minor

docs/make.jl

@@ -11,7 +11,7 @@ using Literate
 using Plots # to not capture precompilation output
 
 # generate examples
-EXAMPLE = joinpath(@__DIR__, "..", "examples", "CalibOneSite.jl")
+EXAMPLE = joinpath(@__DIR__, "..", "examples", "Calib_site.jl")
 OUTPUT = joinpath(@__DIR__, "src/generated")
 
 preprocess(str) = replace(str, "x = 123" => "y = 321"; count=1)
@@ -72,7 +72,7 @@ makedocs(
     # "customprocessing.md",
     # "documenter.md",
     # "tips.md",
-    "generated/CalibOneSite.md",
+    "generated/Calib_site.md",
     # "changelog.md",
   ],
   warnonly=true,

examples/CalibOneIGBP.jl → examples/Calib_IGBP.jl

@@ -1,5 +1,7 @@
 using PML, Ipaper, Test, RTableTools
-df_out, df, par = deserialize(file_FLUXNET_CRO)
+
+df_out, df, _par = deserialize(file_FLUXNET_CRO)
+par = Param_PMLV2(; _par..., hc=0.5)
 df.GPP_obs = df.GPPobs
 df.ET_obs = df.ETobs
 r = PMLV2(df; par)

examples/CalibOneSite.jl → examples/Calib_site.jl

@@ -4,15 +4,19 @@
 using PML, Ipaper
 
 #nb # %% A slide [markdown] {"slideshow": {"slide_type": "slide"}}
-df_out, df, par = deserialize(file_FLUXNET_CRO_USTwt)
+df_out, df, _par = deserialize(file_FLUXNET_CRO_USTwt)
+par = Param_PMLV2(; _par..., hc=0.5)
 df.GPP_obs = df.GPPobs
 df.ET_obs = df.ETobs
 
 # ## 模型参数率定
+parNames = [
+  :α, :η, :g1, :Am_25, :VPDmin, :VPDmax, :D0, :kQ, :kA, :S_sls, :fER0 # :hc
+]
 
 #nb # %% A slide [markdown] {"slideshow": {"slide_type": "fragment"}}
-theta, goal, flag = ModelCalib(df, par0)
-df_out = PMLV2_sites(df; par=theta2par(theta))
+theta, goal, flag = ModelCalib(df, par0, parNames)
+df_out = PMLV2_sites(df; par=theta2par(theta, parNames))
 df_out[1:10, :]
 
 # ## 拟合优度

src/DataType.jl

@@ -1,3 +1,7 @@
+export interm_PML, output_PML
+export to_mat;
+
+
 @with_kw mutable struct interm_PML{T}
   ET::T = 0.0
   GPP::T = 0.0
@@ -74,19 +78,3 @@ function to_df(res::output_PML{T}) where {T<:Real}
   data, names = to_mat(res)
   DataFrame(data, names)
 end
-
-function map_df_tuple(fun::Function, lst::GroupedDataFrame{DataFrame}, args...; kw...)
-  n = length(lst)
-  _keys = keys(lst)
-  map(i -> begin
-      d = lst[i]
-      key = NamedTuple(_keys[i])
-      r = fun(d, args...; kw...)
-      (; key..., r...)
-    end, 1:n)
-end
-
-
-export interm_PML, output_PML
-export to_mat;
-export map_df_tuple

src/ModelCalib.jl

@@ -3,11 +3,9 @@ export model_goal
 export model_gof
 
 include("DataType.jl")
-# include("model_gof.jl")
 
 function ModelCalib_IGBPs(data::AbstractDataFrame;
-  parNames=ParNames,
-  of_gof=:KGE, maxn=2500)
+  parNames=ParNames, of_gof=:KGE, maxn=2500)
 
   vars = ["IGBPname", "IGBPcode", "site", "date", "GPP_obs", "ET_obs",
     "Prcp", "Tavg", "U2", "Rn", "Rs", "VPD", "LAI", "Pa", "Ca"]
@@ -16,8 +14,7 @@ function ModelCalib_IGBPs(data::AbstractDataFrame;
   @time params = par_map(IGBP -> begin
       df = data[data.IGBP.==IGBP, vars]
       IGBPcode = df.IGBPcode[1]
-      theta, _, _ = ModelCalib(df, par0, parNames; IGBPcode, of_gof, maxn, verbose=false)
-      theta
+      ModelCalib(df, par0, parNames; IGBPcode, of_gof, maxn, verbose=false)[1]
     end, IGBPs; progress=false)
 
   # printstyled("[i=$i, IGBP = $IGBP] \n", bold=true, color=:green, underline=false)
@@ -79,17 +76,15 @@ end
 
 ## gof of ET and GPP for ALL IGBP
 function model_gof(df_out::DataFrame; all=true)
-  fun_et(d) = GOF(d.ET_obs, d.ET)
-  fun_gpp(d) = GOF(d.GPP_obs, d.GPP)
-
   lst = groupby(df_out, :IGBP)
 
-  res = map_df_tuple(fun_et, lst)
-  all && push!(res, (; IGBP="ALL", fun_et(df_out)...))
-  gof_et = DataFrame(res)
+  fun_et(d) = GOF(d.ET_obs, d.ET)
+  fun_gpp(d) = GOF(d.GPP_obs, d.GPP)
+  func = (; ET=fun_et, GPP=fun_gpp)
 
-  res = map_df_tuple(fun_gpp, lst)
-  all && push!(res, (; IGBP="ALL", fun_gpp(df_out)...))
-  gof_gpp = DataFrame(res)
-  (; ET=gof_et, GPP=gof_gpp)
+  map(f -> begin
+      res = map_df_tuple(fun_et, lst)
+      all && push!(res, (; IGBP="ALL", fun_et(df_out)...))
+      DataFrame(res)
+    end, func)
 end

src/PML.jl

@@ -25,8 +25,8 @@ file_FLUXNET_CRO = "$dir_proj/data/CRO/FLUXNET_CRO" |> abspath
 file_FLUXNET_CRO_USTwt = "$dir_proj/data/CRO/FLUXNET_CRO_US-Twt" |> abspath
 
 
-include("main_Ipaper.jl")
-include("Params.jl")
+include("utilize.jl")
+include("Parameter.jl")
 include("ModelCalib.jl")
 include("ET_helper.jl")
 include("water_constrain.jl")

src/main_Ipaper.jl → src/utilize.jl

@@ -1,6 +1,7 @@
 # export struct2vec, struct2tuple
 export round2;
 export movmean2, nanmean2, getDataType, replace_miss
+export map_df_tuple
 
 # rounded_data = NamedTuple((field => round(value) for (field, value) in data))
 round2(x::NamedTuple, digits=3; kw...) = map(val -> round(val; digits), x)
@@ -47,7 +48,6 @@ function getDataType(x)
   typeof(type) == Union ? type.b : type
 end
 
-
 # function struct2vec(x)
 #   keys = fieldnames(typeof(x))
 #   vals = [getfield(x, key) for key in keys]
@@ -59,4 +59,17 @@ end
 #   vals = [getfield(x, key) for key in keys]
 #   (; zip(keys, vals)...)
 # end
+
 weighted_mean(x::AbstractVector, w::AbstractVector) = sum(x .* w) / sum(w)
+
+
+function map_df_tuple(fun::Function, lst::GroupedDataFrame{DataFrame}, args...; kw...)
+  n = length(lst)
+  _keys = keys(lst)
+  map(i -> begin
+      d = lst[i]
+      key = NamedTuple(_keys[i])
+      r = fun(d, args...; kw...)
+      (; key..., r...)
+    end, 1:n)
+end