Merge pull request #150 from rabernat/gpcp

Add recipe for gpcp

recipes/gpcp/meta.yaml

@@ -0,0 +1,31 @@
+title: "Global Precipitation Climatology Project"
+description: >
+  Global Precipitation Climatology Project (GPCP) Daily Version 1.3 gridded, merged ty
+  satellite/gauge precipitation Climate data Record (CDR) from 1996 to present.
+pangeo_forge_version: "0.8.3"
+pangeo_notebook_version: "2022.06.02"
+recipes:
+  - id: gpcp
+    object: "recipe:recipe"
+provenance:
+  providers:
+    - name: "NOAA NCEI"
+      description: "National Oceanographic & Atmospheric Administration National Centers for Environmental Information"
+      roles:
+        - host
+        - licensor
+      url: https://www.ncei.noaa.gov/products/global-precipitation-climatology-project
+    - name: "University of Maryland"
+      description: >
+        University of Maryland College Park Earth System Science Interdisciplinary Center
+        (ESSIC) and Cooperative Institute for Climate and Satellites (CICS)
+      roles:
+        - producer
+      url: http://gpcp.umd.edu/
+  license: "No constraints on data access or use."
+maintainers:
+  - name: "Ryan Abernathey"
+    orcid: "0000-0001-5999-4917"
+    github: rabernat
+bakery:
+  id: "pangeo-ldeo-nsf-earthcube"

recipes/gpcp/recipe.py

@@ -0,0 +1,31 @@
+import fsspec
+from fsspec.implementations.http import HTTPFileSystem
+from pangeo_forge_recipes.patterns import pattern_from_file_sequence, FileType
+from pangeo_forge_recipes.recipes import XarrayZarrRecipe
+
+# The GPCP files use an annoying naming convention which embeds the creation date in the file name.
+# e.g. https 1996/gpcp_v01r03_daily_d19961001_c20170530.nc
+# This makes it very hard to create a deterministic function to generate the file names,
+# so instead we crawl the NCEI server.
+
+url_base = "https://www.ncei.noaa.gov/data/global-precipitation-climatology-project-gpcp-daily/access/"
+years = range(1996, 2022)
+file_list = []
+fs = HTTPFileSystem()
+for year in years:
+    file_list += sorted(filter(
+        lambda x: x.endswith('.nc'),
+        fs.ls(url_base + str(year), detail=False)
+    ))
+
+pattern = pattern_from_file_sequence(
+    file_list,
+    "time",
+    nitems_per_file=1
+)
+
+recipe = XarrayZarrRecipe(
+    pattern,
+    inputs_per_chunk=200,
+    xarray_open_kwargs={"decode_coords": "all"}
+)