3-STEP ALBEDO OPTIMIZATION

ORCHIDEE simulation configuration :


STEP 1 : LEAF ALBEDO OPTIMIZATION

Data points (month-pixels) selection : The graph below shows the number of data points (months) selected at each grid cell (from 0 to 12) :

Optimization method : L-BFGS-B with 14 optimized parameters, gradient calculation using analytical derivative

First test : wide ranges of variation, global data

LEAF_ALB_NIRPFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
prior0.230.180.180.20.240.150.260.20.240.270.280.260.240.24
min0.10.10.10.10.10.10.10.10.10.10.10.10.10.1
max0.30.30.30.30.30.30.30.30.40.40.40.40.40.4
post0.2050.1860.150.1760.2350.1290.2710.1860.290.2610.2590.2890.2320.311
LEAF_ALB_VISPFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
prior0.040.040.040.040.030.030.030.030.060.060.060.060.060.06
min0.010.010.010.010.010.010.010.010.010.010.010.010.010.01
max0.050.050.050.050.050.050.050.050.10.10.10.10.10.1
post0.02220.02960.02960.02340.01520.03040.02970.010.04650.03130.05240.06080.010.0217

Second test : tuned ranges of variation, global data

LEAF_ALB_NIRPFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
prior0.230.180.180.20.240.150.260.20.240.270.280.260.240.24
min0.170.170.10.10.160.10.10.160.10.170.10.150.10.2
max0.250.240.230.240.240.190.330.230.360.360.460.360.290.35
post0.2040.1890.150.1760.2280.1290.2630.1780.2960.2580.2590.2920.2220.323
LEAF_ALB_VISPFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
prior0.040.040.040.040.030.030.030.030.060.060.060.060.060.06
min0.020.020.010.010.020.010.020.020.020.040.020.050.020.04
max0.060.080.090.080.070.050.040.040.180.180.230.150.160.1
post0.02250.020.02310.02310.020.02340.020.020.04640.04850.05130.06060.020.04

AMAZON

LEAF_ALB_NIRPFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
prior0.230.180.180.20.240.150.260.20.240.270.280.260.240.24
min0.170.170.10.10.160.10.10.160.130.180.10.240.10.2
max0.240.240.230.220.210.190.330.230.270.270.460.30.260.27
post0.2110.1850.150.220.210.1290.2630.1780.2610.270.2590.2640.260.24
LEAF_ALB_VISPFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
prior0.040.040.040.040.030.030.030.030.060.060.060.060.060.06
min0.020.020.010.020.020.010.020.020.030.040.020.050.020.07
max0.030.050.060.050.040.050.040.040.110.070.230.080.090.1
post0.02260.02370.02310.02080.020.02340.020.020.04540.040.05130.050.020.07

SAHEL

LEAF_ALB_NIRPFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
prior0.230.180.180.20.240.150.260.20.240.270.280.260.240.24
min0.170.190.120.130.160.10.10.160.10.180.110.190.10.2
max0.230.240.230.240.240.190.330.230.360.270.460.360.290.35
post0.180.190.120.2010.160.1290.2630.1780.2960.270.2350.3170.2220.323
LEAF_ALB_VISPFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
prior0.040.040.040.040.030.030.030.030.060.060.060.060.060.06
min0.020.030.020.020.020.010.020.020.020.040.040.080.020.04
max0.060.080.090.080.070.050.040.040.180.070.230.140.160.1
post0.02310.030.090.020.020.02340.020.020.04640.05910.040.09930.020.04

SIBERIA

LEAF_ALB_NIRPFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
prior0.230.180.180.20.240.150.260.20.240.270.280.260.240.24
min0.170.170.10.10.160.120.240.170.190.170.220.150.10.2
max0.250.240.230.240.240.190.330.230.280.360.290.360.290.35
post0.2040.1890.150.1760.2280.1430.2440.1810.190.2580.2740.2920.2220.313
LEAF_ALB_VISPFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
prior0.040.040.040.040.030.030.030.030.060.060.060.060.060.06
min0.020.020.010.010.020.020.020.020.050.040.040.050.020.04
max0.060.080.090.080.070.030.040.040.10.180.060.150.160.1
post0.02250.020.02310.02310.020.020.020.020.07270.04850.040.06060.020.04


STEP 2 : SNOW ALBEDO OPTIMIZATION

Data points (month-pixels) selection : The graph below shows the number of data points (months) selected at each grid cell (from 0 to 12) :

Optimization method : L-BFGS-B with 24 optimized parameters, gradient calculation using finite differences approach

SNOW_ALB_NIRPFT1PFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
snowa_aged_prior0.5000.10.370.080.160.170.270.440.440.440.4400.44
snowa_dec_prior0.13000.10.10.160.040.070.080.120.120.120.1200.12
min(aged+dec)0.05000.050.050.050.050.050.050.050.050.050.0500.05
max(aged+dec)0.95000.950.950.950.950.950.950.950.950.950.9500.95
min(aged/(aged+dec))0.05000.050.050.050.050.050.050.050.050.050.0500.05
max(aged/(aged+dec))0.95000.950.950.950.950.950.950.950.950.950.9500.95
snowa_aged_post0.385000.2030.9020.1330.1750.4380.2590.6360.6980.5360.61500.579
snowa_dec_post0.163000.01070.04750.006990.00920.02310.01360.03350.03680.02820.032400.0305
SNOW_ALB_VISPFT1PFT2PFT3PFT4PFT5PFT6PFT7PFT8PFT9PFT10PFT11PFT12PFT13PFT14PFT15
snowa_aged_prior0.74000.080.240.070.180.180.330.570.570.570.5700.57
snowa_dec_prior0.21000.140.080.170.050.060.090.150.150.150.1500.15
min(aged+dec)0.05000.050.050.050.050.050.050.050.050.050.0500.05
max(aged+dec)0.95000.950.950.950.950.950.950.950.950.950.9500.95
min(aged/(aged+dec))0.05000.050.050.050.050.050.050.050.050.050.0500.05
max(aged/(aged+dec))0.95000.950.950.950.950.950.950.950.950.950.9500.95
snowa_aged_post0.633000.06850.9020.009590.1090.6270.2970.9020.9020.7420.7700.862
snowa_dec_post0.234000.00360.04750.04040.005750.0330.01560.04750.04750.0390.040500.0454


STEP 3 : BACKGROUND ALBEDO OPTIMIZATION

Data points (month-pixels) selection : The graph below shows the number of data points (months) selected at each grid cell (from 0 to 12) :

Optimization method : L-BFGS-B with 1 optimized parameter individually at each grid cell

RESULTS (seasonal cycle across different regional masks)