Draft DRCDP data specifications have been established for data technically aligned with other Earth System Grid Federation (ESGF) projects including CMIP, CORDEX, and obs4MIPs. Here, we provide guidance to help data providers generate compliant data using these specifications.

• Contributors must register their datasets by submitting an issue on this DRCDP GitHub repository and filling out the provided template.

• We recommend data providers prepare their data using Python so that they can more readily leverage the resources described below.

• We recommend using Xarray as in the examples provided below, and xCDAT to generate coordinate bounds.

• Use of the Climate Model Output Writer (CMOR) is required to ensure alignment with other ESGF projects.

The easiest way to configure a suitable computing environment is via the conda package manager, Xarray, xCDAT and CMOR can be obtained and included in a common environment.

The recipe below describes the process of preparing a DRCDP-compliant dataset. This typically involves copying information from the provided demos and modifying it as necessary to prepare a new DRCDP-compliant dataset. By preparing a simple python script (example discussed in demo identified below) and an input JavaScript Object Notation (JSON) file, CMOR is used to prepare a DRCDP-compliant dataset.

1. Register a new source_id, if it does not already exist. An issue can be submitted on this GitHub repo (here) with a proposed "source_id" and other key registered content (RC). A member of the team will quickly review this information and enter it into the database of source_id's or propose an alternative if the proposed identifier does not conform to the data specifications. When opening an issue, a template is provided, so one needs to replace the information in the example (GPCP) with their own proposed source_id. The source_id is intended to identify the product/version and generally closely resembles an existing identifier but may be slightly modified to be consistent with CMIP/obs4MIPs conventions. For example: A compliant source_id for "GPCP v2.4" is "GPCP-2-4". More information on the guidelines for constructing a source_id is available in Table 1 of the obs4MIPs data specifications (see here).

2. Prepare an input table for running CMOR (e.g., see here). The simplest thing to do is to save this file, rename it, and replace the demo content with the relevant information for a new source_id or dataset. Typically, this involves only making changes to the following attributes: "contact", "grid", "grid_label", "institution_id", "nominal_resolution", "references", "outpath", "source_id", "title", "variant_info" and "variant_label". For explanations of these attributes, see the obs4MIPs data specifications (see here).

3. Prepare a Python script for reading in data and writing with CMOR. This is the most time consuming aspect of preparing a DRCDP-compliant dataset, but examples provided on this repo (see here) help to streamline the process. It involves preparing a simple Python script to read the original data and rewrite it using CMOR. As with the steps before, one can start by downloading a demo python script (see examples), renaming, and modifying as needed.

4. Process and generation DRCDP compliant data by executing your Python script. The processed data will be located in a directory defined in the input_table: outpath + output_path_template, the former being the base directory (where the user wants to output the data). The latter being a directory template explicitly defined for DRCDP (<activity_id>/<institution_id>/<source_id>/<variable_id>/<grid_label>/).

5. Contact the DRCDP team with any questions with the above and to arrange verification and data transfer for ESGF publication.