16-Civ-B4 Engineering Hydrology · December 2013
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Paper format. National Examinations, December 2013 — 98-Civ-B4 Engineering Hydrology. Three hours, closed book, one candidate-prepared two-sided 8½″ × 11″ aid sheet, and one approved Casio or Sharp calculator whose model must be declared. Seven problems are printed; page-1 Note 4 states that any five (5) questions constitute a complete paper and that only the first five answers appearing in the workbook are marked. Each problem carries twenty (20) points, so the examinable total is 5 × 20 = 100 points. The page-6 marking scheme breaks each problem into its sub-parts. All seven problems are solved here, because this set is a study resource rather than a timed sitting; the sub-part mark values shown below are the printed ones.
Reference texts. V. T. Chow, D. R. Maidment and L. W. Mays, Applied Hydrology (hydrologic cycle, unit hydrographs, routing, frequency analysis, infiltration); L. W. Mays, Water Resources Engineering, 3rd ed. (design application, rainfall–runoff, reservoir operation); W. Viessman and G. L. Lewis, Introduction to Hydrology, 5th ed. (measurement, areal precipitation, energy budget); V. T. Chow, Open-Channel Hydraulics (1959) (flood-wave propagation, dam-break and gradually varied unsteady flow); C. W. Fetter, Applied Hydrogeology, 4th ed. (Darcy’s law, hydraulic conductivity, recharge). Canadian practice references: Environment and Climate Change Canada / Water Survey of Canada HYDAT archive and the ECCC Engineering Climate Datasets (IDF curves and the IDF_CC climate-adjustment tool); the ISO 1100 / WMO Manual on Stream Gauging series as adopted by the Water Survey of Canada; the Canadian Dam Association Dam Safety Guidelines (inflow design flood and dam-break consequence classification); and the Transportation Association of Canada Guide to Bridge Hydraulics, 2nd ed.
Check — conventions used throughout this paper. A hydrologic year is taken as 365 days = 31 536 000 s unless a question says otherwise. Water density is 1000 kg/m3, gravitational acceleration is 9.81 m/s2, and the latent heat of vaporisation of water is 2.45 MJ/kg at 20 °C. Several sub-parts ask for an explanation with an example rather than for the solution of stated data; in those cases a realistic Canadian data set is declared at the point of use and every number arising from it. Where the printed data are internally inconsistent — and Question 6(i) is such a case — the inconsistency is demonstrated arithmetically, the governing conservation requirement is stated, and the corrected reading actually used is declared, as page-1 Note 1 invites (“the candidate is urged to submit… a clear statement of any assumptions made”).
Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.
Given. A 40 ha urban catchment with a runoff coefficient $C = 0.75$ and a time of concentration of 15 minutes, to be designed for a 25-year return period. The local ECCC IDF curve gives a 15-minute, 25-year intensity of 90 mm/h.
Find. How the curve is entered and used, and how a factor of safety is applied to the result.
How the curve is used. An IDF curve is the output of a frequency analysis performed on the annual maximum rainfall intensity for each of a set of durations: for every duration the annual maxima are fitted, usually with a Gumbel distribution, and the quantiles are read off and re-plotted as one curve per return period. The designer enters it with two independent choices. The duration is taken as the time of concentration of the catchment, because that is the storm length that first brings the whole area to contribute simultaneously and therefore produces the largest peak; shorter storms are more intense but do not engage the whole catchment, and longer ones engage it at a lower intensity. The return period is set by the consequence of failure and by the governing municipal standard — typically 2 to 10 years for a minor storm sewer, 100 years for the major overland system and for a highway crossing. Reading at 15 minutes and 25 years gives 90 mm/h, and the Rational method converts it to a design peak:
$$Q = k\,C\,i\,A = 0.00278 \times 0.75 \times 90 \times 40 = 7.51\ \text{m}^3/\text{s}$$How safety factors address the underlying variability. The curve is a fitted statistic drawn from a record of finite length, so the plotted quantile carries a standard error that grows sharply as the return period exceeds the record length; and the record itself is assumed stationary, which a warming climate makes false. Engineering practice does not hide this variability inside the answer — it adds explicit margin, in four recognised forms. Return-period margin: designing the outlet or the crossing for a return period well beyond the service standard (a 100-year check on a 10-year system), which is a factor of safety expressed in probability rather than in load. Intensity uplift: multiplying the read intensity by a climate-change factor, typically 1.15 to 1.20 over a design life, obtained in Canada from the ECCC IDF_CC tool. Applying a 20 percent uplift here,
$$i_{\text{design}} = 1.20 \times 90 = 108\ \text{mm/h}, \qquad \boxed{Q_{\text{design}} = 0.00278 \times 0.75 \times 108 \times 40 = 9.01\ \text{m}^3/\text{s}}$$a 20 percent increase in the design flow. Freeboard: a physical height margin above the computed water level at culverts, channels and ponds, which absorbs both the statistical error and the modelling error in one allowance. Conservative parameter choice: adopting the fully built-out runoff coefficient rather than the present one, and the rougher end of the Manning range where roughness reduces capacity. The four are cumulative and should be stated separately in the design brief, so that a later reviewer can see how much margin exists and why.
| Quantity | Basis | Result |
|---|---|---|
| Design intensity from the IDF curve | 15 min duration, T = 25 a | 90 mm/h |
| Design peak, Rational method | 0.00278 × 0.75 × 90 × 40 | 7.51 m3/s |
| Climate-uplifted intensity | 1.20 × 90 | 108 mm/h |
| Design peak with uplift | 0.00278 × 0.75 × 108 × 40 | 9.01 m3/s |
| Effect of the safety factor | Qdesign / Q | 1.20 |
Prediction of flood and intense-rainfall impacts in a 100 km2 mixed rural and urban watershed is a probability calculation, and it can be no better than three sets of information that feed it.
Set 1 — a long, homogeneous and quality-controlled hydrometeorological record, with its metadata. The frequency analysis needs the annual maximum series of instantaneous peak discharge at the watershed outlet and of rainfall intensity at the relevant durations, ideally 30 years or more, because the standard error of a T-year quantile grows rapidly once T exceeds the record length n. It equally needs the metadata: gauge moves, rating shifts, ice-affected periods and infilled gaps. And it needs a homogeneity check, because the urbanising part of the watershed makes the series non-stationary — the pre-development and post-development peaks are not draws from the same distribution, and mixing them produces a fitted curve that describes no real condition. Regional information belongs in this set too: for a short local record, an index-flood or regional regression analysis pooling hydrologically similar Canadian basins is more reliable than a long extrapolation of the local one.
Set 2 — the watershed’s physical and land-use characterisation, present and future. Probability enters the answer through the rainfall, but the transformation from rainfall probability to flood probability is done by the catchment: area, slope, drainage density, time of concentration, soil hydrologic groups, storage in wetlands and ponds, imperviousness by sub-catchment, and the location and capacity of the drainage infrastructure. This set is what makes a mixed rural–urban watershed different from either pure case: the urban fraction responds in minutes and the rural fraction in hours, so the critical storm duration differs between them and the two peaks may or may not coincide at the outlet. The set must include the planned land use as well as the present one, since the design must remain valid over the asset’s life, and it must include antecedent-condition information (soil moisture, frozen ground, snowpack water equivalent) because the same rainfall probability maps to very different flood probabilities depending on the state of the basin when it arrives.
Set 3 — the exposure, vulnerability and consequence information that converts a flood probability into an impact probability. A frequency curve alone predicts discharge, not impact. Converting one to the other requires floodplain topography of survey quality (LiDAR-based digital elevation model), hydraulically modelled stage–discharge and inundation extent, an inventory of what lies in each depth band — buildings by type and first-floor elevation, roads, the water and wastewater plants, the hospital, the electrical substations — and depth–damage relations for each class. The product of the three sets is the expected annual damage, $E[D] = \int D(q)\,f(q)\,dq$, which is the quantity that actually justifies mitigation spending. It is also the set most often missing: a great many watersheds have a defensible flood-frequency curve and no credible damage inventory, and in those the prediction of impact is far weaker than the prediction of flow.
What they are. A data outlier is an observation in the sample that departs so far from the trend of the remaining data that it appears not to belong to the same population — formally, a value whose deviation from the fitted distribution is larger than a stated threshold. In flood frequency the standard test operates on the logarithms of the annual peaks and uses the Bulletin 17B/17C thresholds
$$\log Q_{H,L} = \overline{\log Q} \pm K_N\,s_{\log Q}$$where $K_N$ depends on the sample size N. For a 40-year record, $K_N = 2.682$; with a mean log of 2.60 and a log standard deviation of 0.25 the thresholds are $\log Q_H = 2.60 + 0.6705 = 3.2705$ and $\log Q_L = 2.60 - 0.6705 = 1.9295$, that is a high-outlier threshold of 1863 m3/s and a low-outlier threshold of 85.0 m3/s. An outlier is not automatically an error: it may be a genuine rare event, a value from a different physical mechanism (an ice-jam peak or a rain-on-snow peak in an otherwise snowmelt series), or a recording or rating mistake.
Treatment of a high outlier, and why. First check the record: verify the gauging, the rating extrapolation and any high-water marks. If the value is a genuine flood, retain it — and if historical information exists showing it is the largest in a period longer than the systematic record (a documented 1948 flood in a record that starts in 1970), retain it with a historically adjusted weight rather than at the weight of one ordinary year. Deleting a real extreme is the most damaging thing a hydrologist can do to a frequency curve, because it removes precisely the information the upper tail is being fitted from and biases every design quantile downward. Only a demonstrated measurement error justifies removal.
Treatment of a low outlier, and why. Low outliers are treated in the opposite spirit. A very small annual peak — from a drought year, or from a gauge that was ice-affected or out of service — carries almost no information about the flood tail, yet in a log-space fit it exerts great leverage on the mean and, more importantly, inflates the standard deviation and drags the skew, distorting the 100-year estimate. Standard practice is therefore to censor low outliers: remove them from the moment computation and re-fit using the conditional probability adjustment, which re-scales the fitted frequency curve for the fact that a known number of years fell below the censoring threshold. The record of the deletion, the threshold used and the number censored must be reported. The general principle behind both treatments is the same: keep everything that carries information about the part of the distribution being used for design, and censor what only adds leverage without adding information — and never treat either decision as a licence to fit a curve you find convenient.
| Quantity | Basis | Result |
|---|---|---|
| Outlier coefficient, N = 40 | Bulletin 17B table | KN = 2.682 |
| High-outlier threshold (log) | 2.60 + 2.682(0.25) | 3.2705 |
| High-outlier threshold | 103.2705 | 1863 m3/s — retain if genuine |
| Low-outlier threshold (log) | 2.60 − 2.682(0.25) | 1.9295 |
| Low-outlier threshold | 101.9295 | 85.0 m3/s — censor and adjust |
