Dataset and file structure

CPCIR and GridSat

CCIC is derived from two satellite-observations datasets: The GridSat-B1 dataset [KNOAACProgram14] and the NCEP/CPC Merged IR dataset [JJX17]. In the following, these two datasets will be referred to as GridSat and CPCIR, respectively.

CCIC provides estimates for both the GridSat and the CPCIR datasets. The estimates are provided on the same grid as the observations and thus inherit their temporospatial resolution and coverage. With temporospatial resolution of 3h @ 0.07 degree, the GridSat-based data offers lower resolution than the CPCIR data, which has 30 min @ 0.036 degree resolution. However, GridSat is available from 1980, whereas CPCIR only from 2000 (with some retrievals in 1998). The temporospatial coverage and resolution is summarized in table table 1.

Table 1 Temporospatial coverage and resolution of the GridSat and CPCIR variants of CCIC

Dataset	Coverage	Spatial resolution	Temporal resolution
GridSat	1980 - present	0.07 degree	3 h
CPCIR	2000 - present	0.04 degree	30 min

Data organization

The CCIC record is organized into results derived from GridSat input data results derived from CPCIR input data. Below that files are organized into folder by year.

record
├── cpcir
│   ├── 2000
│   ├── 2001
┆   ┆
│   ├── 2022
│   └── 2023
└── gridsat
    ├── 1980
    ├── 1981
    ┆
    ├── 2022
    └── 2023

The files follow the naming pattern

ccic_{product}_{YYYYmmddHH}00.zarr

where {product} is either cpcir or gridsat, and YYYYmmddHH is the timestamp. Each GridSat file contains the retrieval for one timestamp, and each CPCIR file contains the retrieval for two timestamps, the full hour and 30 minutes after the full hour. For each day, GridSat are available at hours HH = 00, 03, 06, 09, 12, 15, 18, and 21, while CPCIR files are available at hours HH = 00, 01, …, 22, 23.

To list the files currently available in the CCIC S3 bucket, e.g., for 2020 and CPCIR, you can use:

• Python:

  import s3fs
  all_files = s3.ls("chalmerscloudiceclimatology/record/cpcir/2020")
  first_day = s3.glob("chalmerscloudiceclimatology/record/cpcir/2020/ccic_cpcir_20200101*zarr")
  

• AWS Command Line Interface (terminal):

  $ aws --no-sign-request s3 ls s3://chalmerscloudiceclimatology/record/cpcir/2020/
  

Variables

The CCIC climate data record provides estimates of the total ice water path (TIWP) and a 2D cloud probability. The data files follow CF conventions. The variables and their meaning are listed in table 2.

Table 2 CCIC Variables and their significance

Variable name	Units	Range	Description
tiwp	kg m-2	≥ 0	Vertically-integrated concentration of frozen hydrometeors
tiwp_ci	kg m-2	≥ 0	90% confidence interval for the retrieved TIWP
p_tiwp		[0, 1]	Probability that tiwp exceeds 10-3 kg m-2
cloud_prob_2d		[0, 1]	Probability of presence of a cloud anywhere in the atmosphere
inpainted		{0, 1}	Input pixel was NaN; the retrieval can be a numeric value (inpainted)

Note: cloud_prob_2d will be published on a second upload phase.

These variables are gridded on the coordinates from the table below, where the spatial grid and the time resolution are constant for each input product.

Table 3 Coordinate conventions used by CCIC

Coordinate name	Units	Range	Description
latitude	Degrees North	CPCIR: (-60, +60) GridSat: [-70, +70)	Latitude
longitude	Degrees East	[-180, +180)	Longitude
time	ns	numpy’s datetime64[ns] (nanoseconds since Unix epoch)|	Nominal retrieval time