This version has no lead-time dimension: it's a same-month, no-lag regression meant to capture the low-frequency signal rather than a propagating subseasonal one.
Map Key
Reading the chart
Color fill
Correlation (r), full ±1 scale
Pearson correlation coefficient (r) between the monthly-mean VP200 index and monthly-mean 500 mb height: red = positive (height tends to rise when the VP200 index rises, i.e. when convection is enhanced), blue = negative. Correlation, not a raw regression in meters, because 500 mb height naturally varies far less in the tropics than at high latitudes: a raw-meters map would make high-latitude features look mechanically "bigger" regardless of how tight the actual relationship is there. Correlation puts every point on the same -1..1 scale.
Dashed contours
Green = enhanced convection, orange = suppressed
The VP200 pattern itself, same month as the VP200 index (no lag): green means enhanced convection (positive VP200 index, upper-level outflow), orange means suppressed convection (negative VP200 index, drier and subsiding).
Black box
Source region for the VP200 index
The region whose VP200 was averaged into the VP200 index driving this map, shaded gray so it stands out from the correlation coloring underneath.
Stippling
Passed an 80% significance test
The correlation at that point passed a significance test at the 80% level (a permutation test that shuffles which year's VP200 value is paired with which year's height field). At 80% there's roughly a 1-in-5 chance any single stippled point is a false positive, so isolated, scattered dots deserve real skepticism; a broad, spatially coherent patch is much better evidence of a genuine signal. Stippling is also intentionally removed within 20 degrees of the equator regardless of significance: VP200 and height are tightly, near-mechanically coupled in the deep tropics. The more informative signal is the stippling that remains, out in the extratropics, where it's patchier and more selective.
★
Star
Texas, shown for reference
Texas, shown for reference throughout.
Signal processing
1
True VP200 (not a proxy)
Computed via spherical harmonics, not raw divergence
200 mb velocity potential (VP200) is derived from 200 mb u/v wind by computing horizontal divergence, then inverting the Poisson equation via a global spherical harmonic transform to recover VP200 itself. VP200 is oriented here so that positive values mean enhanced convection, the opposite of its native sign convention (where enhanced convection is negative). It's flipped throughout this tool so "positive = more active convection" reads the same way as everything else on the page.
2
Detrending
Long-term warming trend removed
Before any of this is computed, both the VP200 index and 500 mb height have a linear trend over the full 1979-2025 record removed (fit at daily resolution before the monthly averaging). 500 mb height has a real, well-documented long-term rise (tropospheric warming raises the height of pressure surfaces), and without removing it, two variables that are each just drifting upward over the decades could look spuriously correlated even with no genuine year-to-year relationship.
Scope & design notes
SCOPE
No lag, by design
Monthly averaging is itself a low-pass filter
Each daily anomaly (day's value minus a smoothed seasonal-cycle average and its long-term trend) is averaged within its calendar month before this correlation is computed. That monthly averaging suppresses fast weather-timescale noise and approximates a low-pass filter, capturing slower-moving variability rather than a propagating subseasonal signal.
GUIDE
Active Convection is a guide
A reliable indicator here, but boxes are a fixed partition
The eight "Active Convection" options are fixed longitude sectors aligned with the standard 8-phase MJO/RMM convention, useful as a starting menu of tropical source regions. Since this is a monthly mean (averaged over roughly 30 days), a strong signal in a given box is a fairly reliable sign that convection really was elevated, or suppressed, there that month, not just daily noise. But the box edges are a convenient fixed partition, not a natural one: the actual footprint of whatever is driving the signal can span parts of two adjacent boxes or sit off-center within one, so it's still worth checking the dashed green/orange contours for the real spatial pattern. During El Nino years specifically, the relevant VP200 signal is often centered in the Phase 6 (Western Pacific) or Phase 7 (Central Pacific) boxes, since that's roughly where El Nino's characteristic warm-water and convective shift sits, so those two are a good starting point to check.