Merge pull request #771 from CliMA/gb/leaderboard

Change units to mm/day for leaderboard

experiments/ClimaEarth/user_io/leaderboard/compare_with_obs.jl

@@ -1,16 +1,15 @@
 const OBS_DS = Dict()
+const SIM_DS_KWARGS = Dict()
 
 function preprocess_pr_fn(data)
-    # 1 mm/day -> - 1 kg/m/s2
-    # The minus sign comes from the different conventions used
-    return data .* Float32(-1 / 86400)
+    # -1 kg/m/s2 -> 1 mm/day
+    return data .* Float32(-86400)
 end
 
-OBS_DS["pr"] = ObsDataSource(;
-    path = joinpath(pr_obs_data_path(), "gpcp.precip.mon.mean.197901-202305.nc"),
-    var_name = "precip",
-    preprocess_data_fn = preprocess_pr_fn,
-)
+OBS_DS["pr"] =
+    ObsDataSource(; path = joinpath(pr_obs_data_path(), "gpcp.precip.mon.mean.197901-202305.nc"), var_name = "precip")
+
+SIM_DS_KWARGS["pr"] = (; preprocess_data_fn = preprocess_pr_fn, new_units = "mm / day")
 
 # OBS_DS["rsut"] = ObsDataSource(;
 #                              path = "OBS/CERES_EBAF-TOA_Ed4.2_Subset_200003-202303.g025.nc",
@@ -24,7 +23,7 @@ OBS_DS["pr"] = ObsDataSource(;
 
 function bias(output_dir::AbstractString, short_name::AbstractString, target_dates::AbstractArray{<:Dates.DateTime})
     obs = OBS_DS[short_name]
-    sim = SimDataSource(; path = output_dir, short_name)
+    sim = SimDataSource(; path = output_dir, short_name, SIM_DS_KWARGS["pr"]...)
     return bias(obs, sim, target_dates)
 end
 

experiments/ClimaEarth/user_io/leaderboard/data_sources.jl

@@ -19,8 +19,7 @@ struct ObsDataSource
     """Name of the latitude dimension in the NetCDF file"""
     lat_name::AbstractString
 
-    """Function that has to be applied to the data to convert it to the same conventions
-    as CliMA"""
+    """Function that has to be applied to the data to convert it to different units"""
     preprocess_data_fn::Function
 
     """The NCDataset associated to the file"""
@@ -75,17 +74,32 @@ struct SimDataSource
 
     """Simulation longitudes and latitudes"""
     lonlat::Tuple{AbstractArray, AbstractArray}
+
+    """Function that has to be applied to the data to convert it to different units"""
+    preprocess_data_fn::Function
+
+    # TODO: This should be handled by ClimaAnalysis
+    """preprocess_data_fn is typically used to change units, so we have to tell ClimaAnalysis what the new
+    units are."""
+    new_units::Union{Nothing, AbstractString}
 end
 
-function SimDataSource(; path, short_name, reduction = "average", period = "10d")
+function SimDataSource(;
+    path,
+    short_name,
+    reduction = "average",
+    period = "10d",
+    preprocess_data_fn = identity,
+    new_units = nothing,
+)
 
     sim = ClimaAnalysis.SimDir(path)
     # TODO: Add period, for the time-being, we just pick up what's there
     var = get(sim; short_name, reduction)
 
     lonlat = (var.dims["lon"], var.dims["lat"])
 
-    return SimDataSource(path, short_name, reduction, period, var, lonlat)
+    return SimDataSource(path, short_name, reduction, period, var, lonlat, preprocess_data_fn, new_units)
 end
 
 """
@@ -96,5 +110,5 @@ Return the simulation data at the given date.
 function data_at_date(sim_ds::SimDataSource, date::Dates.DateTime)
     start_date = Dates.DateTime(sim_ds.var.attributes["start_date"])
     time_diff_seconds = (date - start_date) / Dates.Second(1)
-    return ClimaAnalysis.slice(sim_ds.var, time = time_diff_seconds).data
+    return sim_ds.preprocess_data_fn(ClimaAnalysis.slice(sim_ds.var, time = time_diff_seconds).data)
 end

experiments/ClimaEarth/user_io/leaderboard/utils.jl

@@ -148,7 +148,7 @@ function bias(obs_ds::ObsDataSource, sim_ds::SimDataSource, target_dates::Abstra
     rmse = round(sqrt(integrate_on_sphere(mse_arr, lonlat)); sigdigits = 3)
     global_bias = round(integrate_on_sphere(bias_arr, lonlat); sigdigits = 3)
 
-    units = sim_ds.var.attributes["units"]
+    units = isnothing(sim_ds.new_units) ? sim_ds.var.attributes["units"] : sim_ds.new_units
 
     bias_attribs = Dict{String, Any}(
         "short_name" => "sim-obs_$short_name",

test/experiment_tests/leaderboard.jl

@@ -53,16 +53,18 @@ end
 @testset "Leaderboard" begin
     simdir = ClimaAnalysis.SimDir(@__DIR__)
 
-    sim_datasource = Leaderboard.SimDataSource(path = @__DIR__, short_name = "pr")
+    preprocess_fn = (data) -> data .* Float32(-1 / 86400)
+
+    # The conversion is technically not correct for this data source, but what
+    # we care about here is that preprocess_data_fn works
+    sim_datasource = Leaderboard.SimDataSource(path = @__DIR__, short_name = "pr", preprocess_data_fn = preprocess_fn)
 
     pr = get(simdir, "pr")
 
     @test sim_datasource.lonlat[1] == pr.dims["lon"]
     @test sim_datasource.lonlat[2] == pr.dims["lat"]
 
-    @test Leaderboard.data_at_date(sim_datasource, Dates.DateTime(1979, 1, 2)) == pr.data[1, :, :]
-
-    preprocess_fn = (data) -> data .* Float32(-1 / 86400)
+    @test Leaderboard.data_at_date(sim_datasource, Dates.DateTime(1979, 1, 2)) == preprocess_fn(pr.data[1, :, :])
 
     obs_datasource = Leaderboard.ObsDataSource(;
         path = joinpath(pr_obs_data_path(), "gpcp.precip.mon.mean.197901-202305.nc"),