Beyond conventional engineering assessments: Why SUMP needs a measurable integration of scientific evidence, indigenous knowledge and lived river experience.
Dr M L Gaur
When the river becomes a national question
The Siang Upper Multipurpose Project (SUMP) is no longer merely a proposed hydropower and water-storage intervention. It has become a complex question involving national water security, strategic interests, state development priorities, environmental responsibility and the aspirations of communities whose lives are inseparably connected with the river.
There are legitimate arguments on different sides. India's concerns about upstream developments on the Yarlung Tsangpo deserve serious scientific examination. The proposed storage infrastructure may offer opportunities for energy generation and water regulation. At the same time, questions concerning ancestral land, displacement, ecological change and the future of the Adi communities cannot be treated as peripheral considerations.
I believe the discussion needs to move beyond asking whether a large dam is desirable or undesirable. A more meaningful question is: How do we ensure that the knowledge used to justify, design and evaluate a mega-project is as reliable on the ground as it appears in technical reports? We have the models. Do we have the complete ground truth?
Modern project preparation is supported by sophisticated hydrological models, geological investigations, satellite imagery, seismic assessments, environmental studies and economic evaluations. These are indispensable. The intention here is neither to question their necessity nor to presume that the SUMP assessment team has overlooked established procedures.
However, a technically comprehensive investigation and a sufficiently grounded investigation are not necessarily the same thing.
A satellite image may identify a land-cover class, but may not explain how a community uses that landscape through different seasons. A rainfall record may quantify precipitation, but may not capture every historically remembered flash-flood pathway. A mathematical model may simulate sediment movement, but local observations may reveal a tributary's changing behaviour, a newly active erosion zone or a drainage channel that requires closer examination.
The opportunity, therefore, is not to replace scientific knowledge with traditional knowledge. It is to make the two answerable to each other.
From G-Index wisdom to ground-tested project intelligence
In my professional and intellectual journey, particularly through my writing on G-Index Wisdom and AI Bharat and the G-index, I have explored an important distinction: the difference between what a system measures or reports and what actually works in the real world.
I call attention to the Ground Index (G-Index) as a way of thinking about knowledge through field performance, contextual suitability, adoption, ownership and sustained impact. Technology and artificial intelligence can expand our analytical capabilities, but their sophistication does not automatically establish ground-level effectiveness.
This distinction has direct relevance to large infrastructure projects.
A project may achieve excellent scores on design parameters, installed capacity, estimated investment and construction milestones. But how effectively does it perform within its actual social, ecological and hydrological setting? Are the benefits experienced where the costs are borne? Do the assumptions remain valid after the landscape changes?
For SUMP, I propose an additional assessment philosophy: Ground-to-Model Concordance.
It would systematically compare four forms of evidence:
Instrumented evidence: Rainfall, discharge, groundwater, sediment, geological samples, remote sensing and model outputs.
Inherited evidence: Historical flood memories, customary river knowledge, seasonal observations, traditional land-use practices and ecological indicators.
Observed evidence: Present-day field measurements, community monitoring, erosion locations, changes in river channels and livelihood dependence.
Decision evidence: Documented changes in site selection, design assumptions, safeguards and operational arrangements resulting from the combined findings.
The important innovation is not collecting more information indiscriminately. It is recording where different forms of evidence agree, where they disagree, why the differences exist and how the final decision responds.
The Siang may need its own knowledge architecture
Consider three alternative sites reported for preliminary investigation: Parong, Dite Dime and Ugeng. The objective should not simply be to compare their engineering suitability using conventional parameters. Each site should also have its own ground-intelligence record.
What are the historical high-water marks remembered by neighbouring villages? Which areas experience seasonal inundation? Where have bank failures or slope movements occurred? Which springs, grazing areas, agricultural patches, fishing locations and forest-use zones sustain local livelihoods? Which routes become inaccessible during extreme rainfall?
These questions should be mapped alongside engineering surveys, geological observations and hydrological models.
The Adi community's traditional institutions, including the Kebang system, offer an important foundation for structured deliberation. Their role should not be reduced to receiving presentations or signing documents. With appropriate representation and respect for customary institutions, they can contribute to the systematic documentation and interpretation of local evidence.
A village-level River Knowledge Register could record seasonal observations, historical events, land-use patterns, ecological changes and community-identified risks. Such records should be verified where possible through field measurements, historical imagery and independent scientific assessment. Sensitive cultural knowledge must remain under appropriate community control.
This is how local wisdom becomes usable evidence without being stripped of its social meaning.
From consultation to measurable participation
There is an important difference between informing people, consulting people and allowing people's knowledge to influence decisions.
For SUMP, I would suggest a practical mechanism: a Ground Evidence and Decision Ledger for every potentially affected village.
It should record five things: 1. What the community has reported. 2. What independent field investigation has verified. 3. Where scientific models support or differ from local observations. 4. What changes have been made to project assumptions or safeguards. 5. What issues remain unresolved, who is responsible and when they will be reviewed.
This creates accountability on both sides. Technical agencies must explain their conclusions, while local observations become documented, traceable and open to scientific investigation.
A useful measure would be the Ground-to-Model Concordance Rate: The proportion of identified, testable local observations that have been independently investigated and formally addressed in the relevant technical assessment. It should be reported with the number of observations, their categories and unresolved cases, rather than presented as an unexplained percentage.
Other useful indicators could include the number of village-identified risks incorporated into updated hazard maps, the proportion of community concerns receiving documented technical responses, and the number of design or mitigation decisions modified following field verification.
These would be proposed project-governance indicators, not claims about SUMP's existing performance.
What do my own 42 years of field experience taught me and probably we all?
My professional journey from 1985 to 2026 has taken me across different states, watersheds, rivers, irrigation systems, reservoirs, forests and agricultural landscapes. Working in agricultural engineering, hydrology, watershed development, soil and water conservation and natural-resource management has taught me that the same technical intervention can produce very different outcomes under different ground conditions.
A watershed treatment that performs well in one landscape may need substantial adaptation elsewhere. A soil conservation measure cannot be evaluated simply by counting structures; its effectiveness must be observed through runoff, infiltration, erosion reduction and sustained maintenance. A reservoir cannot be understood exclusively through its storage capacity; its performance is connected with its catchment, sediment supply, operating regime and downstream requirements.
(To be continued in next edition...)