Saṃsāra · public governance · witnessed

Bhandara Water Watch.

Bhandara district · eastern Maharashtra, India · the Wainganga basin

Where should the district build its next check dams, percolation tanks and gully plugs — and which of its old malguzari tanks are worth desilting first? We answer it from the terrain itself: every candidate is sited on the real drainage network, screened against slope, land use and catchment, and ranked by where storage is genuinely missing.

Copernicus GLO-30ESA WorldCover 10 mWhiteboxTools · D8 / StrahlergeoBoundaries ADM230 m native

District area: 4,222.8 km²
Elevation: 206–535.1 m
Dominant soil: HSG-D
Runoff (80 mm): CN 89 · 51 mm
Candidate sites: 57

For the administration

Executive summary.

The whole study in one view — what was found, what it would deliver, and what to do next. Every figure below is read from the witnessed data on this page and is reproducible from scripts/water/.

· What this is. A research-grade, satellite-witnessed siting study for Bhandara’s water conservation — where to build new check dams, percolation tanks and gully plugs, and which old malguzari tanks to desilt first. Every one of the 57 candidates sits on the real 30 m drainage network, not a wish-list.
· A decision-ready programme. From ~163,000 stream cells, 57 candidates were sited and ranked into a 11-site shortlist (corroborated, low-displacement), 14 worth-surveying (one or two resolvable blockers each), and 32 held back (dense settlement or not yet corroborated).
· The prize, counted incrementally. Built in full, the 25-site programme adds ≈28.72 MCM of net-new water a year — ≈5,743 ha of protective irrigation and ≈522,111 drinking person-years across 24 villages — for 414 ha of new land (renovations acquire none). The shortlist alone delivers ≈21.19 MCM net-new.
· Every number is witnessed. A 48-month, four-monsoon dual-sensor record — Sentinel-2 optical and Sentinel-1 radar (calibrated σ⁰) — cross-checks each fill and draw-down; HydroLAKES confirms real tanks; benefits are discounted to the net-new gain over each site’s own historical baseline, not gross capacity.
· Honest by construction. Every site carries a confidence grade; new structures are capped at “medium” (no ground truth a new reservoir will hold); a full 3 m bund is stepped down to 2 m where it would flood a settlement; tank renovations displace no one; low-confidence sites must be ground-truthed before any funds are committed.
· What it is not. A 30 m regional screen, not a Detailed Project Report — it precedes, never replaces, cadastral (7/12) verification, geotechnical bores and level surveys.
· Recommended next step. Sanction the 11-site shortlist to cadastral and geotechnical survey now; commission ground reconnaissance on the 14 worth-surveying sites to resolve their single blocker — land to verify, a pour-point / through-flow check, or independent corroboration — and promote or drop each.

What it delivers

The impact, quantified.

Indicative district benefit if these interventions are built — the corroborated shortlist of 11, the full programme of 25 (shortlist + worth‑surveying), and the theoretical upper bound of all 57 candidates. Under standard norms; actual yield varies with the monsoon.

Shortlist · 11 sites

Water impounded

24.22 MCM

Irrigation

4,843 ha

Drinking (person-yrs)

440,285

Recharge

3,180 TCM

Villages served

Land to acquire

169 ha

≈ 21.19 MCM net-new · 4,237 ha · 385,218 drinking p-yrs · incremental over today

Full programme · 25 sites

Water impounded

33.35 MCM

Irrigation

6,670 ha

Drinking (person-yrs)

606,328

Recharge

4,501 TCM

Villages served

Land to acquire

414 ha

≈ 28.72 MCM net-new · 5,743 ha · 522,111 drinking p-yrs · incremental over today

All candidates · 57

Water impounded

56.66 MCM

Irrigation

11,331 ha

Drinking (person-yrs)

1,030,102

Recharge

11,202 TCM

Villages served

Land to acquire

2,762 ha

≈ 52.02 MCM net-new · 10,405 ha · 945,910 drinking p-yrs · incremental over today

Indicative benefits under stated water-resource norms. Actual yield varies with monsoon reliability, seepage and siltation; not a DPR-grade hydrology. Recharge is limited on HSG-D clay and is not overclaimed. Land-to-acquire (submergence) counts only NEW structures at their 3 m design footprint — tank renovations acquire no new land (the water body already exists).

How it works · terrain to sanction

From bare terrain to a fundable programme.

No wish-lists, no round numbers. The district is read from its own elevation, watched from orbit across four monsoons, screened against every mapped building — and only then ranked. Every step is a witnessed artefact, reproducible end-to-end from scripts/water/.

The task

Find where Bhandara should build and desilt next — sited on the real drainage network, proven by satellite, honest about who it moves — and quantify the net-new water each intervention buys.

Terrain & drainage

Copernicus GLO-30 hillshade · Strahler-ordered nalas · the sited structures

01 / 05

01GLO-30 · WhiteboxTools
Read the terrain
A 30 m elevation model, depression-breached, becomes the district’s full drainage network — every nala ordered 1 to 8, the Wainganga at the top.
02AHP · 7 criteria
Site & score
Slope, soil, land use and catchment gate what can be built where; a Saaty-weighted score ranks 57 candidates on the real network — nothing off-channel.
03Sentinel-2 + Sentinel-1
Witness from orbit
Four monsoons of optical and radar water extent cross-witness every fill and draw-down — radar sees through the cloud that blinds optical exactly when tanks fill.
04Overture · 415,175 buildings
Screen for people
Each design footprint is checked against real buildings; bunds step down to clear settlements, and tank renovations displace no one.
05tiers · delta impact
Rank & deliver
Sites resolve into a 11-site shortlist, a worth-surveying tier and the rest — each with the net-new water it would add, counted against its own history.

The evidence

The district, read from terrain up.

A 30 m elevation model, hydrologically conditioned, becomes a full drainage network; land cover and slope gate what can be built where; and a priority score surfaces the sites that fill the real gaps in Bhandara’s tank country. Switch layers, filter by structure, read the coordinates.

Terrain & drainage

Copernicus GLO-30 hillshade · Strahler-ordered stream network · sited structures

Investment shortlist — corroborated & low-displacement

Sites where two independent sensors agree on the water signal (or an inventory lake confirms it) and the design footprint clears existing settlement. These lead; the full roster and its caveats follow below.

Tank renovationtank-5high

near Dhamlewada

607 ha peak65 km² catchclear displacement

restores ≈ 9099 TCM · 1820 ha irrigable · Δ 8645 net-new

Check damcd-2medium

near Tekadi

83 priority102 km² catchclear displacement

impounds ≈ 235 TCM · 47 ha irrigable

Percolationpt-10medium

near Opara

65 priority1.5 km² catchlow displacement

impounds ≈ 94 TCM · 19 ha irrigable

Percolationpt-14medium

near Dongargaon

63 priority7.5 km² catchclear displacement

impounds ≈ 1096 TCM · 219 ha irrigable

Percolationpt-15medium

near Pimpalgaon

63 priority9.8 km² catchlow displacement

impounds ≈ 248 TCM · 50 ha irrigable

Tank renovationtank-9medium

near Gadkumbhali

98 ha peak658 km² catchclear displacement

restores ≈ 1469 TCM · 294 ha irrigable · Δ 804 net-new

Tank renovationtank-10medium

near Wadad

83 ha peak466 km² catchclear displacement

restores ≈ 1241 TCM · 248 ha irrigable

Tank renovationtank-6medium

near Umari

140 ha peak453 km² catchclear displacement

restores ≈ 2101 TCM · 420 ha irrigable · Δ 1259 net-new

Tank renovationtank-2medium

near Umari

142 ha peak452 km² catchclear displacement

restores ≈ 2123 TCM · 425 ha irrigable · Δ 1224 net-new

Tank renovationtank-4medium

near Kitadi R

280 ha peak105 km² catchclear displacement

restores ≈ 4196 TCM · 839 ha irrigable

Tank renovationtank-12medium

near Kholapur

154 ha peak32 km² catchclear displacement

restores ≈ 2314 TCM · 463 ha irrigable · Δ 2152 net-new

Worth surveying — promising, a blocker or two to resolve

The next tier for the ground team: sites with real water evidence held back by one or two resolvable questions — chiefly land to verify (moderate displacement), a pour-point / through-flow check on the larger nalas, or a signal not yet corroborated by an independent lake. Each still carries its measurable payoff if built. Survey these to promote or drop them.

Tank renovationtank-1needs corroboration

near Khambata

173 ha peaklow confidenceclear displacement

if built · ≈ 2590 TCM · 518 ha irrigable · Δ 2198 net-new

Tank renovationtank-7needs corroboration

near Jamni

116 ha peaklow confidenceclear displacement

if built · ≈ 1744 TCM · 349 ha irrigable · Δ 1213 net-new

Check damcd-15verify land

near Wahani

90 km² catchmedium confidencemoderate displacement

if built · ≈ 188 TCM · 38 ha irrigable

Tank renovationtank-11needs corroboration

near Mahalgaon

82 ha peaklow confidenceclear displacement

if built · ≈ 1235 TCM · 247 ha irrigable · Δ 715 net-new

Percolationpt-3verify land

near Hingana

7.9 km² catchmedium confidencemoderate displacement

if built · ≈ 158 TCM · 32 ha irrigable

Check damcd-1verify land

near Kinhi

133 km² catchmedium confidencemoderate displacement

if built · ≈ 302 TCM · 60 ha irrigable

Percolationpt-5verify land

near Mangli

5.6 km² catchmedium confidencemoderate displacement

if built · ≈ 598 TCM · 120 ha irrigable

Tank renovationtank-8needs corroboration

near Dabha

55 ha peaklow confidenceclear displacement

if built · ≈ 823 TCM · 165 ha irrigable · Δ 665 net-new

Tank renovationtank-3needs corroboration

near Khamari Bk

44 ha peaklow confidenceclear displacement

if built · ≈ 667 TCM · 134 ha irrigable

Check damcd-14verify land

near Silegaon

116 km² catchmedium confidencemoderate displacement

if built · ≈ 158 TCM · 32 ha irrigable

Percolationpt-12verify land

near Panjara

7.4 km² catchmedium confidencemoderate displacement

if built · ≈ 225 TCM · 45 ha irrigable

Check damcd-3verify land

near Chicholi

69 km² catchmedium confidencemoderate displacement

if built · ≈ 158 TCM · 32 ha irrigable

Check damcd-11verify land

near Kolari

34 km² catchmedium confidencemoderate displacement

if built · ≈ 142 TCM · 29 ha irrigable

Check damcd-12verify land

near Panjarepar

48 km² catchmedium confidencemoderate displacement

if built · ≈ 143 TCM · 29 ha irrigable

Tank desilting / strengthening · 12

Satellite-observed water body

Renovate an existing malguzari tank — desilt to restore lost storage.

Site	Lat, Lon	Water-body area	Catchment km²	Peak water	Confidence	Displacement
tank-5near Dhamlewada	21.5204, 79.8014	284 ha	65	607 ha	high	clear
tank-9near Gadkumbhali	21.0500, 80.0021	23 ha	658	98 ha	medium	clear
tank-10near Wadad	21.0334, 80.0165	8 ha	466	83 ha	medium	clear
tank-6near Umari	21.0424, 80.0289	7 ha	453	140 ha	medium	clear
tank-2near Umari	21.0408, 80.0345	5 ha	452	142 ha	medium	clear
tank-4near Kitadi R	20.9975, 79.6692	2 ha	105	280 ha	medium	clear
tank-12near Kholapur	21.0462, 79.6407	5 ha	32	154 ha	medium	clear
tank-7near Jamni	21.2035, 79.6731	17 ha	757	116 ha	low	clear
tank-1near Khambata	21.2144, 79.6752	2 ha	754	173 ha	low	clear
tank-8near Dabha	21.2246, 79.6721	8 ha	753	55 ha	low	clear
tank-11near Mahalgaon	21.0351, 80.0374	2 ha	439	82 ha	low	clear
tank-3near Khamari Bk	21.2660, 79.5963	3 ha	349	44 ha	low	clear

Check dam / nala bund · 15

Sensor-witnessed nala flow

New bund across a 2nd–4th-order nala to hold monsoon flow and recharge upstream.

Site	Lat, Lon	Priority	Strahler	Slope %	Catchment km²	Peak water	Confidence	Displacement
cd-1near Kinhi	20.8237, 79.8038	84	4	2.95	133	69 ha	medium	moderate
cd-3near Chicholi	21.3645, 79.6123	83	4	2.96	69	13 ha	medium	moderate
cd-11near Kolari	20.7035, 79.5519	83	3	2.63	34	6 ha	medium	moderate
cd-2near Tekadi	20.7353, 79.5276	83	4	2.27	102	4 ha	medium	clear
cd-12near Panjarepar	20.7316, 79.5424	83	4	2.85	48	2 ha	medium	moderate
cd-13near Irali Bk.	20.7706, 79.7722	82	4	2.50	42	150 ha	medium	high
cd-9near Lawadi	20.7922, 79.7337	81	3	2.50	19	66 ha	medium	high
cd-8near Mohadi	21.3046, 79.6785	81	4	2.92	68	26 ha	medium	high
cd-5near Dhamani	20.7788, 79.7537	81	4	2.43	33	143 ha	medium	high
cd-7near Asgaon	20.8079, 79.7258	81	3	2.60	14	92 ha	medium	high
cd-14near Silegaon	21.4343, 79.8481	80	4	1.55	116	27 ha	medium	moderate
cd-15near Wahani	21.4528, 79.8613	80	4	2.08	90	114 ha	medium	moderate
cd-4near Mendhegaon	20.7173, 79.5474	83	3	2.87	37	7 ha	low	moderate
cd-10near Bhuyar	20.6991, 79.5660	80	3	1.90	31	11 ha	low	moderate
cd-6near Ghorpad	21.3484, 79.5314	80	3	2.16	24	3 ha	low	moderate

Percolation tank · 15

Sensor-witnessed nala flow

New pond on a 1st–2nd-order stream to pond runoff and recharge the aquifer.

Site	Lat, Lon	Priority	Strahler	Slope %	Catchment km²	Peak water	Confidence	Displacement
pt-9near Dhamani	20.7721, 79.7590	65	2	0.54	5.69	124 ha	medium	high
pt-4near Gawarala	20.7051, 79.8350	65	2	2.19	7.08	69 ha	medium	high
pt-10near Opara	20.7716, 79.8028	65	1	1.49	1.54	94 ha	medium	low
pt-12near Panjara	21.2509, 79.7877	65	2	1.47	7.41	20 ha	medium	moderate
pt-3near Hingana	21.4065, 79.7094	64	2	1.49	7.94	79 ha	medium	moderate
pt-13near Khairi	20.8413, 79.8019	64	2	0.35	2.89	187 ha	medium	high
pt-14near Dongargaon	20.9542, 79.9629	63	2	0.32	7.46	43 ha	medium	clear
pt-15near Pimpalgaon	21.3376, 79.6319	63	2	1.73	9.84	28 ha	medium	low
pt-5near Mangli	21.3614, 79.7385	62	2	1.10	5.64	59 ha	medium	moderate
pt-7near Doke Sarandi	20.7546, 79.8170	66	2	1.89	3.32	130 ha	low	moderate
pt-6near Kudegaon	20.7346, 79.8278	66	2	0.22	3.39	205 ha	low	high
pt-8near Bhojapur	21.1590, 79.6353	66	2	1.60	4.46	15 ha	low	high
pt-11near Dholsar	20.8086, 79.7754	65	2	2.20	3.86	71 ha	low	high
pt-1near Manded	20.7185, 79.8432	64	2	0.35	3.29	148 ha	low	high
pt-2near Keshori Kh.	20.6893, 79.9362	64	2	1.33	7.98	46 ha	low	moderate

Gully plug / gabion · 15

Terrain-modelled (sub-pixel)

Small check in a headwater ravine to slow erosion and seed recharge — sub-pixel for satellite water.

Site	Lat, Lon	Priority	Strahler	Slope %	Catchment km²	Peak water	Confidence	Displacement
gp-8near Khaksi	20.8273, 79.6000	80	1	5.97	1.05	2 ha	low	clear
gp-1near Bundalghat	21.2281, 79.8572	79	1	3.97	0.59	0 ha	low	clear
gp-4near Pangadi	21.2309, 79.9529	79	1	3.78	0.71	0 ha	low	clear
gp-2near Ambagad	21.4572, 79.6522	78	1	3.84	0.91	2 ha	low	clear
gp-5	21.2355, 79.9138	78	1	4.23	0.59	0 ha	low	clear
gp-9near Pangadi	21.2456, 79.9594	78	1	3.99	0.27	0 ha	low	clear
gp-10near Gidalpar	21.2277, 80.0035	77	1	4.36	0.31	0 ha	low	clear
gp-11	21.2470, 79.9250	77	1	3.97	0.63	0 ha	low	clear
gp-6near Pangadi	21.2240, 79.9735	77	1	3.99	0.37	0 ha	low	clear
gp-12	21.4440, 79.4969	77	1	5.10	0.40	0 ha	low	clear
gp-7near Pitezari	21.2077, 80.0216	77	1	3.71	1.04	0 ha	low	clear
gp-13near Alesur	20.9682, 79.7755	77	1	4.36	0.69	0 ha	low	clear
gp-14near Gidalpar	21.2138, 79.9969	75	1	4.53	0.34	0 ha	low	clear
gp-15near Pangadi	21.2169, 79.9611	75	1	4.42	0.74	0 ha	low	clear
gp-3near Mandekasa	21.4524, 79.5281	75	1	4.06	0.45	2 ha	low	clear

Method

How it was derived.

Substrate

Copernicus GLO-30 elevation and ESA WorldCover 10 m land cover, plus ISRIC SoilGrids clay/sand (0–30 cm) → Hydrologic Soil Group. Cross-checked against the witnessed MH36 cube for Bhandara (3,631 cells). Reprojected to UTM 44N at 30 m.

Terrain & hydrology

WhiteboxTools: least-cost depression breaching → D8 flow accumulation → Strahler order (1–8, order-8 = the Wainganga), slope, wetness index and drainage density; plus DEM-derived lineament (fracture) density and an SCS Curve-Number runoff surface.

Decision (AHP)

Seven witnessed criteria — runoff, slope, soil, drainage density, lineaments, land use, wetness — weighted by a Saaty pairwise matrix. Because the district is HSG-D clay, two surfaces are built: storage on the runoff-rich lowlands, recharge only where soil is lighter or fractured.

Seasonal witness (radar + optical)

For every site, a four-year (48-month) monthly time series across four monsoons: Sentinel-2 MNDWI (optical) and Sentinel-1 GRD radiometrically calibrated to σ⁰ (DN²/sigmaNought² from the product calibration LUT, GCP thin-plate-spline geocoded), open water at σ⁰ < −17 dB. Four monsoons capture inter-annual variability — how a tank behaves in a deficit year, not just one snapshot. Radar sees through the monsoon cloud that blinds optical exactly when tanks fill — the two sensors cross-witness each fill and draw-down.

AHP criteria weights

Consistency ratio 0.003 (< 0.10 = consistent)

runoff

0.21

slope

0.22

soil

0.21

draindens

0.12

lineament

0.07

lulc

0.12

twi

0.06

What this is — and is not

· Every site is cross-witnessed — optical (S2 MNDWI) vs radiometrically calibrated radar (S1 σ⁰, water at < −17 dB) water agreement, and an independent HydroLAKES check. Each carries a confidence grade; new structures are capped at medium (no ground truth a new reservoir will hold) and low-confidence sites must be ground-truthed before any funds are committed.
· A 30 m regional screening that narrows ~163,000 stream cells to a shortlist — not an engineering survey. Every new structure is checked for displacement against real building footprints (Overture: OSM + Google + Microsoft, 415,175 in the district) inside its design-bund submergence — stepped down from 3 m to 2 m where a full bund would flood a settlement; tank renovations acquire no new land. This precedes — not replaces — a cadastral (7/12) overlay and ground survey before sanction.
· Reconnaissance context is from map APIs — Ola Maps + Google Places give the nearest habitation, landmark and benefit amenities (schools/health) per site, a second independent witness to displacement. They do not provide cadastral land records, geotechnical bores or level surveys — those remain a statutory ground task.
· Groundwater decline is not yet fused — the WRIS telemetric feed has no Maharashtra coverage, so we did not fabricate a recharge-need layer; CGWB seasonal (pre/post-monsoon) levels are the next join.
· Lineaments are DEM-derived structural proxies (automated Hough extraction), indicative of fracture zones — not a surveyed GSI fracture map.
· The radar record is single-satellite before Dec 2024 — Sentinel-1B was lost in Dec 2021 and S1C only came online Dec 2024, so most of the 48-month window is S1A alone (12-day repeat). Monthly compositing absorbs this, but the earlier years have fewer candidate radar passes than the optical record (Sentinel-2, 2017–).
· Boundary is geoBoundaries ADM2 (4,223 km²); the LGD-2011 record is 3,717 km² — a boundary-vintage delta, not an error.

Provenance

DEM: Copernicus GLO-30 (ESA) tiles N20/N21 x E079/E080
Soil: ISRIC SoilGrids v2.0 clay & sand 0-30 cm (/vsicurl)
Land cover: ESA WorldCover 10 m v200 (2021)
Hydrology: WhiteboxTools: breach_depressions_least_cost, d8_flow_accumulation, strahler_stream_order, slope, wetness_index
Lineaments: DEM multi-azimuth hillshade -> Canny -> probabilistic Hough (skimage)
Boundary: geoBoundaries ADM2 Bhandara (4,223 km2; LGD-2011 = 3,717 km2, boundary-vintage delta)
Cube cross-check: s3://gg-ananda-prithvi-pristine/derived/bharata_strata/sprint-6/mh36-unified-v1/admin_code=471

DEM tile fingerprints (SHA-256 · 16)

N20_00_E079.tif29db2e4f070afc0f
N20_00_E080.tif53d7465f3a735ce3
N21_00_E079.tif3c16332fc0ad8904
N21_00_E080.tif54aa601652335ffb

Generated 2026-07-02 · reproducible from scripts/water/

More of the witnessed substrate