Explore state-level paddy performance

Where is yield higher?

Yield in kg/ha. One fixed colour scale across all years and states in the CSV supports comparison. Grey means excluded or unavailable. Regional panels are enlarged separately and are not at the same map scale.

Hover over a state for yield, area and production. State averages are not paddy-field locations. Boundaries: geoBoundaries / OpenStreetMap contributors; ODbL 1.0 for the inspected cached Malaysia layer.

Distributions and unusual observations

Boxes show Q1–Q3; middle line is the median; whiskers extend to the furthest observation within 1.5 × IQR. All observations are shown. Outliers are descriptive flags, not errors. Mean and sample SD are reported below.

Relationships across variables

Inspect a relationship

One point per state-year. Colour identifies state. Hover for year and values; click legend entries to isolate states.

Units and definitions

Perangkaan Agromakanan Malaysia 2024, Table 2.1.2, printed page 13 (PDF page 27): planted area ha, production mt. Table 2.1.13, printed page 24 (PDF page 38): yield kg/ha. Includes wetland and dryland paddy, main and off-seasons.

Yield here is production (mt) × 1,000 / planted area (ha). It is calculated from the CSV, not transcribed from the rounded published yield table. The annual area is cropped area across seasons, not necessarily unique land area.

GEE definitions follow the supplied harvester and metadata: state-wide means, without a paddy mask. “Median_NDVI” is a legacy name for the spatial mean of annual per-pixel median NDVI. Soil variables are static 0–5 cm estimates. Repeating them across years does not provide independent soil observations.

Correlations are exploratory, not causal. Repeated observations of states are dependent. Removing state means does not remove shared year effects. Area and production are mathematically related to calculated yield. No statistical significance is claimed.

Malaysia Paddy Resilience Digital Twin

Explore observed state-level patterns through time. This is a historical, exploratory twin, not a live monitoring system or a predictive crop model. This page uses all available states and its own year controls; the EDA filters above apply to the other pages.

Click a state, or focus it and press Enter, to inspect its twin. Grey means unavailable, never zero. Colour scales are fixed across 2017–2024 for each layer. Regional panels use different map scales. Boundaries reuse the existing dashboard BOUNDARIES layer and attribution.

State twin

How to interpret this twin

Each baseline is the state's arithmetic mean of available annual observations in 2017–2024, including the displayed year. At least three valid years are required. Baselines are fixed when the replay year changes; they are retrospective reference values, not pre-event counterfactuals. Missing observations are not imputed.

Anomaly = observed value − state baseline. Percentage anomaly = 100 × anomaly / baseline (undefined for a zero baseline). Standardised anomaly = anomaly / sample standard deviation (undefined when there is no variation). Baseline sample counts are shown for every metric.

Exploratory resilience index = 100 × observed yield / the state's mean yield. 100 means the state matches its own baseline. A higher value indicates relatively higher observed yield only. This unvalidated proxy does not measure shock resistance, recovery, adaptation or causal climate effects; it is not an overall resilience ranking. No climate thresholds or arbitrary climate weights are used.

National pulse uses available states: yield = total production × 1,000 / total planted area. Its baseline comparison weights those same states' mean yields by their current-year planted area, requiring a valid baseline for every reporting state. Changes in coverage or cropped area can change the aggregate. It is not a verified national total when coverage is incomplete.

Non-causal and descriptive: temperature, rainfall and NDVI are state-wide environmental summaries, not paddy-field exposure measurements. State annual averages hide local conditions and crop seasons. Climate and yield co-movement does not establish cause; irrigation, management, varieties and other factors are not controlled. Eight annual observations cannot validate resilience, forecasts or policy effects. There is no simulated intervention or prediction.