Run a simulation on the Simulator page first, then return here to view the summary.
Loading…
📊 Key Indicators
Max Simulated
—
Baseline scenario
Average Simulated
—
Mean of draws
Threshold Exceeded
—
—
Risk Level
—
—
Observed vs Simulated — Annual Maxima
Return Levels (Gumbel EVT)
📈 Observed vs Simulated Comparison
Year
Observed (mm)
Sim. Mean (mm)
±1 SD
Obs. within ±1 SD?
How to read this table: If the observed value falls within Sim. Mean ± 1 SD, the simulation is realistic. Values above mean+1SD suggest an extreme event; values below mean−1SD indicate an unusually dry year.
🔬 Validation Results
Wilcoxon p: —
McNemar p: —
t-test p: —
CV (obs): —
Test
Result
Formula
Interpretation
Wilcoxon Signed-Rank
—
W = Σ rank(|dᵢ|)
—
McNemar's Test
—
χ² = (b−c)² / (b+c)
—
Welch's t-Test
—
t = (μ₁−μ₂) / √(σ₁²/n₁+σ₂²/n₂)
—
Coefficient of Variation
—
CV = (σ/μ)×100%
—
Note: p < 0.05 in Wilcoxon or t-test indicates model predictions significantly differ from observations. McNemar p > 0.05 indicates good classification agreement. Lower CV = more stable predictions.
📐 Goodness-of-Fit (Gumbel Distribution)
K-S D: —
A-D A²: —
Test
Statistic
Formula
Interpretation
Kolmogorov–Smirnov
—
D = sup|F₀(x)−Fₙ(x)|
—
Anderson–Darling
—
A² = −n − (1/n)Σ(2i−1)(lnF(xᵢ)+ln(1−F(xₙ₊₁₋ᵢ)))
—
Gumbel fit quality: K-S p > 0.05 and smaller A² both indicate a better-fitting Gumbel distribution. These tests validate whether EVT return level estimates are statistically reliable.
