C
CI-EWS·CARBETA
Log out

Climate is the upstream driver of every disease signal in this system. This page quantifies how rainfall, temperature, humidity, and vegetation couple with malaria — and when, through measurable lag windows — all from real ERA5 reanalysis.

01Climate Coupling

Humidity leads Malaria most at 0 months (ρ = 0.44)

Live weather, climate anomaly stripes, and the measured climate→malaria coupling.

Live weather · CAR
<20°C 20-24°C 24-28°C >28°C
Climate Anomaly Stripes
Annual departure from the 1991–2020 climatology · 2021–2026
26.7°CMean temp · -0.43°C
62mmMean rain / mo · -1%
0.00Veg (LAI) · 0σ
TEMPERATURE ANOMALY (°C)
-0.1
-0.6
-0.2
+0.1
-0.4
-0.4
■ CoolBaselineWarm ■
RAINFALL ANOMALY (%)
-1%
-1%
-1%
-1%
-1%
-1%
■ DryBaselineWet ■
VEGETATION — ERA5 LAI (σ)
0.00
0.00
0.00
0.00
0.00
0.00
■ SparseBaselineLush ■
202120222023202420252026
Source: ERA5 reanalysis (ECMWF), district-level monthly. Vegetation = ERA5 Leaf Area Index; MODIS NDVI (higher-res) is a pending upgrade.
Disease:

Climate-Disease Pathways — Malaria

RAINFALL · 0mo
Malaria couples most with rainfall at same month (ρ = +0.35, p < 0.05).

Malaria climbs as the rains arrive — pooled water breeds Anopheles mosquitoes, so confirmed cases track rainfall with little to no delay (roughly a 0–1 month lead).

CLIMATE DRIVERSSTRONGEST-LAG ρMALARIA+0.35·0mo+0.33·3mo-0.19·1mo+0.34·1moRAINRainfallstanding water & wet-season conditionsTEMPTemperatureambient heat → vector/pathogen cycleRAINRainfall anomalydeparture from the normal rainfallTEMPTemperature anomalydeparture from the normal temperatureMALA+ (more → more)− (more → fewer)not significantsolid node = headline driver · n = 48 months

Malaria — correlation by lag

n = 48 mo
-0.50-0.250.00+0.25+0.500mo1mo2mo3moclimate lead (months)
RainfallTemperatureRainfall anomalyTemperature anomalyfilled dot = significant (p < 0.05)

Which climate signal drives which disease

ALL DISEASES
RainfallTemperatureRainfall anomalyTemperature anomalyMalaria+0.35*lag 0mo+0.33*lag 3mo-0.19lag 1mo+0.34*lag 1moMeningitis-0.47*lag 2mo+0.40*lag 0mo+0.26*lag 2mo-0.40*lag 2moYellow Fever+0.52*lag 0mo+0.31*lag 3mo+0.15lag 0mo+0.29*lag 2moMeasles-0.31*lag 2mo+0.31*lag 0mo+0.22lag 1mo-0.30*lag 1moAFP-0.35*lag 3mo+0.32*lag 0mo+0.34*lag 3mo-0.41*lag 2moMpox+0.13lag 2mo-0.11lag 2mo+0.22lag 1mo+0.21lag 3moPertussis+0.13lag 0mo-0.17lag 0mo-0.07lag 2mo-0.12lag 1mo
Cell = each disease's strongest-lag correlation for that driver. Blue = positive, red = negative; bold* = significant (p < 0.05). Indigo outline = that disease's headline driver. Click a row to load it above.

Climate forecastability ladder

RANKED BY LEAD-TIME
How far ahead each disease can be warned about from climate — lead-time, not just correlation. Bar length is the usable weeks of warning (the lag of the strongest significant coupling); colour intensity is |r|. Diseases with no significant coupling fall into the grey climate-blind zone — don't build climate triggers for them.
MeningitisBEST
8 wk
8 wk (~2 mo)
Rainfall · r -0.47
AFP
8 wk
8 wk (~2 mo)
Temperature anomaly · r -0.41
Measles
8 wk
8 wk (~2 mo)
Rainfall · r -0.31
Malaria
4 wk
4 wk (~1 mo)
Rainfall · r +0.35
Yellow Fever
concurrent — no lead
0 wk lead
Rainfall · r +0.52
Mpox
no useful lead · climate-blind
not forecastable
Rainfall anomaly · r +0.22 n.s.
Pertussis
no useful lead · climate-blind
not forecastable
Temperature · r -0.17 n.s.
0 (concurrent)4 wk8 wk
weeks of warning →
Each disease's strongest significant climate coupling (national, from ciews_gold__climate_coupling); the lag of that coupling is the usable lead. Deeper colour = larger |r|. Grey striped = no significant coupling (climate-blind). Yellow fever couples strongly but concurrently — a strong signal with no lead.
Context, not forecastCommodity-stockout and immunization couplings are descriptive / operational context, not predictive drivers — within-district lag correlations are near-null (Penta3→measles ≈ 0 at every lag). The climate signals ranked above are the platform's actual forecasting inputs.

Each disease answers to a different climate signal — and none act the same month; they lead it. The next section takes the flagship malaria coupling apart lag by lag.

02Lag Analysis

Strongest coupling: Humidity at lag 0mo → Malaria (ρ = 0.44)

National monthly correlation over 48 months. Pick a lag to highlight it.

Lag window:
Positive = climate before the case count (the causal direction). Negative = climate after the case count — a reverse check; a real driver should be strong at positive lag and weak at negative lag.

Lag Correlation Matrix

MALARIA
-6mo-5mo-4mo-3mo-2mo-1mo0mo+1mo+2mo+3mo+4mo+5mo+6morainfall-0.42*-0.34*-0.08+0.17+0.29*+0.40*+0.35*+0.31*+0.17-0.05-0.22-0.39*-0.43*temperature+0.30*+0.220.00-0.21-0.35*-0.40*-0.29*-0.13+0.07+0.33*+0.39*+0.36*+0.35*humidity-0.42*-0.29*-0.08+0.09+0.25+0.41*+0.44*+0.38*+0.21-0.08-0.34*-0.42*-0.40*vegetation0.000.000.000.000.000.000.000.000.000.000.000.000.00
Blue = positive, red = negative coupling. Bold* = significant (p < 0.05).

Rainfall vs malaria

+4mo
Rainfall mm (lag +4mo)Malaria casesρ = -0.22
03Where & when

District climate now, and CAR's seasonal rhythm

The real ERA5 depth: this month per district, and the average year that shapes transmission.

District climate — 2026-06

35 DISTRICTS
DistrictTempRainHumidVegAnomaly
Nanga-Boguila26.1°168mm79%—R+1.5T-0.2
Paoua25.7°150mm78%—R+1.0T-0.3
Batangafo-Kabo26.9°148mm78%—R+1.3T-0.0
Bossangoa25.4°145mm82%—R+1.1T-0.5
Ngaoundaye24.5°120mm77%—R-1.1T+0.2
Nana-Gribizi26.3°114mm78%—R-0.1T+0.4
Bamingui-Bangoran27.6°113mm72%—R+0.4T+1.4
Bozoum-Bossemptele24.3°110mm82%—R-0.6T+0.1
Kouango-Grimari25.6°103mm83%—R-0.3T-0.2
Bocaranga-Koui22.7°98mm81%—R-1.9T+0.3
Bambari25.7°95mm78%—R-1.1T+1.3
Haute-Kotto25.3°93mm73%—R-1.5T+2.5
Bouca25.3°92mm85%—R-0.4T-0.9
Alindao-Mingala25.8°91mm80%—R-0.4T+1.2
Mbaiki25.4°90mm84%—R-1.1T+0.7
Bangassou25.4°89mm76%—R-0.9T+1.8
Kemo25.8°83mm81%—R-1.2T+0.3
Boda24.4°82mm85%—R-0.9T-0.4
Haut-Mbomou26.0°78mm72%—R-0.8T+1.8
Ouango-Gambo25.5°76mm80%—R-1.0T+0.5
Bouar -Baoro23.6°76mm84%—R-2.0T-0.4
Begoua25.8°74mm79%—R-1.1T+0.5
Sangha-Mbaere24.4°72mm85%—R-1.4T+0.6
Bossembele25.1°71mm80%—R-1.5T+0.3
Mobaye-Zangba25.8°70mm79%—R-1.0T+0.3
Kembe-Satema25.6°68mm81%—R-1.1T+0.5
Carnot-Gadzi24.2°62mm84%—R-1.8T+0.3
Baboua-Abba23.3°61mm84%—R-2.2T-0.2
Gamboula24.2°60mm86%—R-1.4T+0.7
Vakaga29.1°58mm59%—R-1.0T+1.4
Bimbo25.9°58mm82%—R-1.2T+0.4
Bangui Ii26.3°52mm81%—R-1.5T+0.4
Berberati24.3°50mm85%—R-2.4T+0.9

Seasonal climate calendar

CLIMATOLOGY
0103206JFMAMJJASOND24°26°28°
rainfall (mm) · bold = rainy seasontemperature (°C)

Steady-state coupling explains the typical season. The final section asks: if the next season shifts, how much should we prepare for?

04Scenarios

Climate scenario

A fitted distributed-lag non-linear model (DLNM).

DLNM has not been fitted yet.
So what do we do about itAlerts — Every district classified, every trigger explained, every action pre-authorised.→