Republic of Serbia · 5 to 11 August

Republic of Serbia had its highest fire week for this point in the year since 2012

Two questions, side by side. How bad was this week, measured against every week like it. How bad is the year, measured against every season on record. Different instruments, different units, and one of them is not finished.

This week · 5 to 11 August

3.8×

active-fire detections against the same-week mean of 111, 2012 to 2025
NASA FIRMS SNPP VIIRS, daily, 375 m
Cleared the anomaly gate for this week. The multiple rests on a thin baseline, so the rank is the sturdier reading. Standardised anomaly z = +5.2

Against the same week, every year

same-week mean 11120122026426

Each year’s detections in the same window of the year, same sensor throughout.

The seven days

05060708091011one seventh of a normal week, 16

7 of 7 days cleared one seventh of a normal week. The peak, 114 on 8 August, is 7 times that line.

This year · since 1 January

6,822 ha

burnt area mapped, 1.4× the average for this point in the year
EFFIS mapped perimeters, weekly, week 31, through 2026-08-05

Cumulative, every season since 2006

021kJanJulDecprevious record, 2024: 38,954 ha2026

Only this year carries hue; every grey line is one earlier season accumulating from January. The record season, 2024, reached 38,954 ha.

Week by week, 2026 so far

JanAprJulOctDec3,205 ha in week 2931 weeks in, 21 still to comeweeks 10 to 11, 1,949 ha

Week by week rather than cumulative, so a single heavy week is visible as one. The cell runs to week 52: 21 weeks of this season have not happened yet.

What this is

Two measurements of the same fire season at two time scales. The multiple is a rate: how much fire activity satellites detected this week against what this week normally looks like. The hectare figure is a stock: how much land has been mapped as burnt since January. A country can have an unremarkable week and still be having a record year, and the reverse is also true.

What this is not

Not one number at two zoom levels. The two figures come from different instruments with different latencies, and they are not convertible into each other. Not an attribution: fire seasons are driven by heat, drought, wind and land use, and the tag on this page states what is and is not established for this event. Not a forecast of where the season ends.

Baseline

Weekly multiple: same-week mean, 2012 to 2025.
Cumulative: complete seasons, 2006 to 2025.

Instruments

NASA FIRMS SNPP VIIRS
thermal anomaly counts, daily, 375 m

EFFIS burnt area
mapped perimeters, weekly. Area lands in the week it is mapped, which need not be the week it burned.

Revision

Mapped area for recent weeks rises as perimeters are completed, and a week’s area may be mapped after the week it burned. Published figures are not edited in place; corrections run forward.

The same week elsewhere

5.3xGeorgiaWell clear of its 2016 record week4.5xSpainWell clear of its 2015 record weeknot ENSO-linked3.9xCubaWell clear of its 2015 record week
Kristjan Lepik (2026). The Long Swell, Fires. thelongswell.com Republic of Serbia · week 31