# IPI 02 Vegetation Condition, The Gambia

Release **2026-09** (version 2026.09.0). Date measured: **2026-09-30**.

Published: **2026-11-12**.

First release. Any correction is published as a new release with a dated note in the corrections log.

Monthly satellite greenness and rain for The Gambia, compared with each place's own normal. The tables describe how green the land looks and how much rain fell; they are not a verdict on any place or person.

## Coverage

The Gambia, vegetated land on a 1 km grid, monthly from 2012-01 to 2026-09. Cells that are not scorable (too little vegetation under the land-cover map) are not listed. Cells and areas that were not seen clearly enough in a month carry the class "Not enough clear observations" and no value.

A month enters the record only when every 16-day satellite composite that overlaps it has arrived and a CHIRPS rain month exists, so the newest month can be weeks behind today. A released month never changes when later months arrive.

## Files

Each view is a folder with the same table as CSV and XLSX (identical rows and columns), a GeoJSON file where there are shapes (WGS84, one feature per cell or district, with an id and a name), and `data_dictionary.csv`.

| View | What it is | Shapes |
|---|---|---|
| `monthly_class_layer` | One row per scorable 1 km cell, month and indicator for the latest month and every month of the latest year. GeoJSON: latest month. | GeoJSON |
| `district_summary_monthly` | Class shares by land type and rain indicators for every district, region and the nation, by calendar month. | GeoJSON |
| `annual_summary` | Class shares, worst month, persistent cell counts and rain flag counts by calendar year (the current year from January to the latest month is marked partial_year). | none |
| `persistent_list` | Cells in a run of three or more months below normal, by calendar year: cell ID and district only. GeoJSON: latest year. | GeoJSON |
| `coverage_monthly` | Share of scorable cells that were seen clearly enough to be given a class, by region and nation and calendar month. | none |
| `satellite_agreement_check` | How closely the Suomi NPP and NOAA-20 monthly values agree over the cells both see, by calendar month and land type. | none |

`monthly_class_months/` holds one CSV per month of the monthly class layer (`monthly_class_YYYY-MM.csv`): one row per cell with its class, rain flag, withheld flag and reason and evidence grades, built from that layer, with its own `data_dictionary.csv`. The map page reads these files; the full layer above remains the complete download. `summary_for_map/` holds the small summary the page reads for its tiles, charts and district table: rows copied unchanged from the tables above, with its own `data_dictionary.csv`.

## Indicator codes

| Code | Unit | Meaning |
|---|---|---|
| `ipi02.greenness` | NDVI | Monthly maximum NDVI of a 1 km cell, from the satellite record. |
| `ipi02.greenness_z` | robust z | How far a cell-month is from the cell's own normal for the same calendar month, in robust z units. |
| `ipi02.greenness_class` | class value | The class of a cell-month: 1 Normal, 2 Below normal, 3 Well below normal. The label is in class_label. |
| `ipi02.coverage_share` | percent | Share of a cell's pixels, or of an area's scorable cells, seen clearly enough in the month. |
| `ipi02.district_class_share` | percent | Share of the scorable cells of a land type in an area that fall in a class. |
| `ipi02.worst_month_share` | percent | Highest monthly share of classed cells that are Below normal or Well below normal in the calendar year. |
| `ipi02.rain_1m` | mm | Rain total of the calendar month, mean over the cells of the area. |
| `ipi02.rain_3m` | mm | Rain total of the month and the two before it, mean over the cells of the area. |
| `ipi02.rain_3m_pct_normal` | percent | The 3-month rain total as a percent of the 2012 to 2025 mean for the same window. |
| `ipi02.spi3` | SPI-3 | Standardised precipitation index over 3 months, mean over the cells of the area. |
| `ipi02.persistence_months` | months | Months of the calendar year a cell spent inside a run of three or more months classed Below or Well below normal. |
| `ipi02.persistent_cells` | cells | Cells on the persistent list of the calendar year, by area. |
| `ipi02.rain_flag` | cells | Rain flag (grade E): in the cell layer, one row per cell classed Below or Well below normal and month, value 1, with the label in flag_label; in the annual summary, cells that carry a label in at least one month of the year. |
| `ipi02.satellite_agreement_ratio` | ratio | Median ratio of the second satellite's monthly value to the first, over the cells both see. |
| `ipi02.satellite_agreement_diff` | NDVI | Median absolute difference of the two satellites' monthly values over the cells both see. |

Every row carries `indicator_code`, `period_start`, `period_end`, `value`, `value_low`, `value_high`, `evidence_grade`, `grade_reason` and `licence`, then the area key, `value_withheld`, `withheld_reason` and the view's own columns. A blank value is never a zero.

## Licence

The data are released under **CC BY 4.0**: https://creativecommons.org/licenses/by/4.0/

## How to cite

Cubixio GMB (2026). IPI 02 Vegetation Condition, The Gambia, release 2026-09 (2026.09.0). https://data.cubixio.org/ipi-02/

## Credits

- **VNP13A1** (Collection 2 (version 002)). Vermote, E. (2023). VIIRS/NPP Vegetation Indices 16-Day L3 Global 500m SIN Grid V002 [Data set]. NASA Land Processes Distributed Active Archive Center. https://doi.org/10.5067/VIIRS/VNP13A1.002 Licence: Openly shared, without restriction, in accordance with the EOSDIS Data Use and Citation Guidance (NASA Earthdata). (https://www.earthdata.nasa.gov/earth-observation-data/data-use-policy)
- **VJ113A1** (Collection 2 (version 002)). Vermote, E. (2023). VIIRS/JPSS1 Vegetation Indices 16-Day L3 Global 500m SIN Grid V002 [Data set]. NASA Land Processes Distributed Active Archive Center. https://doi.org/10.5067/VIIRS/VJ113A1.002 Licence: Openly shared, without restriction, in accordance with the EOSDIS Data Use and Citation Guidance (NASA Earthdata). (https://www.earthdata.nasa.gov/earth-observation-data/data-use-policy)
- **CHIRPS-3.0** (v3.0 (released 2025-01-01 per README)). Climate Hazards Center Infrared Precipitation with Stations version 3 (CHIRPS3) Data Repository (2025). https://doi.org/10.15780/G2JQ0P Funk, C. et al. (2026). The Climate Hazards Center Infrared Precipitation with Stations, Version 3. Scientific Data 13, 718. https://doi.org/10.1038/s41597-026-07096-4 Licence: Public domain: the Climate Hazards Center has waived all copyright; credited as requested. (https://chc.ucsb.edu/node/497)
- **WorldCover2021** (v200 (2021 map)). (c) ESA WorldCover project 2021 / Contains modified Copernicus Sentinel data (2021) processed by ESA WorldCover consortium. DOI 10.5281/zenodo.7254221 Licence: CC BY 4.0 (Creative Commons Attribution 4.0 International) (https://creativecommons.org/licenses/by/4.0/)
- **cod-ab-gmb** (dataset last modified 2026-01-26; dataset date 2022-09-01 to 2025-10-30). Gambia - Subnational Administrative Boundaries. Source: Gambia National Disaster Management Agency (NDMA); quality assured and published by OCHA Field Information Services and HDX. Outlines simplified by Cubixio. Licence: Creative Commons Attribution 3.0 IGO (CC BY 3.0 IGO) (https://creativecommons.org/licenses/by/3.0/igo/)

## Evidence grades

- **A**: directly observed by our own instruments or coders, raw record kept (Not used in this release.)
- **B**: verified secondary, checked against a second unrelated source (Not used in this release.)
- **C**: modelled from partial data with a published model and stated assumptions
- **D**: single unverified source (Not used in this release.)
- **E**: indicative only
- **No grade**: no value is shown on the row.

Grade the evidence, not the conclusion. Where a map shape and the figures on it come from different evidence, grade them separately.

## Limitations

- Rainy-season cloud limits what the satellite sees; coverage is reported for every month.
- Dry-season greenness is low and noisy.
- Fire, harvest, irrigation and flooding change greenness for reasons of their own.
- Land-cover change shows as "below normal".
- Cells of 500 m to 1 km mix land types.
- CHIRPS rain is about 5 km and gauge-sparse.
- The satellite change (NPP to NOAA-20/21) is checked at every join; the satellite agreement check table shows how close the two are.
- The baseline is 14 years, not 25.
- A figure that rests on fewer than 3 scorable cells is withheld (the small-count rule), so region and national figures leave very small districts out.
- Every value is aggregated to 1 km cells, not farms, fields or people.
- Releases lag the calendar (the latency rule): a month enters the record only when every 16-day composite that overlaps it has arrived and a CHIRPS month exists, so the newest month can be weeks behind today. A released month never changes when later months arrive.
- Cells not seen clearly enough (clouds, haze) are shown as such. They are never counted as normal or as below normal.
- The rain flag reports rain and greenness seen together at a set lag. It is graded E (indicative) and is not a finding about what lies behind a change.
- Classes compare each 1 km cell with its own 2012 to 2025 normal for the same calendar month. They describe how green the land looks from satellite; they do not say what is behind a change. Cells not seen clearly enough are shown as "Not enough clear observations". A figure that rests on fewer than 3 cells is withheld and left blank, never shown as zero; region and national figures are built only from the districts that are shown. Months and calendar years only. Evidence grade C for greenness and rain, grade E (indicative) for the rain flag.

## How to read a change

- Cloud and coverage: rainy-season cloud hides cells. Read the coverage beside the month before reading a change in a class share.
- Fire: a burnt cell looks below normal for reasons the data cannot see.
- Harvest: a cell harvested early or late moves against its own normal.
- Irrigation: watered fields stay green outside the rains and read above normal.
- Flooding: standing water changes what the satellite sees.
- Land-cover change: a cleared or built-on cell reads as persistently below normal.
- The satellite join and change: Suomi NPP is left out from September 2026 (Method v0.1.1), weeks before it stops delivering new observations on 2 Nov 2026; NOAA-20 carries the record alone, with NOAA-21 to join when served. Months with two satellites are checked in the satellite agreement check, and the change is marked as a series break.
- CHIRPS revision: a preliminary rain month is revised when its final file arrives; preliminary months are flagged.
- Baseline length: the normal is 14 years, not 25, so one unusual year moves it more.

## Series breaks

A series break is marked in the `series_break` column of every monthly row of that month, and listed here and in `manifest.json`.

- 2018-01: Satellite join: NOAA-20 (VJ113A1) is pooled with Suomi NPP (VNP13A1) from this month
- 2026-09: Satellite change: Suomi NPP left out from this month, NOAA-20 alone (Method v0.1.1)
- 2027-01: CHIRPS version boundary: v2.0 production ends after December 2026; this record uses CHIRPS v3.0 for every month (after the end of this record)

## Satellite check

Months near a satellite change are tested for a disagreement between the satellites (Method v0.1.1, the standing check). A month the check flags is held until it is reviewed. Reviewed months:

- April 2026: flagged by the satellite check; reviewed on 9 Nov 2026 and cleared (no satellite started or stopped in April; a dry-season month, when greenness is low and small differences move squares between classes; the combined share lies between the shares from each satellite alone).
## Small counts and withholding

A figure that rests on fewer than 3 scorable cells is withheld: its value is blank, value_withheld is true and withheld_reason says why. A count of fewer than 3 cells is withheld, and one more count is withheld with it so that it cannot be worked out from the total. Region and national figures are built only from the districts that are shown, so a withheld figure cannot be found by subtraction. Only calendar months and calendar years are published; a blank is never a zero.

## Aggregation

Every value is measured on a 1 km grid. District, region and national figures are shares or counts over the scorable 1 km cells inside each area; nothing finer than a district is named, and no cell is described as a place.

## Questions, errors and links

Questions about this dataset: data@cubixio.org

Found an error? hello@cubixio.org

The record page for this dataset: https://cubixio.org/ipi/02-vegetation-condition/

The method: https://cubixio.org/ipi/02-vegetation-condition/method/

## Method

Method v0.1.1 is published at https://cubixio.org/ipi/02-vegetation-condition/method/

Method version of this build: `0.1.1`.

## Rebuild

From the course folder, with the raw data in `data/raw` unchanged:

    python run.py --steps composite pool baseline anomaly rain summarise release

- Run id: `ed94581-8a17ebb57a22`
- Commit: `ed945813565d925b396ec859955716098013c9dc`
- Config sha256: `aa28f9ca3d71e16e6cf8c9170a7ad0941a93cf4f14c7d707a90dba773acb8131`
- Inputs digest (sha256 over every input file's name and hash): `3af17dbe0b6d813ef2de046cc46bc11978c2bc432cedbb27d847f724ae2d7999`
- Input digest, boundaries: `e0f12a91944f21abb9af1f648d89b61d87bfd85924f150ef624d480e35800d55`
- Input digest, chirps: `dc4c2ebdbcbb232aa53bc5d225900feb07bc18d4c73144ddcc4fc12e2049f67c`
- Input digest, grid: `ed75b15651a37f010013d8f0072e234663b21be18466c441821c2ff3909e964f`
- Input digest, viirs: `c309b77579e06dfd2ee64f1bb8f4ff0e7679fc389bbef9471e6f67e93df7087d`
- Input digest, worldcover: `2cb5c1560c7d8e82b818ead76eba7fa53dbd1c3f099797f0574cb8741dbdfc82`

A rebuild from the same raw files, code and settings gives byte-identical files; `manifest.json` lists the hash of every file.
