Hack's law¶
An empirical power-law fit between a drainage network's longest upstream flow-path length and its contributing area over a stated measurement range.
Core Idea¶
Hack's law is an empirical length–area relation for a specified drainage network: L = C A^h, where A is area contributing flow to a chosen outlet or point, L is the longest upstream channel or flow-path length under a declared measurement rule, C is a unit-dependent coefficient, and h is an exponent fitted across a population and scale range. John T. Hack's 1957 Virginia and Maryland measurements gave L=1.4 A^0.6 in the report's miles and square-mile convention. The equation describes that measured region; neither 1.4 nor 0.6 is a necessary constant of every network.[1]
The same length–area question can be asked of a computed overland-flow network without visible channels. Cheraghi and colleagues tested it on a laboratory sediment surface under rainfall using topographic scans and a D8-derived network. That unlike case supports a bounded Hack-type relation across drainage representations, not a claim that a lab path is a natural stream channel. Deviations and finite range belong in the claim rather than being hidden after the fit.[2][3]
Structural Signature¶
- Network and units: select a natural tributary basin or a declared derived drainage/flow network, and identify the network points or subbasins compared. Arbitrary paired size and distance measurements are not Hack's geomorphic law.[1][2]
- Contributing area A: measure the surface whose drainage reaches each selected point. In Hack's original stream setting, A includes tributaries above the locality; a delta's nourishment area would be a different quantity and requires explicit analogy rather than substitution.[1][4]
- Longest path L: measure along the main channel and bends to the divide at the head of the longest upstream stream, or give the corresponding longest routed flow path in a computed network. Total channel length, straight-line basin diameter, and L are different measurements.[1][2]
- Fitted relation and limits: estimate C and h for a specified scale range, units, and method; inspect residuals and possible breaks. One basin cannot establish an exponent. A fitted regional regularity need not persist at all scales.[1][3]
The field and laboratory cases share those roles while differing in whether channels are observed or inferred by routing surface flow.
What It Is Not¶
The name is not a universal equation with h exactly 0.6. Hack reported an approximately 0.6 exponent in his setting, while Cheraghi's different experiment estimated a finite-range h of roughly 0.54–0.60 by a conditional-length-moment method. Those estimates do not show an invariant coefficient, exponent, or basin shape across environments. A higher exponent in one fitted range also does not entitle us to extrapolate a monotonic shape trend to every larger basin.[1][2][3]
It is not the Stream Power Law. That separate model links river-bed incision rate to drainage area and channel slope; Hack's L–A relation concerns geometric length, not an erosion-rate output. It is not any power-law scatterplot whose variables happen to be a length and an area.[1][5]
A recent river-delta study calls a distributary-length versus nourishment-area relation apparent Hack's law and reports local scale breaks. Nourishment area describes where a distributary supplies land-building sediment, not the tributary catchment in Hack's field report. This is a useful qualified analogue, not a third unexamined proof of one identical exponent.[4]
Scope of Application¶
The relation is useful for comparing drainage-network geometry where a defensible rule pairs contributing area with the longest upstream path over many units. It can summarize natural tributary basins and, with an explicit routing algorithm, derived overland-flow networks. State the dataset, map or scan resolution, network extraction, length convention, units, fitting method, and area/length range when interpreting h and C.[1][2]
The laboratory application is especially bounded. Cheraghi's 2 m × 1 m unconsolidated-sediment surface received nonuniform rainfall for 16 hours. No visible rills formed. The researchers scanned topography, calculated D8 overland-flow paths, and analyzed a computed network. The reported scaling is evidence about that represented flow morphology, not measurement of a natural channel or a guarantee about every hillslope.[2]
Clarity¶
A power law becomes a straight line after taking logarithms: log L = log C + h log A. The slope h belongs to the fitting method and sampled interval. C has units of length divided by area to the h power, so Hack's numerical 1.4 cannot be carried to another unit system unchanged. The expression L∝A^h summarizes a trend, not an exact equality for every point.[1][3]
If h exceeds ½ in a given comparison, L/√A scales upward with A in that fitted range. That mathematical comparison helps explain why Hack scaling is discussed with basin elongation, but measured channel sinuosity and basin geometry also matter. Rigon and colleagues investigate the exponent–elongation connection; the ratio alone is not a universal shape theorem.[6]
Manages Complexity¶
The law compresses many network measurements into a length–area relation and a fitted exponent. It lets one compare field basins, computed networks, or fitting ranges without tracking every channel bend in the final summary. The compression remains accountable to the definition of A and L and to residual variation. Dodds and Rothman show that fluctuations and deviations constrain the range of approximate Hack scaling; simply raising network resolution does not eliminate that issue.[1][3]
The laboratory case also illustrates a controlled transfer. The surface lacks visible rills, yet a D8-derived network yields comparable statistics over a finite interval. What transfers is the length–contributing-area measurement relation, not a claim that channel formation, sediment transport, and field-basin morphology are physically identical.[2]
Abstract Reasoning¶
To use Hack's law, first define the drainage or routed-flow network and the population of outlets or network points. For each unit, compute the upstream contributing area A and the longest upstream path L under the same rule. Then choose a scale interval, fit an exponent and coefficient using a method appropriate to the data, and inspect deviations or scale breaks. Only after that compare the fitted law with another network whose A and L have matched meanings.[1][2][3]
Cheraghi's h≈0.54–0.60 is not an ordinary single regression of each observed L against A. Their Figure 4 estimates h from slopes of ratios of successive conditional upstream-length moments, with a finite-size conditional distribution and roughly two decades of scaling. Describing the result merely as an exact pointwise L=C A^h for all D8 paths would misstate both the statistic and its range.[2]
Knowledge Transfer¶
Natural streams and the unchanneled laboratory surface pose the same structural question: how does the longest path feeding a network point vary with the area feeding it? Hack's field study measures mapped stream channels and basins. Cheraghi's experiment obtains paths from a scanned surface and a flow-routing algorithm. In both, measurements are paired across scales, but their coefficients, exponents, and physical carriers must be interpreted within each method.[1][2]
A distributary delta permits a further analogy if nourishment area is defined as the sediment-supplied area. Dong and colleagues report globally similar apparent scaling but distinct local patterns and scale breaks. That finding transfers a measurement question; it does not transform nourishment area into upstream tributary drainage area.[4]
Examples¶
Hack's Virginia and Maryland streams. For field-measured stream localities, A is upstream drainage area including tributaries and L follows the channel bends to the divide at the head of the longest upstream stream. Hack plotted the paired values and reported L=1.4 A^0.6 in his units. The example instantiates network, contributing area, main path, and empirical power fit; its geography and measurement convention limit the numeric parameters.[1]
Unchanneled laboratory runoff surface. On a 2 m × 1 m sediment surface under 16 hours of nonuniform rainfall, Cheraghi and colleagues found no visible rills. They laser-scanned the evolving morphology, used D8 routing to define upstream flow paths and contributing areas, and measured Hack-type statistical scaling over a finite interval. Figure 4's h≈0.54–0.60 comes from conditional-length-moment ratios. The mapped roles match Hack's field relation, while the network is computationally inferred overland flow rather than a natural channel system.[2]
Structural Tensions¶
Aggregate regularity versus local and scale-specific difference. A single L–A exponent makes network comparisons possible, but it can obscure deviations, uncertain path extraction, and a limited fitting interval. Higher resolution does not automatically extend the scaling range. Diagnostic: Which points and scales fitted h, how were longest paths extracted, and where do residuals or breaks appear? A meaningful law reports both the pattern and the region in which it holds.[3][2]
Structural–Framed Character¶
Formal structure: the measurable relation is a power-law association between contributing area and longest path. Evaluative weight: a fitted exponent is descriptive, not a judgment that one drainage form is better. Human-practice dependence: researchers choose basin boundaries, routing algorithms, units, and fit ranges; these choices change the reported numbers. Institutional origin: Hack's name records a historical field study, while a later laboratory analysis must still meet the same variable definitions.[1][2]
Vocabulary travel: “area” and “length” travel widely, but “contributing drainage area” and “longest upstream flow path” anchor this law in geomorphology. Import versus recognition: recognize Hack-type scaling by paired drainage-network measurements across a stated range; calling any size–length trend “Hack's law” imports the name without its carrier. Its character: structural-dominant but empirically framed. The equation is reusable within network measurements, while coefficients, extraction methods, and validity intervals depend on observations.[1][2][3]
Structural Core vs. Domain Accent¶
The portable skeleton is a power-law relation fitted across scale, a theme related to the live Allometry and Scaling Law. That Prime's present full identity, however, asserts recurring or universal exponents across diverse substrates. Hack's context-fitted h does not establish that stronger condition. The live Scaling and Scale Dependence additionally concerns qualitative mechanism shifts with scale, which this L–A fit does not assert. No strict parent edge has passed all-instance review, so Hack's law is an approved unparented domain-specific root rather than a forced child of either Prime.[1][2][3]
The domain-bound mechanism is pairing upstream contributing area with the longest path in a drainage or routed-flow network and estimating their empirical power relationship. A general cross-domain Prime about measurement-range-dependent scaling would require independent unlike-domain evidence; the field and lab cases here remain within geomorphic flow-network representations. The river-delta analogue changes what “area” means and is therefore qualified separately.[4]
Instantiates / Related Primes¶
The approved DAG placement is a zero-edge root. Allometry and Scaling Law supplies a related power-law idea but its present recurrence/universality requirement is not proven for these context-specific fits. Scaling and Scale Dependence does not follow from an exponent without a qualitative regime shift. The domain-specific Stream Power Law predicts incision using area and slope, not longest-path length. These relations clarify neighbors without declaring an unsupported strict parent.[1][3][5]
Neighborhood in Abstraction Space¶
Hack's law sits in a sparse region of the domain-specific corpus (98th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Unclustered & Miscellaneous (2551 abstractions)
Nearest neighbors
- Watershed — 0.80
- Drainage system (geomorphology) — 0.78
- Mesohabitat Simulation Model — 0.75
- Network Communication Efficiency — 0.74
- Playfair's law — 0.74
Computed from structural-signature embeddings · 2026-10-08
Not to Be Confused With¶
Do not use h=0.6 as an exact universal constant, compare coefficients without units, infer an exponent from one basin, or extrapolate a fit past reported deviations. Do not treat a D8-computed laboratory flow path as a visible stream channel. Do not transfer the tributary basin-shape interpretation automatically to a delta's nourishment area. Hack's law is the bounded empirical L–A relation with its drainage-network variables and measurement conditions made explicit.[1][2][4]
References¶
[1] John T. Hack, Studies of Longitudinal Stream Profiles in Virginia and Maryland (1957), DOI 10.3133/pp294B, U.S. Geological Survey Professional Paper 294-B, printed p. 47 for A and L definitions, printed p. 63 for Eq. 3 and the regional L=1.4 A^0.6 fit, and printed p. 64 for Figure 25 in Relation of Stream Length to Drainage Area. https://pubs.usgs.gov/pp/0294b/report.pdf registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r
[2] Mohsen Cheraghi et al., Catchment Drainage Network Scaling Laws Found Experimentally in Overland Flow Morphologies (2018), Geophysical Research Letters 45, 9614–9622, §2 printed p. 2 for the laboratory setup, §3 printed pp. 3–4 and Figure 4 for the conditional-length-moment analysis and finite exponent range. DOI 10.1029/2018GL078351. https://infoscience.epfl.ch/server/api/core/bitstreams/c90ded1d-cb95-4f82-9eee-80f0b046c95f/content registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p
[3] Peter S. Dodds and Daniel H. Rothman, Geometry of River Networks I: Scaling, Fluctuations, and Deviations (2001), Physical Review E 63, 016115, abstract for restricted approximate scaling and deviations not removed by resolution. DOI 10.1103/PhysRevE.63.016115. https://arxiv.org/abs/physics/0005047 registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j
[4] Dong et al., Apparent Hack's Law in River Deltas (2026), Science 392, 493–498, publisher/PubMed abstract for global apparent length–nourishment-area scaling and local uniform/composite pattern differences. DOI 10.1126/science.ady6805. https://pubmed.ncbi.nlm.nih.gov/42060736/ registry ↩a ↩b ↩c ↩d ↩e
[5] Gregory E. Tucker and Kelin X. Whipple, Topographic Outcomes Predicted by Stream Erosion Models: Sensitivity Analysis and Intermodel Comparison (2002), DOI 10.1029/2001JB000162, Journal of Geophysical Research: Solid Earth 107(B9), 2179, §2.1 for the generalized stream-power incision law with drainage area and channel gradient. https://doi.org/10.1029/2001JB000162 registry ↩a ↩b
[6] Riccardo Rigon et al., On Hack's Law (1996), Water Resources Research 32(11), 3367–3374, publisher abstract for the investigated connection between Hack exponent, basin elongation, and internal structure. DOI 10.1029/96WR02397. https://agupubs.onlinelibrary.wiley.com/doi/10.1029/96WR02397 registry ↩