministic model at 2.8° resolution (NeuralGCM-2.8). This dataset includes snapshots every 6 hours and is used in the NeuralGCM paper to evaluate decadal (AMIP) runs: gs://neuralgcm/amip_runs/v1_deterministic_2_8_deg/1980-to-2020_128x64_gauss_37-level_stride6h.zarr

  • 20-year simulations from a stochastic NeuralGCM model with precipitation at 2.8° resolution. This dataset includes snapshots every 3 hours and accumulated precipitation. It is described in an arXiv pre-print: gs://neuralgcm/amip_runs/v1_precip_stochastic_2_8_deg/2001-to-2021_128x64_gauss_37-level_stride3h.zarr

Below, we show how to open and inspect these datsets. You’ll need Xarray, Zarr and GCSFS installed:

! pip install -q -U zarr xarray gcsfs
import xarray
import zarr

Details on coordinate axes:

• time: The date and time when the simulation was initialized.

• prediction_timedelta: Time delta (stored as timedelta64[ns]) since the simulation was initialized.

• longitude: Longitude in degrees.

• latitude: Latitude in degrees.

• level: Pressure level (in hPa) as the vertical coordinate.

Medium-range weather forecasts for 2020

15 day weather forecasts from the 0.7° deterministic model (NeuralGCM-0.7)

Corresponding model checkpoint: gs://neuralgcm/models/v1/deterministic_0_7_deg.pkl

# output available at different resolutions:
# `2020-512x256.zarr`  (original)
# `2020-240x121_equiangular_with_poles_conservative.zarr`
# `2020-64x32_equiangular_conservative.zarr`
path = 'gs://weatherbench2/datasets/neuralgcm_deterministic/2020-512x256.zarr'
ngcm_0_7 = xarray.open_zarr(path, storage_options=dict(token='anon'))
ngcm_0_7

<xarray.Dataset> Size: 4TB
Dimensions:                              (time: 797, prediction_timedelta: 31,
                                          longitude: 512, latitude: 256,
                                          level: 37)
Coordinates:
  * latitude                             (latitude) float64 2kB -89.46 ... 89.46
  * level                                (level) int64 296B 1 2 3 ... 975 1000
  * longitude                            (longitude) float64 4kB 0.0 ... 359.3
  * prediction_timedelta                 (prediction_timedelta) timedelta64[ns] 248B ...
  * time                                 (time) datetime64[ns] 6kB 2019-12-15...
Data variables:
    P_minus_E_cumulative                 (time, prediction_timedelta, longitude, latitude) float32 13GB dask.array<chunksize=(1, 1, 512, 256), meta=np.ndarray>
    geopotential                         (time, prediction_timedelta, level, longitude, latitude) float32 479GB dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>
    specific_cloud_ice_water_content     (time, prediction_timedelta, level, longitude, latitude) float32 479GB dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>
    specific_cloud_liquid_water_content  (time, prediction_timedelta, level, longitude, latitude) float32 479GB dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>
    specific_humidity                    (time, prediction_timedelta, level, longitude, latitude) float32 479GB dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>
    temperature                          (time, prediction_timedelta, level, longitude, latitude) float32 479GB dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>
    u_component_of_wind                  (time, prediction_timedelta, level, longitude, latitude) float32 479GB dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>
    v_component_of_wind                  (time, prediction_timedelta, level, longitude, latitude) float32 479GB dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>
    wind_speed                           (time, prediction_timedelta, level, longitude, latitude) float32 479GB dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>
Attributes:
    experiment_id:  67001173
    worker_id:      1

xarray.Dataset

• Dimensions:
  • time: 797
  • prediction_timedelta: 31
  • longitude: 512
  • latitude: 256
  • level: 37
• Coordinates: (5)
  • latitude
    (latitude)
    float64
    -89.46 -88.77 ... 88.77 89.46

    array([-89.462822, -88.766951, -88.066972, ...,  88.066972,  88.766951,
            89.462822])

  • level
    (level)
    int64
    1 2 3 5 7 ... 900 925 950 975 1000

    array([   1,    2,    3,    5,    7,   10,   20,   30,   50,   70,  100,  125,
            150,  175,  200,  225,  250,  300,  350,  400,  450,  500,  550,  600,
            650,  700,  750,  775,  800,  825,  850,  875,  900,  925,  950,  975,
           1000])

  • longitude
    (longitude)
    float64
    0.0 0.7031 1.406 ... 358.6 359.3

    array([  0.      ,   0.703125,   1.40625 , ..., 357.890625, 358.59375 ,
           359.296875])

  • prediction_timedelta
    (prediction_timedelta)
    timedelta64[ns]
    0 days 00:00:00 ... 15 days 00:0...

    array([               0,   43200000000000,   86400000000000,  129600000000000,
            172800000000000,  216000000000000,  259200000000000,  302400000000000,
            345600000000000,  388800000000000,  432000000000000,  475200000000000,
            518400000000000,  561600000000000,  604800000000000,  648000000000000,
            691200000000000,  734400000000000,  777600000000000,  820800000000000,
            864000000000000,  907200000000000,  950400000000000,  993600000000000,
           1036800000000000, 1080000000000000, 1123200000000000, 1166400000000000,
           1209600000000000, 1252800000000000, 1296000000000000],
          dtype='timedelta64[ns]')

  • time
    (time)
    datetime64[ns]
    2019-12-15 ... 2021-01-16

    array(['2019-12-15T00:00:00.000000000', '2019-12-15T12:00:00.000000000',
           '2019-12-16T00:00:00.000000000', ..., '2021-01-15T00:00:00.000000000',
           '2021-01-15T12:00:00.000000000', '2021-01-16T00:00:00.000000000'],
          dtype='datetime64[ns]')

• Data variables: (9)
  • P_minus_E_cumulative
    (time, prediction_timedelta, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 512, 256), meta=np.ndarray>

    Array Chunk Bytes 12.06 GiB 512.00 kiB Shape (797, 31, 512, 256) (1, 1, 512, 256) Dask graph 24707 chunks in 2 graph layers Data type float32 numpy.ndarray

  • geopotential
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>

    Array Chunk Bytes 446.37 GiB 18.50 MiB Shape (797, 31, 37, 512, 256) (1, 1, 37, 512, 256) Dask graph 24707 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_cloud_ice_water_content
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>

    Array Chunk Bytes 446.37 GiB 18.50 MiB Shape (797, 31, 37, 512, 256) (1, 1, 37, 512, 256) Dask graph 24707 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_cloud_liquid_water_content
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>

    Array Chunk Bytes 446.37 GiB 18.50 MiB Shape (797, 31, 37, 512, 256) (1, 1, 37, 512, 256) Dask graph 24707 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_humidity
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>

    Array Chunk Bytes 446.37 GiB 18.50 MiB Shape (797, 31, 37, 512, 256) (1, 1, 37, 512, 256) Dask graph 24707 chunks in 2 graph layers Data type float32 numpy.ndarray

  • temperature
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>

    Array Chunk Bytes 446.37 GiB 18.50 MiB Shape (797, 31, 37, 512, 256) (1, 1, 37, 512, 256) Dask graph 24707 chunks in 2 graph layers Data type float32 numpy.ndarray

  • u_component_of_wind
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>

    Array Chunk Bytes 446.37 GiB 18.50 MiB Shape (797, 31, 37, 512, 256) (1, 1, 37, 512, 256) Dask graph 24707 chunks in 2 graph layers Data type float32 numpy.ndarray

  • v_component_of_wind
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>

    Array Chunk Bytes 446.37 GiB 18.50 MiB Shape (797, 31, 37, 512, 256) (1, 1, 37, 512, 256) Dask graph 24707 chunks in 2 graph layers Data type float32 numpy.ndarray

  • wind_speed
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 512, 256), meta=np.ndarray>

    Array Chunk Bytes 446.37 GiB 18.50 MiB Shape (797, 31, 37, 512, 256) (1, 1, 37, 512, 256) Dask graph 24707 chunks in 2 graph layers Data type float32 numpy.ndarray

• Indexes: (5)
  • latitude
    PandasIndex

    PandasIndex(Index([-89.46282156857811, -88.76695135284248, -88.06697164743062,
           -87.36606343308205, -86.66480301344076, -85.96337216088044,
            -85.2618460607125, -84.56026138305344, -83.85863812860755,
           -83.15698812854168,
           ...
            83.15698812854168,  83.85863812860755,  84.56026138305344,
             85.2618460607125,  85.96337216088044,  86.66480301344076,
            87.36606343308205,  88.06697164743062,  88.76695135284248,
            89.46282156857811],
          dtype='float64', name='latitude', length=256))

  • level
    PandasIndex

    PandasIndex(Index([   1,    2,    3,    5,    7,   10,   20,   30,   50,   70,  100,  125,
            150,  175,  200,  225,  250,  300,  350,  400,  450,  500,  550,  600,
            650,  700,  750,  775,  800,  825,  850,  875,  900,  925,  950,  975,
           1000],
          dtype='int64', name='level'))

  • longitude
    PandasIndex

    PandasIndex(Index([               0.0,           0.703125,            1.40625,
                     2.109375,             2.8125,           3.515625,
                      4.21875,           4.921875,              5.625,
            6.328124999999999,
           ...
           352.96875000000006,         353.671875,            354.375,
                   355.078125,          355.78125, 356.48437499999994,
           357.18750000000006,         357.890625,          358.59375,
                   359.296875],
          dtype='float64', name='longitude', length=512))

  • prediction_timedelta
    PandasIndex

    PandasIndex(TimedeltaIndex([ '0 days 00:00:00',  '0 days 12:00:00',  '1 days 00:00:00',
                     '1 days 12:00:00',  '2 days 00:00:00',  '2 days 12:00:00',
                     '3 days 00:00:00',  '3 days 12:00:00',  '4 days 00:00:00',
                     '4 days 12:00:00',  '5 days 00:00:00',  '5 days 12:00:00',
                     '6 days 00:00:00',  '6 days 12:00:00',  '7 days 00:00:00',
                     '7 days 12:00:00',  '8 days 00:00:00',  '8 days 12:00:00',
                     '9 days 00:00:00',  '9 days 12:00:00', '10 days 00:00:00',
                    '10 days 12:00:00', '11 days 00:00:00', '11 days 12:00:00',
                    '12 days 00:00:00', '12 days 12:00:00', '13 days 00:00:00',
                    '13 days 12:00:00', '14 days 00:00:00', '14 days 12:00:00',
                    '15 days 00:00:00'],
                   dtype='timedelta64[ns]', name='prediction_timedelta', freq=None))

  • time
    PandasIndex

    PandasIndex(DatetimeIndex(['2019-12-15 00:00:00', '2019-12-15 12:00:00',
                   '2019-12-16 00:00:00', '2019-12-16 12:00:00',
                   '2019-12-17 00:00:00', '2019-12-17 12:00:00',
                   '2019-12-18 00:00:00', '2019-12-18 12:00:00',
                   '2019-12-19 00:00:00', '2019-12-19 12:00:00',
                   ...
                   '2021-01-11 12:00:00', '2021-01-12 00:00:00',
                   '2021-01-12 12:00:00', '2021-01-13 00:00:00',
                   '2021-01-13 12:00:00', '2021-01-14 00:00:00',
                   '2021-01-14 12:00:00', '2021-01-15 00:00:00',
                   '2021-01-15 12:00:00', '2021-01-16 00:00:00'],
                  dtype='datetime64[ns]', name='time', length=797, freq=None))

• Attributes: (2)

  experiment_id :

  67001173

  worker_id :

  1

15 day weather forecasts from the 1.4° stochastic model (NeuralGCM-ENS)

Corresponding model checkpoint: gs://neuralgcm/models/v1/stochastic_1_4_deg.pkl

# output available at different resolutions:
# `2020-256x128.zarr`  (original)
# `2020-240x121_equiangular_with_poles_conservative.zarr`
# `2020-240x121_equiangular_with_poles_conservative_mean.zarr`
# `2020-64x32_equiangular_conservative.zarr`
# `2020-64x32_equiangular_conservative_mean.zarr
path = 'gs://weatherbench2/datasets/neuralgcm_ens/2020-256x128.zarr'
ngcm_ens = xarray.open_zarr(path, storage_options=dict(token='anon'))
ngcm_ens

<xarray.Dataset> Size: 49TB
Dimensions:                              (realization: 50, time: 797,
                                          prediction_timedelta: 32, level: 37,
                                          longitude: 256, latitude: 128)
Coordinates:
  * latitude                             (latitude) float64 1kB -88.93 ... 88.93
  * level                                (level) int64 296B 1 2 3 ... 975 1000
  * longitude                            (longitude) float64 2kB 0.0 ... 358.6
  * prediction_timedelta                 (prediction_timedelta) timedelta64[ns] 256B ...
  * realization                          (realization) int64 400B 0 1 ... 48 49
  * time                                 (time) datetime64[ns] 6kB 2019-12-15...
Data variables:
    geopotential                         (realization, time, prediction_timedelta, level, longitude, latitude) float32 6TB dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>
    specific_cloud_ice_water_content     (realization, time, prediction_timedelta, level, longitude, latitude) float32 6TB dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>
    specific_cloud_liquid_water_content  (realization, time, prediction_timedelta, level, longitude, latitude) float32 6TB dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>
    specific_humidity                    (realization, time, prediction_timedelta, level, longitude, latitude) float32 6TB dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>
    temperature                          (realization, time, prediction_timedelta, level, longitude, latitude) float32 6TB dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>
    u_component_of_wind                  (realization, time, prediction_timedelta, level, longitude, latitude) float32 6TB dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>
    v_component_of_wind                  (realization, time, prediction_timedelta, level, longitude, latitude) float32 6TB dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>
    wind_speed                           (realization, time, prediction_timedelta, level, longitude, latitude) float32 6TB dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>
Attributes:
    experiment_id:  73974210
    worker_id:      3

xarray.Dataset

• Dimensions:
  • realization: 50
  • time: 797
  • prediction_timedelta: 32
  • level: 37
  • longitude: 256
  • latitude: 128
• Coordinates: (6)
  • latitude
    (latitude)
    float64
    -88.93 -87.54 ... 87.54 88.93

    array([-88.927735, -87.538705, -86.141472, -84.742386, -83.342596, -81.942466,
           -80.542146, -79.14171 , -77.741196, -76.340629, -74.940023, -73.539389,
           -72.138732, -70.738059, -69.337372, -67.936673, -66.535966, -65.135251,
           -63.73453 , -62.333803, -60.933072, -59.532337, -58.131598, -56.730857,
           -55.330112, -53.929366, -52.528617, -51.127867, -49.727115, -48.326361,
           -46.925606, -45.52485 , -44.124093, -42.723335, -41.322576, -39.921816,
           -38.521056, -37.120294, -35.719532, -34.31877 , -32.918007, -31.517244,
           -30.11648 , -28.715716, -27.314951, -25.914186, -24.513421, -23.112655,
           -21.71189 , -20.311124, -18.910357, -17.509591, -16.108824, -14.708057,
           -13.30729 , -11.906523, -10.505756,  -9.104989,  -7.704221,  -6.303454,
            -4.902687,  -3.501919,  -2.101151,  -0.700384,   0.700384,   2.101151,
             3.501919,   4.902687,   6.303454,   7.704221,   9.104989,  10.505756,
            11.906523,  13.30729 ,  14.708057,  16.108824,  17.509591,  18.910357,
            20.311124,  21.71189 ,  23.112655,  24.513421,  25.914186,  27.314951,
            28.715716,  30.11648 ,  31.517244,  32.918007,  34.31877 ,  35.719532,
            37.120294,  38.521056,  39.921816,  41.322576,  42.723335,  44.124093,
            45.52485 ,  46.925606,  48.326361,  49.727115,  51.127867,  52.528617,
            53.929366,  55.330112,  56.730857,  58.131598,  59.532337,  60.933072,
            62.333803,  63.73453 ,  65.135251,  66.535966,  67.936673,  69.337372,
            70.738059,  72.138732,  73.539389,  74.940023,  76.340629,  77.741196,
            79.14171 ,  80.542146,  81.942466,  83.342596,  84.742386,  86.141472,
            87.538705,  88.927735])

  • level
    (level)
    int64
    1 2 3 5 7 ... 900 925 950 975 1000

    array([   1,    2,    3,    5,    7,   10,   20,   30,   50,   70,  100,  125,
            150,  175,  200,  225,  250,  300,  350,  400,  450,  500,  550,  600,
            650,  700,  750,  775,  800,  825,  850,  875,  900,  925,  950,  975,
           1000])

  • longitude
    (longitude)
    float64
    0.0 1.406 2.812 ... 357.2 358.6

    array([  0.     ,   1.40625,   2.8125 , ..., 355.78125, 357.1875 , 358.59375])

  • prediction_timedelta
    (prediction_timedelta)
    timedelta64[ns]
    0 days 00:00:00 ... 15 days 12:0...

    array([               0,   43200000000000,   86400000000000,  129600000000000,
            172800000000000,  216000000000000,  259200000000000,  302400000000000,
            345600000000000,  388800000000000,  432000000000000,  475200000000000,
            518400000000000,  561600000000000,  604800000000000,  648000000000000,
            691200000000000,  734400000000000,  777600000000000,  820800000000000,
            864000000000000,  907200000000000,  950400000000000,  993600000000000,
           1036800000000000, 1080000000000000, 1123200000000000, 1166400000000000,
           1209600000000000, 1252800000000000, 1296000000000000, 1339200000000000],
          dtype='timedelta64[ns]')

  • realization
    (realization)
    int64
    0 1 2 3 4 5 6 ... 44 45 46 47 48 49

    array([ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16, 17,
           18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35,
           36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49])

  • time
    (time)
    datetime64[ns]
    2019-12-15 ... 2021-01-16

    array(['2019-12-15T00:00:00.000000000', '2019-12-15T12:00:00.000000000',
           '2019-12-16T00:00:00.000000000', ..., '2021-01-15T00:00:00.000000000',
           '2021-01-15T12:00:00.000000000', '2021-01-16T00:00:00.000000000'],
          dtype='datetime64[ns]')

• Data variables: (8)
  • geopotential
    (realization, time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 5.62 TiB 12.50 MiB Shape (50, 797, 32, 37, 256, 128) (50, 1, 2, 1, 256, 128) Dask graph 471824 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_cloud_ice_water_content
    (realization, time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 5.62 TiB 12.50 MiB Shape (50, 797, 32, 37, 256, 128) (50, 1, 2, 1, 256, 128) Dask graph 471824 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_cloud_liquid_water_content
    (realization, time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 5.62 TiB 12.50 MiB Shape (50, 797, 32, 37, 256, 128) (50, 1, 2, 1, 256, 128) Dask graph 471824 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_humidity
    (realization, time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 5.62 TiB 12.50 MiB Shape (50, 797, 32, 37, 256, 128) (50, 1, 2, 1, 256, 128) Dask graph 471824 chunks in 2 graph layers Data type float32 numpy.ndarray

  • temperature
    (realization, time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 5.62 TiB 12.50 MiB Shape (50, 797, 32, 37, 256, 128) (50, 1, 2, 1, 256, 128) Dask graph 471824 chunks in 2 graph layers Data type float32 numpy.ndarray

  • u_component_of_wind
    (realization, time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 5.62 TiB 12.50 MiB Shape (50, 797, 32, 37, 256, 128) (50, 1, 2, 1, 256, 128) Dask graph 471824 chunks in 2 graph layers Data type float32 numpy.ndarray

  • v_component_of_wind
    (realization, time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 5.62 TiB 12.50 MiB Shape (50, 797, 32, 37, 256, 128) (50, 1, 2, 1, 256, 128) Dask graph 471824 chunks in 2 graph layers Data type float32 numpy.ndarray

  • wind_speed
    (realization, time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(50, 1, 2, 1, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 5.62 TiB 12.50 MiB Shape (50, 797, 32, 37, 256, 128) (50, 1, 2, 1, 256, 128) Dask graph 471824 chunks in 2 graph layers Data type float32 numpy.ndarray

• Indexes: (6)
  • latitude
    PandasIndex

    PandasIndex(Index([-88.92773535229591, -87.53870521302731, -86.14147210152788,
           -84.74238559071426, -83.34259604407038, -81.94246629917325,
           -80.54214643461712,  -79.1417096486217, -77.74119586551384,
           -76.34062870237155,
           ...
            76.34062870237155,  77.74119586551384,   79.1417096486217,
            80.54214643461712,  81.94246629917325,  83.34259604407038,
            84.74238559071426,  86.14147210152788,  87.53870521302731,
            88.92773535229591],
          dtype='float64', name='latitude', length=128))

  • level
    PandasIndex

    PandasIndex(Index([   1,    2,    3,    5,    7,   10,   20,   30,   50,   70,  100,  125,
            150,  175,  200,  225,  250,  300,  350,  400,  450,  500,  550,  600,
            650,  700,  750,  775,  800,  825,  850,  875,  900,  925,  950,  975,
           1000],
          dtype='int64', name='level'))

  • longitude
    PandasIndex

    PandasIndex(Index([               0.0,            1.40625,             2.8125,
                      4.21875,              5.625,            7.03125,
                       8.4375,            9.84375,              11.25,
           12.656249999999998,
           ...
                     345.9375, 347.34374999999994,             348.75,
                    350.15625, 351.56249999999994, 352.96875000000006,
                      354.375,          355.78125, 357.18750000000006,
                    358.59375],
          dtype='float64', name='longitude', length=256))

  • prediction_timedelta
    PandasIndex

    PandasIndex(TimedeltaIndex([ '0 days 00:00:00',  '0 days 12:00:00',  '1 days 00:00:00',
                     '1 days 12:00:00',  '2 days 00:00:00',  '2 days 12:00:00',
                     '3 days 00:00:00',  '3 days 12:00:00',  '4 days 00:00:00',
                     '4 days 12:00:00',  '5 days 00:00:00',  '5 days 12:00:00',
                     '6 days 00:00:00',  '6 days 12:00:00',  '7 days 00:00:00',
                     '7 days 12:00:00',  '8 days 00:00:00',  '8 days 12:00:00',
                     '9 days 00:00:00',  '9 days 12:00:00', '10 days 00:00:00',
                    '10 days 12:00:00', '11 days 00:00:00', '11 days 12:00:00',
                    '12 days 00:00:00', '12 days 12:00:00', '13 days 00:00:00',
                    '13 days 12:00:00', '14 days 00:00:00', '14 days 12:00:00',
                    '15 days 00:00:00', '15 days 12:00:00'],
                   dtype='timedelta64[ns]', name='prediction_timedelta', freq=None))

  • realization
    PandasIndex

    PandasIndex(Index([ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16, 17,
           18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35,
           36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49],
          dtype='int64', name='realization'))

  • time
    PandasIndex

    PandasIndex(DatetimeIndex(['2019-12-15 00:00:00', '2019-12-15 12:00:00',
                   '2019-12-16 00:00:00', '2019-12-16 12:00:00',
                   '2019-12-17 00:00:00', '2019-12-17 12:00:00',
                   '2019-12-18 00:00:00', '2019-12-18 12:00:00',
                   '2019-12-19 00:00:00', '2019-12-19 12:00:00',
                   ...
                   '2021-01-11 12:00:00', '2021-01-12 00:00:00',
                   '2021-01-12 12:00:00', '2021-01-13 00:00:00',
                   '2021-01-13 12:00:00', '2021-01-14 00:00:00',
                   '2021-01-14 12:00:00', '2021-01-15 00:00:00',
                   '2021-01-15 12:00:00', '2021-01-16 00:00:00'],
                  dtype='datetime64[ns]', name='time', length=797, freq=None))

• Attributes: (2)

  experiment_id :

  73974210

  worker_id :

  3

Yearly to decadal climate simulations

2-year 1.4 degree determinisic NeuralGCM model

Corresponding model checkpoint: gs://neuralgcm/models/v1/deterministic_1_4_deg.pkl

path = 'gs://neuralgcm/amip_runs/v1_deterministic_1_4_deg/2019-to-2021_256x128_gauss_37-level_stride12h.zarr/'
ngcm_1_4 = xarray.open_zarr(path, storage_options=dict(token='anon'))
ngcm_1_4

<xarray.Dataset> Size: 2TB
Dimensions:                              (time: 37, prediction_timedelta: 1461,
                                          longitude: 256, latitude: 128,
                                          level: 37)
Coordinates:
  * latitude                             (latitude) float64 1kB -88.93 ... 88.93
  * level                                (level) int64 296B 1 2 3 ... 975 1000
  * longitude                            (longitude) float64 2kB 0.0 ... 358.6
  * prediction_timedelta                 (prediction_timedelta) timedelta64[ns] 12kB ...
  * time                                 (time) datetime64[ns] 296B 2019-01-0...
Data variables:
    P_minus_E_cumulative                 (time, prediction_timedelta, longitude, latitude) float32 7GB dask.array<chunksize=(1, 1, 256, 128), meta=np.ndarray>
    geopotential                         (time, prediction_timedelta, level, longitude, latitude) float32 262GB dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>
    specific_cloud_ice_water_content     (time, prediction_timedelta, level, longitude, latitude) float32 262GB dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>
    specific_cloud_liquid_water_content  (time, prediction_timedelta, level, longitude, latitude) float32 262GB dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>
    specific_humidity                    (time, prediction_timedelta, level, longitude, latitude) float32 262GB dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>
    temperature                          (time, prediction_timedelta, level, longitude, latitude) float32 262GB dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>
    u_component_of_wind                  (time, prediction_timedelta, level, longitude, latitude) float32 262GB dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>
    v_component_of_wind                  (time, prediction_timedelta, level, longitude, latitude) float32 262GB dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>
Attributes:
    experiment_id:  66950211
    worker_id:      1

xarray.Dataset

• Dimensions:
  • time: 37
  • prediction_timedelta: 1461
  • longitude: 256
  • latitude: 128
  • level: 37
• Coordinates: (5)
  • latitude
    (latitude)
    float64
    -88.93 -87.54 ... 87.54 88.93

    array([-88.927735, -87.538705, -86.141472, -84.742386, -83.342596, -81.942466,
           -80.542146, -79.14171 , -77.741196, -76.340629, -74.940023, -73.539389,
           -72.138732, -70.738059, -69.337372, -67.936673, -66.535966, -65.135251,
           -63.73453 , -62.333803, -60.933072, -59.532337, -58.131598, -56.730857,
           -55.330112, -53.929366, -52.528617, -51.127867, -49.727115, -48.326361,
           -46.925606, -45.52485 , -44.124093, -42.723335, -41.322576, -39.921816,
           -38.521056, -37.120294, -35.719532, -34.31877 , -32.918007, -31.517244,
           -30.11648 , -28.715716, -27.314951, -25.914186, -24.513421, -23.112655,
           -21.71189 , -20.311124, -18.910357, -17.509591, -16.108824, -14.708057,
           -13.30729 , -11.906523, -10.505756,  -9.104989,  -7.704221,  -6.303454,
            -4.902687,  -3.501919,  -2.101151,  -0.700384,   0.700384,   2.101151,
             3.501919,   4.902687,   6.303454,   7.704221,   9.104989,  10.505756,
            11.906523,  13.30729 ,  14.708057,  16.108824,  17.509591,  18.910357,
            20.311124,  21.71189 ,  23.112655,  24.513421,  25.914186,  27.314951,
            28.715716,  30.11648 ,  31.517244,  32.918007,  34.31877 ,  35.719532,
            37.120294,  38.521056,  39.921816,  41.322576,  42.723335,  44.124093,
            45.52485 ,  46.925606,  48.326361,  49.727115,  51.127867,  52.528617,
            53.929366,  55.330112,  56.730857,  58.131598,  59.532337,  60.933072,
            62.333803,  63.73453 ,  65.135251,  66.535966,  67.936673,  69.337372,
            70.738059,  72.138732,  73.539389,  74.940023,  76.340629,  77.741196,
            79.14171 ,  80.542146,  81.942466,  83.342596,  84.742386,  86.141472,
            87.538705,  88.927735])

  • level
    (level)
    int64
    1 2 3 5 7 ... 900 925 950 975 1000

    array([   1,    2,    3,    5,    7,   10,   20,   30,   50,   70,  100,  125,
            150,  175,  200,  225,  250,  300,  350,  400,  450,  500,  550,  600,
            650,  700,  750,  775,  800,  825,  850,  875,  900,  925,  950,  975,
           1000])

  • longitude
    (longitude)
    float64
    0.0 1.406 2.812 ... 357.2 358.6

    array([  0.     ,   1.40625,   2.8125 , ..., 355.78125, 357.1875 , 358.59375])

  • prediction_timedelta
    (prediction_timedelta)
    timedelta64[ns]
    0 days 00:00:00 ... 730 days 00:...

    array([                0,    43200000000000,    86400000000000, ...,
           62985600000000000, 63028800000000000, 63072000000000000],
          dtype='timedelta64[ns]')

  • time
    (time)
    datetime64[ns]
    2019-01-01 ... 2019-12-27

    array(['2019-01-01T00:00:00.000000000', '2019-01-11T00:00:00.000000000',
           '2019-01-21T00:00:00.000000000', '2019-01-31T00:00:00.000000000',
           '2019-02-10T00:00:00.000000000', '2019-02-20T00:00:00.000000000',
           '2019-03-02T00:00:00.000000000', '2019-03-12T00:00:00.000000000',
           '2019-03-22T00:00:00.000000000', '2019-04-01T00:00:00.000000000',
           '2019-04-11T00:00:00.000000000', '2019-04-21T00:00:00.000000000',
           '2019-05-01T00:00:00.000000000', '2019-05-11T00:00:00.000000000',
           '2019-05-21T00:00:00.000000000', '2019-05-31T00:00:00.000000000',
           '2019-06-10T00:00:00.000000000', '2019-06-20T00:00:00.000000000',
           '2019-06-30T00:00:00.000000000', '2019-07-10T00:00:00.000000000',
           '2019-07-20T00:00:00.000000000', '2019-07-30T00:00:00.000000000',
           '2019-08-09T00:00:00.000000000', '2019-08-19T00:00:00.000000000',
           '2019-08-29T00:00:00.000000000', '2019-09-08T00:00:00.000000000',
           '2019-09-18T00:00:00.000000000', '2019-09-28T00:00:00.000000000',
           '2019-10-08T00:00:00.000000000', '2019-10-18T00:00:00.000000000',
           '2019-10-28T00:00:00.000000000', '2019-11-07T00:00:00.000000000',
           '2019-11-17T00:00:00.000000000', '2019-11-27T00:00:00.000000000',
           '2019-12-07T00:00:00.000000000', '2019-12-17T00:00:00.000000000',
           '2019-12-27T00:00:00.000000000'], dtype='datetime64[ns]')

• Data variables: (8)
  • P_minus_E_cumulative
    (time, prediction_timedelta, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 6.60 GiB 128.00 kiB Shape (37, 1461, 256, 128) (1, 1, 256, 128) Dask graph 54057 chunks in 2 graph layers Data type float32 numpy.ndarray

  • geopotential
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 244.15 GiB 4.62 MiB Shape (37, 1461, 37, 256, 128) (1, 1, 37, 256, 128) Dask graph 54057 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_cloud_ice_water_content
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 244.15 GiB 4.62 MiB Shape (37, 1461, 37, 256, 128) (1, 1, 37, 256, 128) Dask graph 54057 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_cloud_liquid_water_content
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 244.15 GiB 4.62 MiB Shape (37, 1461, 37, 256, 128) (1, 1, 37, 256, 128) Dask graph 54057 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_humidity
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 244.15 GiB 4.62 MiB Shape (37, 1461, 37, 256, 128) (1, 1, 37, 256, 128) Dask graph 54057 chunks in 2 graph layers Data type float32 numpy.ndarray

  • temperature
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 244.15 GiB 4.62 MiB Shape (37, 1461, 37, 256, 128) (1, 1, 37, 256, 128) Dask graph 54057 chunks in 2 graph layers Data type float32 numpy.ndarray

  • u_component_of_wind
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 244.15 GiB 4.62 MiB Shape (37, 1461, 37, 256, 128) (1, 1, 37, 256, 128) Dask graph 54057 chunks in 2 graph layers Data type float32 numpy.ndarray

  • v_component_of_wind
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 256, 128), meta=np.ndarray>

    Array Chunk Bytes 244.15 GiB 4.62 MiB Shape (37, 1461, 37, 256, 128) (1, 1, 37, 256, 128) Dask graph 54057 chunks in 2 graph layers Data type float32 numpy.ndarray

• Indexes: (5)
  • latitude
    PandasIndex

    PandasIndex(Index([-88.92773535229591, -87.53870521302731, -86.14147210152788,
           -84.74238559071426, -83.34259604407038, -81.94246629917325,
           -80.54214643461712,  -79.1417096486217, -77.74119586551384,
           -76.34062870237155,
           ...
            76.34062870237155,  77.74119586551384,   79.1417096486217,
            80.54214643461712,  81.94246629917325,  83.34259604407038,
            84.74238559071426,  86.14147210152788,  87.53870521302731,
            88.92773535229591],
          dtype='float64', name='latitude', length=128))

  • level
    PandasIndex

    PandasIndex(Index([   1,    2,    3,    5,    7,   10,   20,   30,   50,   70,  100,  125,
            150,  175,  200,  225,  250,  300,  350,  400,  450,  500,  550,  600,
            650,  700,  750,  775,  800,  825,  850,  875,  900,  925,  950,  975,
           1000],
          dtype='int64', name='level'))

  • longitude
    PandasIndex

    PandasIndex(Index([               0.0,            1.40625,             2.8125,
                      4.21875,              5.625,            7.03125,
                       8.4375,            9.84375,              11.25,
           12.656249999999998,
           ...
                     345.9375, 347.34374999999994,             348.75,
                    350.15625, 351.56249999999994, 352.96875000000006,
                      354.375,          355.78125, 357.18750000000006,
                    358.59375],
          dtype='float64', name='longitude', length=256))

  • prediction_timedelta
    PandasIndex

    PandasIndex(TimedeltaIndex([  '0 days 00:00:00',   '0 days 12:00:00',   '1 days 00:00:00',
                      '1 days 12:00:00',   '2 days 00:00:00',   '2 days 12:00:00',
                      '3 days 00:00:00',   '3 days 12:00:00',   '4 days 00:00:00',
                      '4 days 12:00:00',
                    ...
                    '725 days 12:00:00', '726 days 00:00:00', '726 days 12:00:00',
                    '727 days 00:00:00', '727 days 12:00:00', '728 days 00:00:00',
                    '728 days 12:00:00', '729 days 00:00:00', '729 days 12:00:00',
                    '730 days 00:00:00'],
                   dtype='timedelta64[ns]', name='prediction_timedelta', length=1461, freq=None))

  • time
    PandasIndex

    PandasIndex(DatetimeIndex(['2019-01-01', '2019-01-11', '2019-01-21', '2019-01-31',
                   '2019-02-10', '2019-02-20', '2019-03-02', '2019-03-12',
                   '2019-03-22', '2019-04-01', '2019-04-11', '2019-04-21',
                   '2019-05-01', '2019-05-11', '2019-05-21', '2019-05-31',
                   '2019-06-10', '2019-06-20', '2019-06-30', '2019-07-10',
                   '2019-07-20', '2019-07-30', '2019-08-09', '2019-08-19',
                   '2019-08-29', '2019-09-08', '2019-09-18', '2019-09-28',
                   '2019-10-08', '2019-10-18', '2019-10-28', '2019-11-07',
                   '2019-11-17', '2019-11-27', '2019-12-07', '2019-12-17',
                   '2019-12-27'],
                  dtype='datetime64[ns]', name='time', freq=None))

• Attributes: (2)

  experiment_id :

  66950211

  worker_id :

  1

40-year runs 2.8 degree determinitic NeuralGCM model

Corresponding model checkpoint: gs://neuralgcm/models/v1/deterministic_2_8_deg.pkl

path = 'gs://neuralgcm/amip_runs/v1_deterministic_2_8_deg/1980-to-2020_128x64_gauss_37-level_stride6h.zarr/'
ngcm_2_8 = xarray.open_zarr(path, storage_options=dict(token='anon'))
ngcm_2_8

<xarray.Dataset> Size: 18TB
Dimensions:                              (time: 37,
                                          prediction_timedelta: 58449,
                                          level: 37, longitude: 128,
                                          latitude: 64)
Coordinates:
  * latitude                             (latitude) float64 512B -87.86 ... 8...
  * level                                (level) int64 296B 1 2 3 ... 975 1000
  * longitude                            (longitude) float64 1kB 0.0 ... 357.2
  * prediction_timedelta                 (prediction_timedelta) timedelta64[ns] 468kB ...
  * time                                 (time) datetime64[ns] 296B 1980-01-0...
Data variables:
    geopotential                         (time, prediction_timedelta, level, longitude, latitude) float32 3TB dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>
    specific_cloud_ice_water_content     (time, prediction_timedelta, level, longitude, latitude) float32 3TB dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>
    specific_cloud_liquid_water_content  (time, prediction_timedelta, level, longitude, latitude) float32 3TB dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>
    specific_humidity                    (time, prediction_timedelta, level, longitude, latitude) float32 3TB dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>
    temperature                          (time, prediction_timedelta, level, longitude, latitude) float32 3TB dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>
    u_component_of_wind                  (time, prediction_timedelta, level, longitude, latitude) float32 3TB dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>
    v_component_of_wind                  (time, prediction_timedelta, level, longitude, latitude) float32 3TB dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>
Attributes:
    experiment_id:  70001146
    worker_id:      1

xarray.Dataset

• Dimensions:
  • time: 37
  • prediction_timedelta: 58449
  • level: 37
  • longitude: 128
  • latitude: 64
• Coordinates: (5)
  • latitude
    (latitude)
    float64
    -87.86 -85.1 -82.31 ... 85.1 87.86

    array([-87.863799, -85.096527, -82.312913, -79.525607, -76.7369  , -73.947515,
           -71.157752, -68.367756, -65.577607, -62.787352, -59.99702 , -57.206632,
           -54.4162  , -51.625734, -48.835241, -46.044727, -43.254195, -40.463648,
           -37.67309 , -34.882521, -32.091944, -29.30136 , -26.510769, -23.720174,
           -20.929574, -18.138971, -15.348365, -12.557756,  -9.767146,  -6.976534,
            -4.185921,  -1.395307,   1.395307,   4.185921,   6.976534,   9.767146,
            12.557756,  15.348365,  18.138971,  20.929574,  23.720174,  26.510769,
            29.30136 ,  32.091944,  34.882521,  37.67309 ,  40.463648,  43.254195,
            46.044727,  48.835241,  51.625734,  54.4162  ,  57.206632,  59.99702 ,
            62.787352,  65.577607,  68.367756,  71.157752,  73.947515,  76.7369  ,
            79.525607,  82.312913,  85.096527,  87.863799])

  • level
    (level)
    int64
    1 2 3 5 7 ... 900 925 950 975 1000

    array([   1,    2,    3,    5,    7,   10,   20,   30,   50,   70,  100,  125,
            150,  175,  200,  225,  250,  300,  350,  400,  450,  500,  550,  600,
            650,  700,  750,  775,  800,  825,  850,  875,  900,  925,  950,  975,
           1000])

  • longitude
    (longitude)
    float64
    0.0 2.812 5.625 ... 354.4 357.2

    array([  0.    ,   2.8125,   5.625 ,   8.4375,  11.25  ,  14.0625,  16.875 ,
            19.6875,  22.5   ,  25.3125,  28.125 ,  30.9375,  33.75  ,  36.5625,
            39.375 ,  42.1875,  45.    ,  47.8125,  50.625 ,  53.4375,  56.25  ,
            59.0625,  61.875 ,  64.6875,  67.5   ,  70.3125,  73.125 ,  75.9375,
            78.75  ,  81.5625,  84.375 ,  87.1875,  90.    ,  92.8125,  95.625 ,
            98.4375, 101.25  , 104.0625, 106.875 , 109.6875, 112.5   , 115.3125,
           118.125 , 120.9375, 123.75  , 126.5625, 129.375 , 132.1875, 135.    ,
           137.8125, 140.625 , 143.4375, 146.25  , 149.0625, 151.875 , 154.6875,
           157.5   , 160.3125, 163.125 , 165.9375, 168.75  , 171.5625, 174.375 ,
           177.1875, 180.    , 182.8125, 185.625 , 188.4375, 191.25  , 194.0625,
           196.875 , 199.6875, 202.5   , 205.3125, 208.125 , 210.9375, 213.75  ,
           216.5625, 219.375 , 222.1875, 225.    , 227.8125, 230.625 , 233.4375,
           236.25  , 239.0625, 241.875 , 244.6875, 247.5   , 250.3125, 253.125 ,
           255.9375, 258.75  , 261.5625, 264.375 , 267.1875, 270.    , 272.8125,
           275.625 , 278.4375, 281.25  , 284.0625, 286.875 , 289.6875, 292.5   ,
           295.3125, 298.125 , 300.9375, 303.75  , 306.5625, 309.375 , 312.1875,
           315.    , 317.8125, 320.625 , 323.4375, 326.25  , 329.0625, 331.875 ,
           334.6875, 337.5   , 340.3125, 343.125 , 345.9375, 348.75  , 351.5625,
           354.375 , 357.1875])

  • prediction_timedelta
    (prediction_timedelta)
    timedelta64[ns]
    0 days 00:00:00 ... 14612 days 0...

    array([                  0,      21600000000000,      43200000000000, ...,
           1262433600000000000, 1262455200000000000, 1262476800000000000],
          dtype='timedelta64[ns]')

  • time
    (time)
    datetime64[ns]
    1980-01-01 ... 1980-12-26

    array(['1980-01-01T00:00:00.000000000', '1980-01-11T00:00:00.000000000',
           '1980-01-21T00:00:00.000000000', '1980-01-31T00:00:00.000000000',
           '1980-02-10T00:00:00.000000000', '1980-02-20T00:00:00.000000000',
           '1980-03-01T00:00:00.000000000', '1980-03-11T00:00:00.000000000',
           '1980-03-21T00:00:00.000000000', '1980-03-31T00:00:00.000000000',
           '1980-04-10T00:00:00.000000000', '1980-04-20T00:00:00.000000000',
           '1980-04-30T00:00:00.000000000', '1980-05-10T00:00:00.000000000',
           '1980-05-20T00:00:00.000000000', '1980-05-30T00:00:00.000000000',
           '1980-06-09T00:00:00.000000000', '1980-06-19T00:00:00.000000000',
           '1980-06-29T00:00:00.000000000', '1980-07-09T00:00:00.000000000',
           '1980-07-19T00:00:00.000000000', '1980-07-29T00:00:00.000000000',
           '1980-08-08T00:00:00.000000000', '1980-08-18T00:00:00.000000000',
           '1980-08-28T00:00:00.000000000', '1980-09-07T00:00:00.000000000',
           '1980-09-17T00:00:00.000000000', '1980-09-27T00:00:00.000000000',
           '1980-10-07T00:00:00.000000000', '1980-10-17T00:00:00.000000000',
           '1980-10-27T00:00:00.000000000', '1980-11-06T00:00:00.000000000',
           '1980-11-16T00:00:00.000000000', '1980-11-26T00:00:00.000000000',
           '1980-12-06T00:00:00.000000000', '1980-12-16T00:00:00.000000000',
           '1980-12-26T00:00:00.000000000'], dtype='datetime64[ns]')

• Data variables: (7)
  • geopotential
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 2.38 TiB 1.16 MiB Shape (37, 58449, 37, 128, 64) (1, 1, 37, 128, 64) Dask graph 2162613 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_cloud_ice_water_content
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 2.38 TiB 1.16 MiB Shape (37, 58449, 37, 128, 64) (1, 1, 37, 128, 64) Dask graph 2162613 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_cloud_liquid_water_content
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 2.38 TiB 1.16 MiB Shape (37, 58449, 37, 128, 64) (1, 1, 37, 128, 64) Dask graph 2162613 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_humidity
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 2.38 TiB 1.16 MiB Shape (37, 58449, 37, 128, 64) (1, 1, 37, 128, 64) Dask graph 2162613 chunks in 2 graph layers Data type float32 numpy.ndarray

  • temperature
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 2.38 TiB 1.16 MiB Shape (37, 58449, 37, 128, 64) (1, 1, 37, 128, 64) Dask graph 2162613 chunks in 2 graph layers Data type float32 numpy.ndarray

  • u_component_of_wind
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 2.38 TiB 1.16 MiB Shape (37, 58449, 37, 128, 64) (1, 1, 37, 128, 64) Dask graph 2162613 chunks in 2 graph layers Data type float32 numpy.ndarray

  • v_component_of_wind
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 1, 37, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 2.38 TiB 1.16 MiB Shape (37, 58449, 37, 128, 64) (1, 1, 37, 128, 64) Dask graph 2162613 chunks in 2 graph layers Data type float32 numpy.ndarray

• Indexes: (5)
  • latitude
    PandasIndex

    PandasIndex(Index([ -87.86379883923264,   -85.0965269883173,  -82.31291294788628,
            -79.52560657265944,  -76.73689968036832,  -73.94751515398967,
            -71.15775201158732,  -68.36775610831317,  -65.57760701082782,
           -62.787351798963066,  -59.99702010849129, -57.206631527643246,
            -54.41619952608621, -51.625733674938246,  -48.83524096625059,
           -46.044726631101675,  -43.25419466535094,  -40.46364817811504,
            -37.67308962904533, -34.882520993773454,  -32.09194388174401,
           -29.301359621762735, -26.510769325210994, -23.720173933534753,
           -20.929574254489513, -18.138970990239358, -15.348364759491494,
           -12.557756115230678,  -9.767145559195567,  -6.976533553948635,
           -4.1859205331891545,  -1.395306910819497,   1.395306910819497,
            4.1859205331891545,   6.976533553948635,   9.767145559195567,
            12.557756115230678,  15.348364759491494,  18.138970990239358,
            20.929574254489513,  23.720173933534753,  26.510769325210994,
            29.301359621762735,   32.09194388174401,  34.882520993773454,
             37.67308962904533,   40.46364817811504,   43.25419466535094,
            46.044726631101675,   48.83524096625059,  51.625733674938246,
             54.41619952608621,  57.206631527643246,   59.99702010849129,
            62.787351798963066,   65.57760701082782,   68.36775610831317,
             71.15775201158732,   73.94751515398967,   76.73689968036832,
             79.52560657265944,   82.31291294788628,    85.0965269883173,
             87.86379883923264],
          dtype='float64', name='latitude'))

  • level
    PandasIndex

    PandasIndex(Index([   1,    2,    3,    5,    7,   10,   20,   30,   50,   70,  100,  125,
            150,  175,  200,  225,  250,  300,  350,  400,  450,  500,  550,  600,
            650,  700,  750,  775,  800,  825,  850,  875,  900,  925,  950,  975,
           1000],
          dtype='int64', name='level'))

  • longitude
    PandasIndex

    PandasIndex(Index([               0.0,             2.8125,              5.625,
                       8.4375,              11.25,            14.0625,
                       16.875,            19.6875,               22.5,
           25.312499999999996,
           ...
                      331.875,           334.6875, 337.49999999999994,
                     340.3125, 343.12499999999994,           345.9375,
                       348.75, 351.56249999999994,            354.375,
           357.18750000000006],
          dtype='float64', name='longitude', length=128))

  • prediction_timedelta
    PandasIndex

    PandasIndex(TimedeltaIndex([    '0 days 00:00:00',     '0 days 06:00:00',
                        '0 days 12:00:00',     '0 days 18:00:00',
                        '1 days 00:00:00',     '1 days 06:00:00',
                        '1 days 12:00:00',     '1 days 18:00:00',
                        '2 days 00:00:00',     '2 days 06:00:00',
                    ...
                    '14609 days 18:00:00', '14610 days 00:00:00',
                    '14610 days 06:00:00', '14610 days 12:00:00',
                    '14610 days 18:00:00', '14611 days 00:00:00',
                    '14611 days 06:00:00', '14611 days 12:00:00',
                    '14611 days 18:00:00', '14612 days 00:00:00'],
                   dtype='timedelta64[ns]', name='prediction_timedelta', length=58449, freq=None))

  • time
    PandasIndex

    PandasIndex(DatetimeIndex(['1980-01-01', '1980-01-11', '1980-01-21', '1980-01-31',
                   '1980-02-10', '1980-02-20', '1980-03-01', '1980-03-11',
                   '1980-03-21', '1980-03-31', '1980-04-10', '1980-04-20',
                   '1980-04-30', '1980-05-10', '1980-05-20', '1980-05-30',
                   '1980-06-09', '1980-06-19', '1980-06-29', '1980-07-09',
                   '1980-07-19', '1980-07-29', '1980-08-08', '1980-08-18',
                   '1980-08-28', '1980-09-07', '1980-09-17', '1980-09-27',
                   '1980-10-07', '1980-10-17', '1980-10-27', '1980-11-06',
                   '1980-11-16', '1980-11-26', '1980-12-06', '1980-12-16',
                   '1980-12-26'],
                  dtype='datetime64[ns]', name='time', freq=None))

• Attributes: (2)

  experiment_id :

  70001146

  worker_id :

  1

20-year runs 2.8 degree stochastic NeuralGCM model with precipitation

Corresponding model checkpoint: gs://neuralgcm/models/v1_precip/stochastic_precip_2_8_deg.pkl

path = 'gs://neuralgcm/amip_runs/v1_precip_stochastic_2_8_deg/2001-to-2021_128x64_gauss_37-level_stride3h.zarr'
ngcm_2_8_precip = xarray.open_zarr(path, storage_options=dict(token='anon'))
ngcm_2_8_precip

<xarray.Dataset> Size: 2TB
Dimensions:                              (time: 37,
                                          prediction_timedelta: 58408,
                                          surface: 1, longitude: 128,
                                          latitude: 64, level: 3)
Coordinates:
  * latitude                             (latitude) float64 512B -87.86 ... 8...
  * level                                (level) int64 24B 500 700 850
  * longitude                            (longitude) float64 1kB 0.0 ... 357.2
  * prediction_timedelta                 (prediction_timedelta) timedelta64[ns] 467kB ...
  * surface                              (surface) int64 8B 1
  * time                                 (time) datetime64[ns] 296B 2001-01-0...
Data variables:
    evaporation                          (time, prediction_timedelta, surface, longitude, latitude) float32 71GB dask.array<chunksize=(1, 8, 1, 128, 64), meta=np.ndarray>
    geopotential                         (time, prediction_timedelta, level, longitude, latitude) float32 212GB dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>
    precipitation_cumulative_mean        (time, prediction_timedelta, surface, longitude, latitude) float32 71GB dask.array<chunksize=(1, 8, 1, 128, 64), meta=np.ndarray>
    specific_cloud_ice_water_content     (time, prediction_timedelta, level, longitude, latitude) float32 212GB dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>
    specific_cloud_liquid_water_content  (time, prediction_timedelta, level, longitude, latitude) float32 212GB dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>
    specific_humidity                    (time, prediction_timedelta, level, longitude, latitude) float32 212GB dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>
    temperature                          (time, prediction_timedelta, level, longitude, latitude) float32 212GB dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>
    u_component_of_wind                  (time, prediction_timedelta, level, longitude, latitude) float32 212GB dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>
    v_component_of_wind                  (time, prediction_timedelta, level, longitude, latitude) float32 212GB dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>
Attributes:
    experiment_id:  121052169
    worker_id:      2

xarray.Dataset

• Dimensions:
  • time: 37
  • prediction_timedelta: 58408
  • surface: 1
  • longitude: 128
  • latitude: 64
  • level: 3
• Coordinates: (6)
  • latitude
    (latitude)
    float64
    -87.86 -85.1 -82.31 ... 85.1 87.86

    array([-87.863799, -85.096527, -82.312913, -79.525607, -76.7369  , -73.947515,
           -71.157752, -68.367756, -65.577607, -62.787352, -59.99702 , -57.206632,
           -54.4162  , -51.625734, -48.835241, -46.044727, -43.254195, -40.463648,
           -37.67309 , -34.882521, -32.091944, -29.30136 , -26.510769, -23.720174,
           -20.929574, -18.138971, -15.348365, -12.557756,  -9.767146,  -6.976534,
            -4.185921,  -1.395307,   1.395307,   4.185921,   6.976534,   9.767146,
            12.557756,  15.348365,  18.138971,  20.929574,  23.720174,  26.510769,
            29.30136 ,  32.091944,  34.882521,  37.67309 ,  40.463648,  43.254195,
            46.044727,  48.835241,  51.625734,  54.4162  ,  57.206632,  59.99702 ,
            62.787352,  65.577607,  68.367756,  71.157752,  73.947515,  76.7369  ,
            79.525607,  82.312913,  85.096527,  87.863799])

  • level
    (level)
    int64
    500 700 850

    array([500, 700, 850])

  • longitude
    (longitude)
    float64
    0.0 2.812 5.625 ... 354.4 357.2

    array([  0.    ,   2.8125,   5.625 ,   8.4375,  11.25  ,  14.0625,  16.875 ,
            19.6875,  22.5   ,  25.3125,  28.125 ,  30.9375,  33.75  ,  36.5625,
            39.375 ,  42.1875,  45.    ,  47.8125,  50.625 ,  53.4375,  56.25  ,
            59.0625,  61.875 ,  64.6875,  67.5   ,  70.3125,  73.125 ,  75.9375,
            78.75  ,  81.5625,  84.375 ,  87.1875,  90.    ,  92.8125,  95.625 ,
            98.4375, 101.25  , 104.0625, 106.875 , 109.6875, 112.5   , 115.3125,
           118.125 , 120.9375, 123.75  , 126.5625, 129.375 , 132.1875, 135.    ,
           137.8125, 140.625 , 143.4375, 146.25  , 149.0625, 151.875 , 154.6875,
           157.5   , 160.3125, 163.125 , 165.9375, 168.75  , 171.5625, 174.375 ,
           177.1875, 180.    , 182.8125, 185.625 , 188.4375, 191.25  , 194.0625,
           196.875 , 199.6875, 202.5   , 205.3125, 208.125 , 210.9375, 213.75  ,
           216.5625, 219.375 , 222.1875, 225.    , 227.8125, 230.625 , 233.4375,
           236.25  , 239.0625, 241.875 , 244.6875, 247.5   , 250.3125, 253.125 ,
           255.9375, 258.75  , 261.5625, 264.375 , 267.1875, 270.    , 272.8125,
           275.625 , 278.4375, 281.25  , 284.0625, 286.875 , 289.6875, 292.5   ,
           295.3125, 298.125 , 300.9375, 303.75  , 306.5625, 309.375 , 312.1875,
           315.    , 317.8125, 320.625 , 323.4375, 326.25  , 329.0625, 331.875 ,
           334.6875, 337.5   , 340.3125, 343.125 , 345.9375, 348.75  , 351.5625,
           354.375 , 357.1875])

  • prediction_timedelta
    (prediction_timedelta)
    timedelta64[ns]
    0 days 00:00:00 ... 7300 days 21...

    array([                 0,     10800000000000,     21600000000000, ...,
           630774000000000000, 630784800000000000, 630795600000000000],
          dtype='timedelta64[ns]')

  • surface
    (surface)
    int64
    1

    array([1])

  • time
    (time)
    datetime64[ns]
    2001-01-01 ... 2001-12-27

    array(['2001-01-01T00:00:00.000000000', '2001-01-11T00:00:00.000000000',
           '2001-01-21T00:00:00.000000000', '2001-01-31T00:00:00.000000000',
           '2001-02-10T00:00:00.000000000', '2001-02-20T00:00:00.000000000',
           '2001-03-02T00:00:00.000000000', '2001-03-12T00:00:00.000000000',
           '2001-03-22T00:00:00.000000000', '2001-04-01T00:00:00.000000000',
           '2001-04-11T00:00:00.000000000', '2001-04-21T00:00:00.000000000',
           '2001-05-01T00:00:00.000000000', '2001-05-11T00:00:00.000000000',
           '2001-05-21T00:00:00.000000000', '2001-05-31T00:00:00.000000000',
           '2001-06-10T00:00:00.000000000', '2001-06-20T00:00:00.000000000',
           '2001-06-30T00:00:00.000000000', '2001-07-10T00:00:00.000000000',
           '2001-07-20T00:00:00.000000000', '2001-07-30T00:00:00.000000000',
           '2001-08-09T00:00:00.000000000', '2001-08-19T00:00:00.000000000',
           '2001-08-29T00:00:00.000000000', '2001-09-08T00:00:00.000000000',
           '2001-09-18T00:00:00.000000000', '2001-09-28T00:00:00.000000000',
           '2001-10-08T00:00:00.000000000', '2001-10-18T00:00:00.000000000',
           '2001-10-28T00:00:00.000000000', '2001-11-07T00:00:00.000000000',
           '2001-11-17T00:00:00.000000000', '2001-11-27T00:00:00.000000000',
           '2001-12-07T00:00:00.000000000', '2001-12-17T00:00:00.000000000',
           '2001-12-27T00:00:00.000000000'], dtype='datetime64[ns]')

• Data variables: (9)
  • evaporation
    (time, prediction_timedelta, surface, longitude, latitude)
    float32
    dask.array<chunksize=(1, 8, 1, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 65.95 GiB 256.00 kiB Shape (37, 58408, 1, 128, 64) (1, 8, 1, 128, 64) Dask graph 270137 chunks in 2 graph layers Data type float32 numpy.ndarray

  • geopotential
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 197.85 GiB 768.00 kiB Shape (37, 58408, 3, 128, 64) (1, 8, 3, 128, 64) Dask graph 270137 chunks in 2 graph layers Data type float32 numpy.ndarray

  • precipitation_cumulative_mean
    (time, prediction_timedelta, surface, longitude, latitude)
    float32
    dask.array<chunksize=(1, 8, 1, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 65.95 GiB 256.00 kiB Shape (37, 58408, 1, 128, 64) (1, 8, 1, 128, 64) Dask graph 270137 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_cloud_ice_water_content
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 197.85 GiB 768.00 kiB Shape (37, 58408, 3, 128, 64) (1, 8, 3, 128, 64) Dask graph 270137 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_cloud_liquid_water_content
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 197.85 GiB 768.00 kiB Shape (37, 58408, 3, 128, 64) (1, 8, 3, 128, 64) Dask graph 270137 chunks in 2 graph layers Data type float32 numpy.ndarray

  • specific_humidity
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 197.85 GiB 768.00 kiB Shape (37, 58408, 3, 128, 64) (1, 8, 3, 128, 64) Dask graph 270137 chunks in 2 graph layers Data type float32 numpy.ndarray

  • temperature
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 197.85 GiB 768.00 kiB Shape (37, 58408, 3, 128, 64) (1, 8, 3, 128, 64) Dask graph 270137 chunks in 2 graph layers Data type float32 numpy.ndarray

  • u_component_of_wind
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 197.85 GiB 768.00 kiB Shape (37, 58408, 3, 128, 64) (1, 8, 3, 128, 64) Dask graph 270137 chunks in 2 graph layers Data type float32 numpy.ndarray

  • v_component_of_wind
    (time, prediction_timedelta, level, longitude, latitude)
    float32
    dask.array<chunksize=(1, 8, 3, 128, 64), meta=np.ndarray>

    Array Chunk Bytes 197.85 GiB 768.00 kiB Shape (37, 58408, 3, 128, 64) (1, 8, 3, 128, 64) Dask graph 270137 chunks in 2 graph layers Data type float32 numpy.ndarray

• Indexes: (6)
  • latitude
    PandasIndex

    PandasIndex(Index([ -87.86379883923264,   -85.0965269883173,  -82.31291294788628,
            -79.52560657265944,  -76.73689968036832,  -73.94751515398967,
            -71.15775201158732,  -68.36775610831317,  -65.57760701082782,
           -62.787351798963066,  -59.99702010849129, -57.206631527643246,
            -54.41619952608621, -51.625733674938246,  -48.83524096625059,
           -46.044726631101675,  -43.25419466535094,  -40.46364817811504,
            -37.67308962904533, -34.882520993773454,  -32.09194388174401,
           -29.301359621762735, -26.510769325210994, -23.720173933534753,
           -20.929574254489513, -18.138970990239358, -15.348364759491494,
           -12.557756115230678,  -9.767145559195567,  -6.976533553948635,
           -4.1859205331891545,  -1.395306910819497,   1.395306910819497,
            4.1859205331891545,   6.976533553948635,   9.767145559195567,
            12.557756115230678,  15.348364759491494,  18.138970990239358,
            20.929574254489513,  23.720173933534753,  26.510769325210994,
            29.301359621762735,   32.09194388174401,  34.882520993773454,
             37.67308962904533,   40.46364817811504,   43.25419466535094,
            46.044726631101675,   48.83524096625059,  51.625733674938246,
             54.41619952608621,  57.206631527643246,   59.99702010849129,
            62.787351798963066,   65.57760701082782,   68.36775610831317,
             71.15775201158732,   73.94751515398967,   76.73689968036832,
             79.52560657265944,   82.31291294788628,    85.0965269883173,
             87.86379883923264],
          dtype='float64', name='latitude'))

  • level
    PandasIndex

    PandasIndex(Index([500, 700, 850], dtype='int64', name='level'))

  • longitude
    PandasIndex

    PandasIndex(Index([               0.0,             2.8125,              5.625,
                       8.4375,              11.25,            14.0625,
                       16.875,            19.6875,               22.5,
           25.312499999999996,
           ...
                      331.875,           334.6875, 337.49999999999994,
                     340.3125, 343.12499999999994,           345.9375,
                       348.75, 351.56249999999994,            354.375,
           357.18750000000006],
          dtype='float64', name='longitude', length=128))

  • prediction_timedelta
    PandasIndex

    PandasIndex(TimedeltaIndex([   '0 days 00:00:00',    '0 days 03:00:00',
                       '0 days 06:00:00',    '0 days 09:00:00',
                       '0 days 12:00:00',    '0 days 15:00:00',
                       '0 days 18:00:00',    '0 days 21:00:00',
                       '1 days 00:00:00',    '1 days 03:00:00',
                    ...
                    '7299 days 18:00:00', '7299 days 21:00:00',
                    '7300 days 00:00:00', '7300 days 03:00:00',
                    '7300 days 06:00:00', '7300 days 09:00:00',
                    '7300 days 12:00:00', '7300 days 15:00:00',
                    '7300 days 18:00:00', '7300 days 21:00:00'],
                   dtype='timedelta64[ns]', name='prediction_timedelta', length=58408, freq=None))

  • surface
    PandasIndex

    PandasIndex(Index([1], dtype='int64', name='surface'))

  • time
    PandasIndex

    PandasIndex(DatetimeIndex(['2001-01-01', '2001-01-11', '2001-01-21', '2001-01-31',
                   '2001-02-10', '2001-02-20', '2001-03-02', '2001-03-12',
                   '2001-03-22', '2001-04-01', '2001-04-11', '2001-04-21',
                   '2001-05-01', '2001-05-11', '2001-05-21', '2001-05-31',
                   '2001-06-10', '2001-06-20', '2001-06-30', '2001-07-10',
                   '2001-07-20', '2001-07-30', '2001-08-09', '2001-08-19',
                   '2001-08-29', '2001-09-08', '2001-09-18', '2001-09-28',
                   '2001-10-08', '2001-10-18', '2001-10-28', '2001-11-07',
                   '2001-11-17', '2001-11-27', '2001-12-07', '2001-12-17',
                   '2001-12-27'],
                  dtype='datetime64[ns]', name='time', freq=None))

• Attributes: (2)

  experiment_id :

  121052169

  worker_id :

  2

In this dataset, precipitation_cumulative_mean indicates cumulative precipitation (in meters) from the start of the simulation (i.e., from prediction_timedelta=0).

This makes it straightforward to calculate mean precipitation without needing to load the entire 71 GB dataset from storage:

# Calculate mean precipitation
m_to_mm = 1000
measurement_freq = 3  # in hours
measurements_in_a_day = 24 / measurement_freq
daily_precip = ngcm_2_8_precip.precipitation_cumulative_mean.isel(
    time=-1, surface=0, prediction_timedelta=-1
) * (
    m_to_mm
    * measurements_in_a_day
    / (len(ngcm_2_8_precip.prediction_timedelta) - 1)
)
daily_precip.plot(x="longitude", cbar_kwargs={"label": "(mm/day)"})

<matplotlib.collections.QuadMesh at 0x79abc2dd4880>

To calculate precipitation over shorter internals, use .diff('prediction_timedelta'):

ngcm_2_8_precip.precipitation_cumulative_mean.isel(
    time=-1, surface=0, prediction_timedelta=slice(10, 15)
).diff('prediction_timedelta').plot(
    x='longitude',
    col='prediction_timedelta',
    cbar_kwargs={'label': '(meter/3-hour)'},
)  # In m per 3 hours

<xarray.plot.facetgrid.FacetGrid at 0x79abcb421cf0>