Chapter 3 · Per-IADA Phenology Atlas

Ten Jelapang Padi, ten cropping calendars.

Each of Malaysia's ten national paddy granaries plants on its own schedule — shaped by river or pumped irrigation, by monsoon timing, by single- versus double-cropping, and by the design decade of its irrigation works. This page presents the seasonal phenological fingerprint of each IADA, derived from 72 months of Sentinel-2 NDVI (2020-2025) sampled at every v4 paddy parcel, then cross-validated against published authority cropping calendars.

Total parcels
9,636
across all 10 IADAs
Total area
176,840 ha
v4 boundary 2020-2025
Audit-pass rate
93.1%
PASS + RECENT_PADDY_PASS
NDVI observations
684,339
parcel-month rows

Reading a paddy cycle in NDVI

Lowland rice has a recognisable 4-6 month phenological signature in Sentinel-2 NDVI. Each cropping cycle moves through five stages tied to canopy density and water level.

NDVI < 0.20 · Land prep / flooded 0.20 - 0.35 · Tillering / early vegetative 0.35 - 0.55 · Mid-vegetative 0.55 - 0.75 · Heading / peak biomass > 0.75 · Maturity / dense canopy
Per-IADA cropping calendar — Sentinel-2 mean monthly NDVI 2020-2025
Per-IADA cropping calendar — Sentinel-2 mean monthly NDVI 2020-2025. Background colour bands assign each month a phase by NDVI; shaded grey band = ± year-to-year standard deviation; dashed lines mark peak (green) and trough (red) months.

Headline numbers

IADA Parcels v4 area (ha) Audit % Peak Trough Amp Cropping type
MADA3,27878,12898.4%Jan 0.71May 0.190.52Double (Musim Utama + Musim Luar)
KADA5745,54294.0%Sep 0.68Dec 0.250.43Mixed double + single
IADA Kerian1,42027,11095.7%Dec 0.65Mar 0.280.37Mixed single + double
IADA Barat Laut Selangor57016,72791.1%May 0.55Aug 0.330.23Single-cropping dominant
IADA Pulau Pinang1,58011,10994.4%Dec 0.70Apr 0.280.42Double-cropping
IADA Seberang Perak2876,41598.1%Nov 0.63Aug 0.330.30Mixed
IADA Kemasin-Semerak94011,64180.6%Sep 0.67Dec 0.370.30Single-cropping dominant
IADA Sungai Manik2807,89898.5%Nov 0.65Aug 0.300.35Mixed
IADA Ketara4604,31496.7%Mar 0.70Jan 0.230.47Double-cropping
IADA Rompin2477,95640.9%Aug 0.58Dec 0.430.15Single-cropping

When each Jelapang Padi peaks

All ten IADAs plotted on a single calendar reveal three phenological groups — early-year (Jan-Mar peak), late-year (Sep-Dec peak), and the two out-of-phase outliers BLS and Rompin.

Cross-IADA phenology fingerprint — when each Jelapang Padi peaks
Left: all v4 paddy parcels averaged per IADA. Right: only the double-cropping subset — two clear cycles per year visible in MADA, Kerian, PP, KADA, and Ketara.

Individual IADA profiles

Each card shows the IADA's phenological signature, cropping intensity breakdown, published authority schedule, and a one-paragraph interpretation. Twelve mini month cells show the assigned phase by colour and mean NDVI for that calendar month averaged 2020-2025.

MADA mada
Parcels
3,278
v4 area
78,128 ha
Audit kept
98.4%
Peak
Jan + Aug
Trough
May + Oct
Amplitude
0.52
Jan0.71
Feb0.54
Mar0.30
Apr0.23
May0.19
Jun0.45
Jul0.64
Aug0.69
Sep0.43
Oct0.20
Nov0.25
Dec0.61

Single: 283 · Double: 2,195 · Unknown: 800

Type: Double-cropping (Musim Utama + Musim Luar).
Main: Aug-Jan · Off: Feb-Jul.
Source: MADA Cropping Calendar 2025; Wong et al. (2017). Synchronised water release via Pedu and Muda dams.

Textbook double-cropping. Musim Utama planted Sep-Oct → peak Jan (harvest Feb), then Musim Luar planted May-Jun → peak Aug (harvest Sep). Two flood troughs in May + Oct align with the published cropping windows within ±1 month. Highest amplitude (0.52) of any IADA — the synchronised water management of Malaysia's premier granary visible directly in the satellite signal.

KADA kada
Parcels
574
v4 area
5,542 ha
Audit kept
94.0%
Peak
Mar + Sep
Trough
Dec
Amplitude
0.43
Jan0.28
Feb0.48
Mar0.63
Apr0.59
May0.46
Jun0.35
Jul0.32
Aug0.51
Sep0.68
Oct0.61
Nov0.34
Dec0.25

Single: 250 · Double: 83 · Unknown: 241

Type: Mixed double + single (recently shifted toward single).
Main: Aug-Feb · Off: Mar-Jul (declining area).
Source: KADA Annual Report 2024. Eastern Kelantan, drought-prone.

Two clear NDVI cycles (peaks Mar + Sep), but timing shifted 2-3 months earlier than the published Aug-Feb / Mar-Jul windows. The north-east monsoon (Nov-Feb) likely forces farmers to plant OUTSIDE the official window — Off-season Mar-Jun → Sep harvest; Main-season Oct-Dec → Mar harvest. The data reveals operational practice that diverges from the historical schedule.

IADA Kerian kerian
Parcels
1,420
v4 area
27,110 ha
Audit kept
95.7%
Peak
Jun + Dec
Trough
Mar + Sep
Amplitude
0.37
Jan0.53
Feb0.39
Mar0.28
Apr0.39
May0.51
Jun0.64
Jul0.57
Aug0.43
Sep0.30
Oct0.42
Nov0.58
Dec0.65

Single: 606 · Double: 424 · Unknown: 390

Type: Mixed single + double.
Main: Sep-Feb · Off: Mar-Aug (irrigation supplemented).
Source: IADA Kerian schedule, northern Perak design.

The cleanest 6-month rotation in the entire system — peaks Jun + Dec, troughs Mar + Sep, exactly half a year apart. Despite the v1 label tilting toward single-crop (606 vs 424), the IADA-average NDVI shows BOTH cycles active across the irrigated area. Aligns with the published Sep-Feb (main) + Mar-Aug (off) windows within ±1 month.

IADA Barat Laut Selangor bls
Parcels
570
v4 area
16,727 ha
Audit kept
91.1%
Peak
May + Nov
Trough
Aug
Amplitude
0.23
Jan0.41
Feb0.34
Mar0.45
Apr0.49
May0.55
Jun0.42
Jul0.38
Aug0.33
Sep0.43
Oct0.50
Nov0.53
Dec0.52

Single: 404 · Double: 24 · Unknown: 142

Type: Single-cropping dominant.
Main: Mar-Aug (Sungai Selangor-fed) · Off: Limited (water-stressed).
Source: Selangor agriculture department.

Lowest amplitude of any major IADA (0.23) — confirms shallow flooding. The Selangor River pumped supply doesn't drop fields to deep negative NDVI like MADA's deep-bunded paddies. Authority says single-crop dominant but data shows two cycles (May + Nov): some BLS parcels run an opportunistic off-season when water permits. Despite the modest swing, BLS has the lowest year-to-year variance of any IADA (σ 0.053) — modest signal but reliable.

IADA Pulau Pinang pp
Parcels
1,580
v4 area
11,109 ha
Audit kept
94.4%
Peak
Jul + Dec
Trough
Apr
Amplitude
0.42
Jan0.65
Feb0.44
Mar0.28
Apr0.28
May0.38
Jun0.63
Jul0.66
Aug0.54
Sep0.36
Oct0.30
Nov0.45
Dec0.70

Single: 237 · Double: 1,059 · Unknown: 284

Type: Double-cropping.
Main: Aug-Jan · Off: Feb-Jul.
Source: Pulau Pinang DOA cropping calendar. Small but intensive double-crop zone on northwestern Penang mainland.

Textbook double-cropping. Jul peak (off-season harvest) → Sep-Oct land-prep → Dec main-season peak. Aligns with the published Aug-Jan main and Feb-Jul off windows within ±1 month. 67% of PP parcels are double-cropping in v1's label — confirmed in the NDVI signature.

IADA Seberang Perak sp
Parcels
287
v4 area
6,415 ha
Audit kept
98.1%
Peak
May + Nov
Trough
Aug
Amplitude
0.30
Jan0.38
Feb0.39
Mar0.46
Apr0.53
May0.58
Jun0.53
Jul0.41
Aug0.33
Sep0.43
Oct0.54
Nov0.63
Dec0.58

Single: 106 · Double: 88 · Unknown: 93

Type: Mixed (double on better-drained, single on flood-prone).
Main: Sep-Feb · Off: Mar-Aug.
Source: IADA Seberang Perak schedule.

Two cycles: May + Nov peaks, with the August land-prep window the most synchronised feature (year-to-year σ just 0.013 — lowest single-month variance of any IADA). The cropping label split (106 single / 88 double) suggests different parcels do different cycles; the IADA average shows both. Aligns with published windows within ±1 month.

IADA Kemasin-Semerak ks
Parcels
940
v4 area
11,641 ha
Audit kept
80.6%
Peak
Mar + Sep
Trough
Dec
Amplitude
0.30
Jan0.39
Feb0.55
Mar0.63
Apr0.56
May0.48
Jun0.45
Jul0.43
Aug0.56
Sep0.67
Oct0.63
Nov0.47
Dec0.37

Single: 575 · Double: 101 · Unknown: 264

Type: Single-cropping dominant.
Main: Aug-Feb · Off: Limited.
Source: IADA Kemasin-Semerak schedule. Kelantan east coast, drought + flood risk.

Authority says single-crop dominant but the IADA-average shows two NDVI peaks (Mar + Sep). Explanation: per-parcel asynchrony — many KS parcels do single cycles but on DIFFERENT schedules (some plant Mar-Aug, others Aug-Feb), so the average shows both peaks even when individual parcels only crop once. The 80.6% audit pass rate reflects this irregularity — KS is heterogeneous, not mistracked.

IADA Sungai Manik sm
Parcels
280
v4 area
7,898 ha
Audit kept
98.5%
Peak
May + Nov
Trough
Aug
Amplitude
0.35
Jan0.37
Feb0.36
Mar0.38
Apr0.51
May0.61
Jun0.55
Jul0.41
Aug0.30
Sep0.35
Oct0.55
Nov0.65
Dec0.59

Single: 91 · Double: 91 · Unknown: 98

Type: Mixed single + double.
Main: Sep-Feb · Off: Mar-Aug (where pumped).
Source: IADA Sungai Manik (Perak central). Pumped irrigation from Bernam River.

Near-perfectly even cropping split (91 single / 91 double). Two cycles May + Nov visible. Aug land-prep window exceptionally synchronised (year-to-year σ 0.010, lowest single-month variance of any IADA) — pumped-irrigation discipline from the Bernam River source visible in the data. Aligns with the published Sep-Feb + Mar-Aug windows.

IADA Ketara ketara
Parcels
460
v4 area
4,314 ha
Audit kept
96.7%
Peak
Mar + Sep
Trough
Jan + Jul
Amplitude
0.47
Jan0.23
Feb0.49
Mar0.70
Apr0.67
May0.48
Jun0.35
Jul0.31
Aug0.45
Sep0.70
Oct0.65
Nov0.45
Dec0.28

Single: 107 · Double: 186 · Unknown: 167

Type: Double-cropping.
Main: Jul-Dec · Off: Jan-Jun.
Source: IADA Ketara (Terengganu) schedule. Newest IADA (2010s development); modernised infrastructure.

The textbook modern double-cropping IADA — two EXACTLY-EQUAL peaks (Mar 0.70, Sep 0.70), synchronised land-prep in Jan and Jul. Amplitude 0.47 is the second-highest in the system. However the schedule has shifted 1-3 months earlier than the published Jul-Dec / Jan-Jun windows — likely an operational adjustment to plant ahead of the north-east monsoon. The data reveals modern synchronised water management still operating with discipline.

IADA Rompin rompin
Parcels
247
v4 area
7,956 ha
Audit kept
40.9%
Peak
Aug
Trough
Dec
Amplitude
0.15
Jan0.43
Feb0.51
Mar0.49
Apr0.51
May0.55
Jun0.56
Jul0.55
Aug0.58
Sep0.55
Oct0.52
Nov0.50
Dec0.43

Single: 198 · Double: 11 · Unknown: 38

Type: Single-cropping.
Main: Apr-Sep · Off: None (rain-fed).
Source: IADA Rompin (Pahang). Flood-prone east coast.

NDVI is nearly flat year-round (always between 0.43 and 0.58, amplitude just 0.15) with a barely-detectable peak in August. Aligns with the published Apr-Sep main season → Aug peak biologically expected. However the persistent ground cover (weeds + rice mixed) means the NDVI never returns to a flood-low. The 40.9% audit pass rate reflects this rain-fed asynchrony — not a v4 boundary defect.

Are these calendars consistent with authority schedules?

A biophysical lag of ~3 months separates planting from peak NDVI (the heading / flowering stage). Comparing observed NDVI peaks against the expected peak window from each authority's published calendar:

IADA Authority's planting / season window Expected NDVI peak Observed NDVI peak Match
MADAMusim Utama Aug-Jan; Musim Luar Feb-JulNov-Jan + May-AugJan + Augexcellent
KerianMain Sep-Feb; Off Mar-AugDec-Mar + Jun-SepJun + Decexcellent
PPMain Aug-Jan; Off Feb-JulNov-Jan + May-JulJul + Decexcellent
RompinSingle, Apr-SepJul-SepAugexcellent
SPMain Sep-Feb; Off Mar-AugDec-Feb + Jun-AugNov + Maygood
SMMain Sep-Feb; Off Mar-AugDec-Feb + Jun-AugNov + Maygood
BLSSingle, Mar-AugJun-AugMay + Novpartial
KADAMain Aug-Feb; Off Mar-JulNov-Feb + Jun-JulMar + Septiming shift
KetaraMain Jul-Dec; Off Jan-JunOct-Dec + Apr-JunMar + Septiming shift
KSMain Aug-Feb (single-dominant)Nov-FebMar + Sep (two peaks)mismatch
Defensible finding for the thesis. Four IADAs match published calendars within ±1 month (MADA, Kerian, PP, Rompin); three match within ±1 month with caveats (SP, SM, BLS); three show systematic 1–3 month shifts (KADA, KS, Ketara). The shifts plausibly reflect monsoon-driven retreat from the Nov-Feb planting window in KADA, per-parcel asynchrony between single-cropping farmers in KS, and modern synchronised-water-release-driven schedule advance in Ketara.

Methodology in brief

Each IADA's per-month NDVI value is the mean Sentinel-2 NDVI (cloud-masked SCL) across all v4 paddy parcels in that IADA, averaged over the six calendar years 2020-2025. The year-to-year standard deviation shown in the shaded band measures cycle reliability. Phase assignment maps NDVI ranges to lowland-rice growth stages; the boundaries are heuristic but consistent across all 10 IADAs.

Full pipeline documentation: docs/v4_pipeline_methodology.md (6,563 words, 10 sections covering boundary derivation, audit, and cross-validation).
Source data: 72-month Sentinel-2 NDVI archive 2020-2025 (706 monthly rasters across 10 IADAs).
Citation: Sukiman M.A. (2026), Spatiotemporal framework for paddy vulnerability in Malaysia's Jelapang Padi zones, PhD thesis, UTM.