YEt another coefficient update

Author: tapios

dev/SoilAlbedo/README.md

@@ -72,15 +72,15 @@ by ~0.3 to reduce the overall upwelling shortwave bias in coupled simulations.
 
 ### PAR Band
 ```julia
-η₀_PAR = -2.0      # Bias-corrected intercept (raw fit: -3.04)
+η₀_PAR = -2.25     # Bias-corrected intercept (raw fit: -3.04)
 c_om_PAR = -0.13   # Organic matter (darkening)
 c_vgn_PAR = 1.24   # van Genuchten n (texture)
 c_cf_PAR = 0.15    # Coarse fragments (rocks)
 ```
 
 ### NIR Band
 ```julia
-η₀_NIR = -2.05     # Bias-corrected intercept (raw fit: -3.10)
+η₀_NIR = -2.3      # Bias-corrected intercept (raw fit: -3.10)
 c_om_NIR = -0.14
 c_vgn_NIR = 1.28
 c_cf_NIR = 0.16

src/standalone/Soil/soil_albedo.jl

@@ -278,8 +278,8 @@ end
 
 """
     CompositionBasedSoilAlbedo{FT}(;
-        η₀_PAR = FT(-2.0),
-        η₀_NIR = FT(-2.05),
+        η₀_PAR = FT(-2.25),
+        η₀_NIR = FT(-2.3),
         c_om_PAR = FT(-0.13),
         c_om_NIR = FT(-0.14),
         c_vgn_PAR = FT(1.24),
@@ -306,13 +306,13 @@ Coefficients fitted via logistic regression on:
 # Expected albedo values
 
 **High vg_n desert (sandy)**: vg_n ≈ 2.5, cf ≈ 0.3, om ≈ 0
-- η ≈ -2.0 + 1.24*2.5 + 0.15*0.3 = 1.15 → α ≈ 0.47
+- η ≈ -2.25 + 1.24*2.5 + 0.15*0.3 = 0.90 → α ≈ 0.44
 
 **Rocky desert**: vg_n ≈ 1.5, cf ≈ 0.6, om ≈ 0
-- η ≈ -2.0 + 1.24*1.5 + 0.15*0.6 = -0.04 → α ≈ 0.31
+- η ≈ -2.25 + 1.24*1.5 + 0.15*0.6 = -0.29 → α ≈ 0.28
 
 **Organic soil**: vg_n ≈ 1.5, cf ≈ 0.1, om ≈ 0.1
-- η ≈ -2.0 + 1.24*1.5 - 0.13*0.1 + 0.15*0.1 = -0.12 → α ≈ 0.30
+- η ≈ -2.25 + 1.24*1.5 - 0.13*0.1 + 0.15*0.1 = -0.37 → α ≈ 0.27
 
 # Arguments
 - `η₀_PAR`, `η₀_NIR`: Base log-odds (intercept in logistic model)
@@ -325,8 +325,8 @@ Coefficients fitted via logistic regression on:
 - `albedo_calc_top_thickness`: Depth for surface moisture averaging (m)
 """
 function CompositionBasedSoilAlbedo{FT}(;
-    η₀_PAR = FT(-2.0),
-    η₀_NIR = FT(-2.05),
+    η₀_PAR = FT(-2.25),
+    η₀_NIR = FT(-2.3),
     c_om_PAR = FT(-0.13),
     c_om_NIR = FT(-0.14),
     c_vgn_PAR = FT(1.24),