diff --git a/papers/ice_nucleation_2024/calibration.jl b/papers/ice_nucleation_2024/calibration.jl
index 62ddb8328..21b15baf1 100644
--- a/papers/ice_nucleation_2024/calibration.jl
+++ b/papers/ice_nucleation_2024/calibration.jl
@@ -263,8 +263,8 @@ function calibrate_J_parameters(FT, IN_mode, params, IC, y_truth, Γ,; perfect_m
     rng = Random.seed!(Random.GLOBAL_RNG, rng_seed)
 
     prior = create_prior(FT, IN_mode, perfect_model = perfect_model)
-    N_ensemble = 10      # runs N_ensemble trials per iteration
-    N_iterations = 150    # number of iterations the inverse problem goes through
+    N_ensemble = 25      # runs N_ensemble trials per iteration
+    N_iterations = 100    # number of iterations the inverse problem goes through
 
     # Generate initial ensemble and set up EKI
     initial_ensemble = EKP.construct_initial_ensemble(rng, prior, N_ensemble)
diff --git a/papers/ice_nucleation_2024/calibration_setup.jl b/papers/ice_nucleation_2024/calibration_setup.jl
index 63855b44d..6a0a6039b 100644
--- a/papers/ice_nucleation_2024/calibration_setup.jl
+++ b/papers/ice_nucleation_2024/calibration_setup.jl
@@ -137,7 +137,7 @@ function perf_model_pseudo_data(FT, IN_mode, params, IC)
     G_truth = run_model(params, coeff_true, IN_mode, FT, IC)
     dim_output = length(G_truth)
 
-    Γ = 0.03 * LinearAlgebra.I * (maximum(G_truth) - minimum(G_truth))
+    Γ = 0.05 * LinearAlgebra.I * (maximum(G_truth) - minimum(G_truth))
     noise_dist = Distributions.MvNormal(zeros(dim_output), Γ)
 
     y_truth = zeros(length(G_truth), n_samples) # where noisy ICNC will be stored