Pumping Subsidence in the Resettlement Districts: Property-Damage Claims Likely Track Estimated Settlement of Up to Several Centimetres in Frequency, While Their Size Remains Unresolved
Abstract
Assessors of the Australian Resettlement claims have attributed cracked dwellings to wear from crowded occupation, arguing that weeks to months of emergency pumping cannot move the ground. This premise was tested on 140 survey cells in six districts of the containment zone, with 3,309 property-damage claims lodged in BkE 2–6. Pumping was reconstructed from installed well capacity, because stations outside Australia logged nothing and the records inside are partial, and calibrated on 58 re-levelled cells. Estimated pumping subsidence had a median of 9.7 mm and a maximum of 55.6 mm, which fits under the drainage ceiling only for a window of about 2.8 months or more, an assumption of the paper. A negative binomial model with building age and occupancy held fixed gave 1.27 times the expected claims per 10 mm (95% CI 1.19 to 1.36), the measured settlement of the surveyed cells gave 1.26 (95% CI 1.18 to 1.34), and the 127 cells under a one-month ceiling gave 1.29. Severity rose by a factor of 1.060 per 10 mm over all cells but not clearly on measured settlement (1.031, 95% CI 0.993 to 1.070). Claim frequency likely tracks the subsidence; the size of claims remains unresolved.
1. Introduction
The assessors who handle property-damage claims from the Australian Resettlement attribute cracked walls and sticking doors to wear from crowded occupation, on the ground that the Post-Deterrence Era lasted only weeks to months and that pumping over so short a time cannot move the ground (Szymanski, 5 BkE). This is probably wrong, and the error can be measured, because clay-rich columns under concentrated extraction can compact by up to several centimetres, and the re-survey returns show settlement of that order.
Economic stakes come before physical ones here, and they are hard to apprehend from any single dwelling. The forced relocation crowded humanity into Australia, the designated containment zone, and emergency wells supplied the shelters. The Resettlement Restitution Office (the Office), whose register is used here, later received 3,309 property-damage claims from 140 survey cells between BkE 2 and BkE 6. The sums paid depend on whether the damage is assigned to the pumping, to crowded use or to age.
We write in Bunker Era, Year 7 (BkE 7), and every source used here predates it. Stations outside Australia logged nothing during the Resettlement, and the records inside are partial (Haldorsen, 6 BkE), so the subsidence has to be estimated from surveys made afterwards by the Secretariat, the Water Allocation arm of the UN Planetary Defense Council (PDC), the one body whose levelling network was still in service. In order to test the assessors' premise we will study two things, namely the settlement of benchmarks that the Secretariat re-levelled, and the claims lodged in the same cells.
The method follows an earlier study of drought in the Trisolaran Crisis Water Allocation Zones, in which a count model described how many claims reached the insurers and a separate model described how large they were (Gu, 57 DE). Tosun and Valderrama (41 DE) traced the cracking of masonry on shallow footings to the swelling and shrinking of the clay beneath, and Eskildsen (46 DE) linked that cycle to the lowering of the water table. Similarly, Maruyama (50 DE) found settlement concentrated where and when the table fell furthest, and Ostrovsky (3 BkE) showed that thick aquitards release their water slowly. And Lindqvist and Okoro (49 DE) fitted hurdle and count models to claims under catastrophe covers.
Even if the claims follow the estimated subsidence, much of the variation in claim rates between cells is likely to stay unexplained.
2. Methods
Each of the 140 survey cells is a square of 5 km on a side (25 km²) in six Resettlement districts, with 24, 22, 26, 20, 25 and 23 cells, chosen by the Secretariat because each held an emergency well field and eligible housing, whether or not a claim had been lodged. The Office register gives the claims lodged in BkE 2–6 and the dwellings registered in each cell (486,806 in all, from 1,228 to 12,390 per cell), and it was compiled in BkE 7.
The Re-survey Programme levelled the cells' benchmarks in BkE 1–4 against the pre-Resettlement network, so the levelled interval holds the compaction of the pumping together with any residual consolidation after it stopped, seasonal shrink-swell and drift. Only 58 of the 140 cells (41.4%) kept an intact baseline benchmark, with 10, 9, 12, 8, 10 and 9 cells in the six districts, and these cells alone give a measured settlement, with a closure standard deviation of 6 mm assumed for each benchmark pair. The well inventory gave installed pump capacity, and borehole logs gave the clay share of the column. The Office's intake roll gave the persons recorded per dwelling (the standardised occupancy index – OCC), and the dwelling records gave a standardised index of building age.
Pumping cannot be observed, so it was reconstructed. The pumped depth q of a cell is its installed capacity (median 9,646 m³/h) multiplied by a duty factor of 0.55 and an operating window, divided by the cell area. Only the length of the Post-Deterrence Era is known, as weeks to months, and we adopt a window of three months (2,190 h) as a convention and not as a record. A different window rescales every q alike, which the calibration absorbs, but the ceiling on what the ground can do depends on it. The compaction-weighted pumping load (the pumped depth times the clay share – CWL) is the single predictor of settlement.
The measured settlement of the 58 surveyed cells was regressed on CWL by least squares with an intercept for background drift, and the slope was used to predict the pumping-caused settlement of all 140 cells (the estimated pumping subsidence – EPS). Intervals use t with 56 degrees of freedom. For a confined column with 120 m of compressible material, a skeletal specific storage of 5 × 10-5 per metre and a drawdown of 25 m, the equilibrium compaction is 150 mm. A thick aquitard drains with a time constant of about six months, and a window of t months realises the share 1 − exp(−t/6) of that value, which is 59 mm at three months. These inputs are our own choices for a clay-rich district, not records, and residual consolidation after pumping can only lift the ceiling toward the equilibrium value.
Claim frequency was modelled as a count per cell by negative binomial regression with the log of registered dwellings as offset, district fixed effects, and the standardised indexes of building age and occupancy (overdispersion 0.103, about 3.4 times the Poisson variance at the mean count). Five models were compared by Akaike's criterion (AIC): controls only (M0), controls plus pumped depth (M1), plus EPS (M2), plus pumped depth and clay share as separate terms (M3), and plus those two terms and their product (M4). Because the EPS is a fitted value, its multiplier scales inversely with the calibration slope, so intervals are given conditional on the slope and once more with the slope's 95% limits carried through. Severity, the mean amount paid per claim in a cell, was modelled by a gamma regression with log link, weighted by claims and given the same controls.
3. Results
Settlement rises with CWL, and the calibration slope was 57.0 mm per metre of clay-weighted pumped depth (95% CI 44.9 to 69.0), R² was 0.62 and the residual standard deviation was 8.14 mm. Predicting each district from the other five gave an error of 8.15 mm.
The EPS of the 140 cells had a median of 9.7 mm (interquartile range 5.6 to 17.1 mm), a 95th percentile of 31.5 mm and a maximum of 55.6 mm. Nine cells reached 30 mm, and one reached 50 mm. The measured settlement, net of the 0.9 mm intercept, had a median of 10.7 mm and a maximum of 48.0 mm. Every estimate lies below the 59 mm ceiling of a three-month window. Ceilings fall with the window. They are 12.0 mm at half a month, 23.0 mm at one month and 42.5 mm at two months. The largest EPS fits under them only for a window of 2.8 months or more, and the nine cells at 30 mm or more for about 1.3 months or more. A window of one month leaves 13 cells above the ceiling and half a month leaves 52. The estimates as a whole therefore need the longer part of the range of weeks to months, an assumption of this paper.
In all, the register holds 3,309 claims, 6.80 per 1,000 registered dwellings, and every cell had at least one, although the cells were chosen by eligibility and not by claims.
Frequency follows the EPS. In M2 each 10 mm of EPS multiplied the expected number of claims in a cell by 1.27 (95% CI 1.19 to 1.36, p < 0.001), holding district, building age and occupancy fixed; with the calibration slope's own limits carried through, the interval is 1.16 to 1.48. Building age multiplied the count by 1.25 per standard deviation (95% CI 1.17 to 1.34) and occupancy by 1.26 (95% CI 1.18 to 1.36, p < 0.001), so crowded use does carry claims, but OCC and EPS were almost uncorrelated across cells (r = -0.02). The fitted total was 3,253 claims against 3,309 observed.
Using only the 58 surveyed cells and their measured settlement, the multiplier was 1.26 per 10 mm (95% CI 1.18 to 1.34). With the EPS of the same cells it was 1.35 (95% CI 1.24 to 1.47), the lower figure for measured settlement being the one expected under levelling error. The 127 cells under the one-month ceiling gave 1.29 (95% CI 1.15 to 1.44).
Pumped depth alone fits worse than the settlement estimate. In M1 the multiplier was 1.81 per metre (95% CI 1.51 to 2.19), with an AIC of 981.8 against 971.1 for M2. In M3 it was 1.77 per metre and 1.09 per 0.1 of clay share (95% CI 1.04 to 1.14, p < 0.001), so that at a fixed pumped depth the cells with more clay had more claims. In M4 the product of pumped depth and clay share had a multiplier of 1.06 per metre and 0.1 of clay (95% CI 0.95 to 1.18, p = 0.27), so the data do not separate a product term from additive terms.
Severity rose more weakly than frequency did, and across all 140 cells the mean amount paid per claim was multiplied by 1.060 per 10 mm of EPS (95% CI 1.029 to 1.092), and the EPS removed 10.9% of the deviance left by the controls. In the 58 surveyed cells it was 1.068 per 10 mm of EPS (95% CI 1.022 to 1.115), but 1.031 per 10 mm of measured settlement (95% CI 0.993 to 1.070), which includes 1.
4. Discussion
The claims likely track the subsidence in frequency. Each 10 mm of estimated settlement raised the expected count by 27%, the measured settlement of the surveyed cells gave 26%, and the top EPS quartile carried 1.82 times the claim rate of the lowest. The assessors' premise, that the pumping could not have moved the ground, therefore likely fails for frequency.
A short window does not forbid compaction, but it limits it. The largest estimate of 55.6 mm fits under the ceiling only if the pumping lasted about 2.8 months or more, and a window of weeks would need thicker columns than assumed, consolidation continuing after the pumping, or another cause for the largest settlements. The 127 cells under the one-month ceiling gave 1.29.
We name one alternative. Pumping and crowding may have coincided in the same cells, and crowded dwellings would have been worn by use whatever the ground did. Occupancy does carry claims (1.26 per standard deviation), but it is almost uncorrelated with the EPS (r = -0.02), and the EPS multiplier is estimated with it held fixed. The clay share cannot help in its place, since at a fixed pumped depth the claims rose with clay (1.09 per 0.1), but natural shrink-swell damage predicts the same main effect, and the interaction that would separate compaction from it (1.06, 95% CI 0.95 to 1.18) is unresolved. The case against crowding therefore rests on the control for occupancy, a count of persons and not of load on each wall.
Severity is a weaker case. Across all 140 cells the multiplier of 1.060 would mean that a cell with 30 mm of estimated settlement pays about 19% more per claim than one without. Power does not explain the difference from frequency, because on the 58 surveyed cells the EPS still gave 1.068 while measured settlement gave 1.031. Either the EPS carries pump or clay effects on cost, or levelling error hides a real effect, and these data cannot tell which.
We write within the Luo-Ji / Ye framework. Its first axiom, that survival is the primary need of a civilisation, makes the cracked dwellings of a resettled humanity a survival cost that a ledger of claims can price whatever cause is assigned to them.
5. Limitations
The best model still cannot say what share of any one cell's claims the pumping caused. The EPS rests on installed capacity, a convention for the window and a calibration on 58 cells, and the window varied between cells in ways no record shows, a variation that probably shortens the multipliers toward 1. Sophons were active throughout and could have altered any single reading as it was logged, so no subset of the sheets is treated as free of them; the claims register, filled in by clerks, is independent of the levelling only in so far as a sophon had no reason to touch both.
Register entries end with BkE 6, while claims could still be lodged afterwards, so the counts are incomplete, and if late claims concentrate in the cells where drainage continued longest the multipliers are biased toward 1. Levelling gave total and not differential settlement, and differential settlement is what cracks masonry (Tosun and Valderrama, 41 DE). Clay soils also crack housing in dry spells without any pumping, and no earlier claims record could separate the two.
Monthly levelling, differential settlement across individual footings and a recorded pumping history would each improve these estimates more than any change of model, and they would settle the severity question. The estimates nevertheless give the Office a defensible basis for a damage schedule by cell, which it can revise as the late claims arrive.
References
- PDC Water Allocation Secretariat (5 BkE). Re-survey Programme returns for the Resettlement districts, benchmark and well-inventory sheets, BkE 1–4. PDC Declassified Holdings, New York, series WAS-R.
- Resettlement Restitution Office (7 BkE). Register of lodged property-damage claims by survey cell, BkE 2–6. Reconstituted Chinese Academy of Sciences, Archival Series, accession RRO-C.
- Tosun, B., and Valderrama, R. (41 DE). Cracking of shallow-footed masonry dwellings on expansive clay in the middle reach. Journal of Crisis-Era Civil Engineering, 12(2), 88–107.
- Eskildsen, M. (46 DE). Drawdown below the Three Gorges reservoir and the shrink-swell cycle of the middle-reach clays. Journal of Crisis-Era Civil Engineering, 17(1), 5–29.
- Lindqvist, A., and Okoro, C. (49 DE). Hurdle and zero-inflated count models for claims under catastrophe covers. PDC Working Papers in Strategic Studies, 52, 1–34.
- Maruyama, T. (50 DE). Levelling surveys of settlement in rebuilt housing, Zone III. Journal of Crisis-Era Civil Engineering, 20(3), 211–238.
- Ostrovsky, L. (3 BkE). Delayed drainage of thick clay aquitards under pumping periods of weeks to months. Proceedings of the Reconstituted Chinese Academy of Sciences, 79, 41–66.
- Szymanski, R. (5 BkE). Assessment practice for Resettlement property claims, with the schedule of exclusions. Bulletin of Post-Deterrence Historical Studies, 3, 71–98.
- Haldorsen, V. (6 BkE). Instrument gaps in the records of the Australian Resettlement, an inventory of what was logged. Bulletin of Post-Deterrence Historical Studies, 4, 63–91.
- Wanqing, G. (2026). Hydrology Predicts How Many Subsidence Claims Reach Insurers in the Water Allocation Zones but Not How Large They Are: Evidence from Four Insurers, DE 31–54. Uncited Press.
Open in Uncited Press →