HDX archive and place series. Part of what the twin can answer. Every figure below is the value the route serves, with its source; the raw response is at /api/twin/brain/hdx.
Descriptive, sourced: no model
Basis: sourced rows (URL, sha256 and locator per row); the twin runs no model, while the publisher's WFP rainfall/NDVI series are remote-sensing estimates and OPHI MPI is a survey-based index.
Every dataset the archive index lists is inventoried (publisher, formats, resources listed vs held, sizes, last_modified). The served slice — WFP dekadal rainfall and NDVI by division/district pcode and OPHI MPI national and by division — is read from the manifest-pinned lake tables built from the stored files; sums, means, ranks and changes are computed in code from stored values. WFP values are served as published (unit undocumented) and labelled remote-sensing estimates, MPI a survey-based index; the twin runs no model of its own. The duplicate-series defect is handled with one primary series per pcode.
As of: 2 May 2026.
As of: 2025.
Remote-sensing estimates published by WFP (archive title 'Bangladesh: Rainfall Indicators at Subnational Level', https://data.humdata.org/dataset/4b4650bf-6faa-4659-adf5-521a8be17bf6); the stored files carry no methodology note. The twin runs no model over them.
| Division | Total, as published |
|---|---|
| Sylhet | Total, as published: 3,942 |
| Chittagong | Total, as published: 3,351.9 |
| Mymensingh | Total, as published: 2,755.1 |
| Barisal | Total, as published: 2,694.2 |
| Rangpur | Total, as published: 2,339.6 |
| Dhaka | Total, as published: 2,230.4 |
| Khulna | Total, as published: 1,968.6 |
| Rajshahi | Total, as published: 1,764.9 |
| Division | Latest dekad, as published |
|---|---|
| Sylhet | Latest dekad, as published: 163.74 |
| Mymensingh | Latest dekad, as published: 95.59 |
| Rangpur | Latest dekad, as published: 65.43 |
| Dhaka | Latest dekad, as published: 44.59 |
| Rajshahi | Latest dekad, as published: 30.21 |
| Chittagong | Latest dekad, as published: 27.02 |
| Khulna | Latest dekad, as published: 17.11 |
| Barisal | Latest dekad, as published: 14.77 |
Remote-sensing estimates published by WFP (archive title 'Bangladesh: NDVI at Subnational Level', https://data.humdata.org/dataset/1ad634b6-dc20-4b6b-940a-4037d71ef947); the stored files carry no methodology note. The twin runs no model over them.
| Division | Latest reading, as published |
|---|---|
| Mymensingh | Latest reading, as published: 0.8067 |
| Rangpur | Latest reading, as published: 0.7523 |
| Sylhet | Latest reading, as published: 0.7155 |
| Chittagong | Latest reading, as published: 0.69 |
| Barisal | Latest reading, as published: 0.6844 |
| Dhaka | Latest reading, as published: 0.6821 |
| Rajshahi | Latest reading, as published: 0.6812 |
| Khulna | Latest reading, as published: 0.6523 |
A survey-based index published by OPHI (archive title 'Bangladesh Multidimensional Poverty Index', https://data.humdata.org/dataset/3dff0e65-7aea-4eb0-9be8-af68af882911; surveys named in the stored files: DHS, MICS). The twin runs no model over it.
| Division | Index | Headcount | Intensity |
|---|---|---|---|
| Sylhet | Index: 0.1279 | Headcount: 27.79% | Intensity: 46.02% |
| Mymensingh | Index: 0.1061 | Headcount: 23.02% | Intensity: 46.11% |
| Chattogram | Index: 0.0817 | Headcount: 17.6% | Intensity: 46.42% |
| Rangpur | Index: 0.0712 | Headcount: 16.14% | Intensity: 44.09% |
| Barishal | Index: 0.0532 | Headcount: 12.38% | Intensity: 42.94% |
| Dhaka | Index: 0.0476 | Headcount: 11.19% | Intensity: 42.55% |
| Rajshahi | Index: 0.0454 | Headcount: 10.78% | Intensity: 42.09% |
| Khulna | Index: 0.0404 | Headcount: 9.4% | Intensity: 43.01% |
| Division | Index, earlier survey | Index, later survey | Fall |
|---|---|---|---|
| Chattogram | Index, earlier survey: 0.1427 | Index, later survey: 0.0687 | Fall: 51.9% |
| Sylhet | Index, earlier survey: 0.2513 | Index, later survey: 0.0988 | Fall: 60.7% |
| Rangpur | Index, earlier survey: 0.1398 | Index, later survey: 0.0491 | Fall: 64.9% |
| Dhaka | Index, earlier survey: 0.1356 | Index, later survey: 0.0463 | Fall: 65.9% |
| Khulna | Index, earlier survey: 0.0993 | Index, later survey: 0.0298 | Fall: 70% |
| Barishal | Index, earlier survey: 0.1625 | Index, later survey: 0.0396 | Fall: 75.6% |
| Rajshahi | Index, earlier survey: 0.1347 | Index, later survey: 0.0306 | Fall: 77.3% |
| Publisher | Datasets | Resources held | Bytes held |
|---|---|---|---|
| unosat | Datasets: 58 | Resources held: 55 | Bytes held: 13.3 GB |
| meta | Datasets: 11 | Resources held: 208 | Bytes held: 2.1 GB |
| hot | Datasets: 11 | Resources held: 47 | Bytes held: 1.1 GB |
| geoboundaries | Datasets: 1 | Resources held: 20 | Bytes held: 864.8 MB |
| heidelberg-institute-for-geoinformation-technology | Datasets: 4 | Resources held: 16 | Bytes held: 725.1 MB |
| ocha-roap | Datasets: 13 | Resources held: 19 | Bytes held: 401.0 MB |
| Dataset | Publisher | Place levels | Resources held | Resources listed |
|---|---|---|---|---|
| Bangladesh Multidimensional Poverty Index | Publisher: Oxford Poverty and Human Development Initiative (OPHI) | Place levels: division | Resources held: 2 | Resources listed: 2 |
| Bangladesh: NDVI at Subnational Level | Publisher: World Food Programme (WFP) | Place levels: division,district | Resources held: 2 | Resources listed: 2 |
| Bangladesh: Rainfall Indicators at Subnational Level | Publisher: World Food Programme (WFP) | Place levels: division,district | Resources held: 2 | Resources listed: 2 |
| Bangladesh - Subnational Administrative Boundaries | Publisher: OCHA Field Information Services (COD-AB) | Place levels: division,district | Resources held: 3 | Resources listed: 3 |
| Question | Answer | Computed from |
|---|---|---|
| Which division recorded the highest 2025 rainfall total, and which the lowest? | Answer: Sylhet highest (rfh sum 3942.0 as published), Rajshahi lowest (1764.9) | Computed from: hdx rainfall dekadal joined to hdx admin pcodes, primary series |
| Which divisions are poorest by MPI in the DHS 2022 snapshot? | Answer: Sylhet (0.1279), Mymensingh (0.1061), Chattogram (0.0817); national 0.0665 | Computed from: hdx mpi division, series latest snapshot |
| How far did MPI fall between DHS 2014 and DHS 2022 (trends series), and which division fell least? | Answer: national MPI 0.1452 to 0.0502; least fall Chattogram (51.9%), most Rajshahi (77.3%) | Computed from: hdx mpi division, series trends |
| How much of the indexed HDX archive is actually held on disk? | Answer: 1230 of 1264 resources (97.3%), 20.7 GB across 344 listed datasets | Computed from: hdx datasets (counts and sizes per dataset) |
| What is the rainfall and NDVI coverage of the twin's HDX slice? | Answer: dekadal 1981-01-01 to 2026-04-21 (1632 dekads) over 72 primary series; NDVI 2002-07-01 to 2026-04-01 | Computed from: hdx rainfall dekadal, hdx ndvi dekadal |
| Caveat | What | How handled |
|---|---|---|
| Duplicate units | What: PCODE BD10 (Barisal, division series) carries adm id 900396 (55 pixels), adm id 900397 (345 pixels); PCODE BD1009 (Bhola, district series) carries adm id 1006181 (55 pixels), adm id 1006187 (59 pixels). Adm id 900396's rfh record equals adm id 1006181's on every dekad of their 1,632-dekad common record. | How handled: Every stored row is kept. Aggregates use one primary series per pcode (the crosswalk's primary adm id): for an adm1 duplicate, the series whose n pixels sits at or above the 345-pixel floor the primary division series occupy; for an adm2 duplicate, the series no other unit's record equals. The other series stay readable in the lake table and are excluded from aggregates. |
| WFP units undocumented | What: The WFP snapshots publish no unit or methodology note in the stored files, so rfh and vim values and their *_avg neighbours are served as published; sums and means over them are sums and means of the published values. Rainfall (rfh) and NDVI (vim) are remote-sensing estimates published by WFP, and MPI is a survey-based index published by OPHI; the twin runs no model of its own over them. | How handled: not available |
| MPI two series | What: The two stored OPHI files disagree for DHS 2022 (national MPI 0.0665 in the latest-snapshot file against 0.0502 in the trends file; the units differ too). The stored files carry no note explaining the difference. | How handled: The two are served as two series (latest snapshot, trends), each verbatim from its own file, never mixed; the trends file publishes 7 divisions (no Mymensingh), the snapshot 8. |
| Snapshot not refresh | What: The pinned snapshot ends at dekad 2026-04-21 (rainfall) and 2026-04-01 (NDVI). The archive root also holds fresher unindexed copies (rainfall.csv, ndvi.csv) that appear in no index file; this route serves the manifest-pinned snapshot, not them. | How handled: not available |
| Held subset | What: The index lists 1,264 resources; 1,230 are present in the tree. The inventory carries both counts per dataset; absence is shown, not inferred. | How handled: not available |
| Collection | Archive dataset | Last updated |
|---|---|---|
| Rainfall | Archive dataset: bgd-rainfall-subnational | Last updated: 23 April 2026 |
| NDVI | Archive dataset: bgd-ndvi-subnational | Last updated: 23 April 2026 |
| MPI | Archive dataset: bangladesh-mpi | Last updated: 5 March 2026 |
| Pcodes | Archive dataset: cod-ab-bgd | Last updated: 26 January 2026 |
bgd_admpop_2022, adm0-adm3): Held on disk and inventoried, but outside this lane's slice; the twin serves population from its own registers and lane BRN48 serves climate and poverty place series only.ipc_bgd_area_long.csv, district phase classifications): The highest-value remaining place series in the archive; named here so the next lane starts from the file.dhs_sub.csv): Served by the dhs brain (BRN37); not repeated here.hdx_datasets; either served elsewhere by the twin (prices) or global/admin0 extracts without place keys this lane serves.snapshot_not_refresh); the served coverage fields state the snapshot's ends.