Edit ADRIA.Domain() and generalize to incorporate loading of GBR-wide domain package

src/ADRIA.jl

@@ -102,7 +102,7 @@ export RMEResultSet
 # export dims, ndims
 
 # List out compatible domain datapackages
-const COMPAT_DPKG = ["0.7.0-rc", "0.7.0"]
+const COMPAT_DPKG = ["0.7.0-rc", "0.7.0", "0.1.0-gbr"]
 # This adds ~30 seconds to package load times
 if ccall(:jl_generating_output, Cint, ()) == 1
     Base.precompile(Tuple{typeof(load_domain),String})   # time: 19.120537

src/ExtInterface/ADRIA/Domain.jl

@@ -154,8 +154,10 @@ function Domain(
         timeframe
     )
 
-    # Sort data to maintain consistent order
-    sort!(location_data, Symbol[Symbol(location_id_col)])
+    if cluster_id_col == "cluster_id"
+        # Sort data to maintain consistent order (only if domain is cluster-scale).
+        sort!(location_data, Symbol[Symbol(location_id_col)])
+    end
     u_sids::Vector{String} = string.(collect(location_data[!, location_id_col]))
     # If location id column is missing then derive it from the Unique IDs
     if !in(location_id_col, names(location_data))
@@ -173,7 +175,25 @@ function Domain(
     location_data = location_data[
         coalesce.(in.(conn_ids, [connectivity.loc_ids]), false), (:)
     ]
-    location_data.k .= location_data.k / 100.0  # Make `k` non-dimensional (provided as a percent)
+    if ("k" ∉ names(location_data)) &
+        ("ReefMod_habitable_proportion" ∈ names(location_data))
+        @warn "
+        k column not found in gbr-wide canonical-reefs gpkg. 
+        # Defaulting to ReefMod_habitable_proportion (in proportion 0-1 scale).
+        "
+        rename!(location_data, :ReefMod_habitable_proportion => :k)
+    else
+        location_data.k .= location_data.k / 100.0  # Make `k` non-dimensional (provided as a percent in cluster-scale data).
+    end
+
+    if ("area" ∉ names(location_data)) &
+        ("ReefMod_area_m2" ∈ names(location_data))
+        @warn " 
+        # area column not found in gbr-wide canonical-reefs gpkg. 
+        # Defaulting to ReefMod_area_m2 (Total possible coral area in m2).
+        "
+        rename!(location_data, :ReefMod_area_m2 => :area)
+    end
 
     n_locs::Int64 = nrow(location_data)
     n_groups::Int64, n_sizes::Int64 = size(linear_extensions())
@@ -191,7 +211,8 @@ function Domain(
     waves = load_env_data(waves_params...)
 
     cyc_params =
-        ispath(cyclone_mortality_fn) ? (cyclone_mortality_fn,) : (timeframe, location_data)
+        ispath(cyclone_mortality_fn) ? (cyclone_mortality_fn,) :
+        (timeframe, location_data, location_id_col)
     cyclone_mortality = load_cyclone_mortality(cyc_params...)
 
     # Add compatability with non-migrated datasets but always default current coral spec
@@ -207,6 +228,14 @@ function Domain(
     msg::String = "Provided time frame must match timesteps in DHW and wave data"
     msg = msg * "\n Got: $(length(timeframe)) | $(size(dhw, 1)) | $(size(waves, 1))"
 
+    # ReefMod DHW timeseries data (2000:2100) do not match the length of waves/cyclone 
+    # data (2025:2099). Can be fixed when AIMS-sourced DHW trajectories become available. 
+    if size(dhw, 1) > size(waves, 1)
+        first_year = findfirst(collect(2000:2100) .== first(timeframe))
+        last_year = findfirst(collect(2000:2100) .== last(timeframe))
+        dhw = dhw[first_year:last_year, :, :]
+    end
+
     @assert length(timeframe) == size(dhw, 1) == size(waves, 1) msg
 
     return Domain(
@@ -245,6 +274,8 @@ function load_domain(::Type{ADRIADomain}, path::String, rcp::String)::ADRIADomai
     end
 
     dpkg_details::Dict{String,Any} = _load_dpkg(path)
+    location_id_col = _get_id_col(dpkg_details)
+    cluster_id_col::String = "cluster_id"
 
     # Handle compatibility
     # Extract the time frame represented in this data package
@@ -262,9 +293,6 @@ function load_domain(::Type{ADRIADomain}, path::String, rcp::String)::ADRIADomai
         timeframe = parse.(Int64, md_timeframe)
     end
 
-    location_id_col::String = "reef_siteid"
-    cluster_id_col::String = "cluster_id"
-
     conn_path::String = joinpath(path, "connectivity/")
     spatial_path::String = joinpath(path, "spatial")
 
@@ -290,6 +318,7 @@ function load_domain(::Type{ADRIADomain}, path::String, rcp::String)::ADRIADomai
         cyclone_mortality_fn
     )
 end
+
 function load_domain(path::String, rcp::String)::ADRIADomain
     return load_domain(ADRIADomain, path, rcp)
 end
@@ -307,6 +336,15 @@ function get_wave_data(d::ADRIADomain, RCP::String)::String
     return joinpath(d.env_layer_md.dpkg_path, "waves", "wave_RCP$(RCP).nc")
 end
 
+function _get_id_col(dpkg_details)
+    spatial_resources = findfirst(
+        [x["name"] for x in dpkg_details["resources"]] .== "spatial_data"
+    )
+    spatial_resources = dpkg_details["resources"][spatial_resources]
+
+    return first(spatial_resources["data"])
+end
+
 """
     switch_RCPs!(d::Domain, RCP::String)::Domain
 

src/io/inputs.jl

@@ -166,10 +166,12 @@ function load_cyclone_mortality(data_fn::String)::YAXArray
     cyclone_cube::YAXArray = Cube(data_fn)
     return sort_axis(cyclone_cube, :locations)
 end
-function load_cyclone_mortality(timeframe::Vector{Int64}, loc_data::DataFrame)::YAXArray
+function load_cyclone_mortality(
+    timeframe::Vector{Int64}, loc_data::DataFrame, location_id_col::String
+)::YAXArray
     return ZeroDataCube(;
         timesteps=1:length(timeframe),
-        locations=sort(loc_data.reef_siteid),
+        locations=loc_data[:, location_id_col],
         species=ADRIA.coral_spec().taxa_names,
         scenarios=[1]
     )