Add wind example and expand test coverage

Author: joshday

examples/wind.jl

@@ -0,0 +1,5 @@
+import RapidRefreshData as RRD
+
+using Rasters, ArchGDAL, Dates
+
+datasets = [RRD.RAPDataset(date = Date(2024, 6, 15), cycle_time = "t0$(i)z") for i in 0:5]

src/RapidRefreshData.jl

@@ -3,14 +3,19 @@ module RapidRefreshData
 using Dates, Scratch
 
 const rap_base_url = "https://noaa-rap-pds.s3.amazonaws.com/rap."
-const gfs_base_url = "https://noaa-gfs-bdp-pds.s3.amazonaws.com/"
+const gfs_base_url = "https://noaa-gfs-bdp-pds.s3.amazonaws.com/gfs"
+const hrrr_base_url = "https://noaa-hrrr-bdp-pds.s3.amazonaws.com/hrrr"
 
 
 #-----------------------------------------------------------------------------# __init__
-dir::String = ""
+rap_dir::String = ""
+gfs_dir::String = ""
+hrrr_dir::String = ""
 
 function __init__()
-    global dir = Scratch.@get_scratch!("datasets")
+    global rap_dir = Scratch.@get_scratch!("rap")
+    global gfs_dir = Scratch.@get_scratch!("gfs")
+    global hrrr_dir = Scratch.@get_scratch!("hrrr")
 end
 
 #-----------------------------------------------------------------------------# RAPDataset
@@ -65,7 +70,7 @@ end
 Get local cache path for the RAP dataset.
 """
 function local_path(dset::RAPDataset)
-    joinpath(dir, "rap_$(Dates.format(dset.date, "yyyymmdd"))_$(dset.cycle_time)_$(dset.grid)_$(dset.forecast).grib2")
+    joinpath(rap_dir, "rap_$(Dates.format(dset.date, "yyyymmdd"))_$(dset.cycle_time)_$(dset.grid)_$(dset.forecast).grib2")
 end
 
 """
@@ -83,8 +88,7 @@ end
 List all cached RAP datasets.
 """
 function local_datasets(::Type{RAPDataset})
-    files = filter(f -> startswith(f, "rap_") && endswith(f, ".grib2"), readdir(dir))
-    read.(RAPDataset, files)
+    read.(RAPDataset, readdir(rap_dir))
 end
 
 """
@@ -140,7 +144,7 @@ end
 Generate AWS S3 URL for the GFS dataset.
 """
 function url(dset::GFSDataset)
-    "$(gfs_base_url)gfs.$(Dates.format(dset.date, "yyyymmdd"))/$(dset.cycle)/$(dset.product)/gfs.t$(dset.cycle)z.pgrb2.$(dset.resolution).$(dset.forecast)"
+    "$(gfs_base_url).$(Dates.format(dset.date, "yyyymmdd"))/$(dset.cycle)/$(dset.product)/gfs.t$(dset.cycle)z.pgrb2.$(dset.resolution).$(dset.forecast)"
 end
 
 """
@@ -149,7 +153,7 @@ end
 Get local cache path for the GFS dataset.
 """
 function local_path(dset::GFSDataset)
-    joinpath(dir, "gfs_$(Dates.format(dset.date, "yyyymmdd"))_$(dset.cycle)_$(dset.resolution)_$(dset.product)_$(dset.forecast).grib2")
+    joinpath(gfs_dir, "gfs_$(Dates.format(dset.date, "yyyymmdd"))_$(dset.cycle)_$(dset.resolution)_$(dset.product)_$(dset.forecast).grib2")
 end
 
 """
@@ -167,8 +171,7 @@ end
 List all cached GFS datasets.
 """
 function local_datasets(::Type{GFSDataset})
-    files = filter(f -> startswith(f, "gfs_"), readdir(dir))
-    read.(GFSDataset, files)
+    read.(GFSDataset, readdir(gfs_dir))
 end
 
 """
@@ -180,5 +183,88 @@ function clear_local_dataset!(dset::GFSDataset)
     isfile(local_path(dset)) && rm(local_path(dset))
 end
 
+#-----------------------------------------------------------------------------# HRRRDataset
+"""
+    HRRRDataset(; date, cycle, region, product, forecast)
+
+NOAA High-Resolution Rapid Refresh (HRRR) dataset descriptor.
+
+# Fields
+- `date::Date`: Forecast initialization date (default: today)
+- `cycle::String`: Model run hour - "00" to "23" (default: "00")
+- `region::String`: Geographic region - "conus" (Continental US) or "alaska" (default: "conus")
+- `product::String`: Product type - "wrfsfc" (surface), "wrfprs" (pressure), "wrfnat" (native), or "wrfsub" (subhourly) (default: "wrfsfc")
+- `forecast::String`: Forecast hour - "f00" to "f48" (default: "f00")
+"""
+@kwdef struct HRRRDataset
+    date::Date = today()
+    cycle::String = "00"  # "00" to "23"
+    region::String = "conus"  # "conus", "alaska"
+    product::String = "wrfsfc"  # "wrfsfc", "wrfprs", "wrfnat", "wrfsub"
+    forecast::String = "f00"  # "f00" to "f48"
+end
+
+function Base.show(io::IO, dset::HRRRDataset)
+    (;date, cycle, region, product, forecast) = dset
+    print(io, "HRRRDataset", (; date=string(date), cycle, region, product, forecast))
+end
+
+function Base.read(::Type{HRRRDataset}, local_path::String)
+    filename = split(basename(local_path), ".")[1]
+    parts = split(filename, "_")
+    # Skip "hrrr" prefix (parts[1]) and parse remaining parts
+    date = Date(parts[2], "yyyymmdd")
+    cycle = parts[3]
+    region = parts[4]
+    product = parts[5]
+    forecast = parts[6]
+    return HRRRDataset(; date, cycle, region, product, forecast)
+end
+
+"""
+    url(dset::HRRRDataset) -> String
+
+Generate AWS S3 URL for the HRRR dataset.
+"""
+function url(dset::HRRRDataset)
+    "$(hrrr_base_url).$(Dates.format(dset.date, "yyyymmdd"))/$(dset.region)/hrrr.t$(dset.cycle)z.$(dset.product)$(dset.forecast).grib2"
+end
+
+"""
+    local_path(dset::HRRRDataset) -> String
+
+Get local cache path for the HRRR dataset.
+"""
+function local_path(dset::HRRRDataset)
+    joinpath(hrrr_dir, "hrrr_$(Dates.format(dset.date, "yyyymmdd"))_$(dset.cycle)_$(dset.region)_$(dset.product)_$(dset.forecast).grib2")
+end
+
+"""
+    get(dset::HRRRDataset) -> String
+
+Download HRRR dataset to cache (if needed) and return filepath.
+"""
+function Base.get(dset::HRRRDataset)
+    isfile(local_path(dset)) ? local_path(dset) : Base.download(url(dset), local_path(dset))
+end
+
+"""
+    local_datasets(HRRRDataset) -> Vector{HRRRDataset}
+
+List all cached HRRR datasets.
+"""
+function local_datasets(::Type{HRRRDataset})
+    read.(HRRRDataset, readdir(hrrr_dir))
+end
+
+"""
+    clear_local_dataset!(dset::HRRRDataset)
+
+Remove HRRR dataset from local cache.
+"""
+function clear_local_dataset!(dset::HRRRDataset)
+    isfile(local_path(dset)) && rm(local_path(dset))
+end
+
 
 end