Marsh, Terry; Harvey, Catherine L. 2012 The Thames flood series: a lack of trend in flood magnitude and a decline in maximum levels.Contact CEH NORA team at noraceh@ceh.ac.ukThe NERC and CEH trademarks and logos ('the Trademarks') are registered trademarks of NERC in the UK and other countries, and may not be used without the prior written consent of the Trademark owner. The flow series for the River Thames near its tidal limit is one of the most studied in the 28 world. Its length and completeness, and the richness of the historical information which 29 augments the formal flow record, ensures that the series is of immense value. However, 30 interpretation of the variability in flood magnitude and frequency that it captures needs to 31 be undertaken with caution. The homogeneity of the time series is influenced by a wide 32 range of factors -including changes in the hydrometric capability of the gauging station 33 and the impact of differing water, river and land management practices on the flow 34 regime. 35 36Nevertheless, both the daily flow series and the record of lock levels provide some 37 reassuring signals regarding the resilience of the Thames to fluvial flood risk in a 38 warming world. Since routine flow measurement began in 1883, the Thames basin has 39 seen a substantial rise in air temperature, and a tendency for both winter rainfall and 40 annual runoff to increase. However, there is no trend in fluvial flood magnitude, 41 reflecting in part a decline in snowmelt contributions to major floods; and annual 42 maximum lock levels show a significant decline, reflecting a very sustained programme 43 of river management. 44 45 46

River flow records are also a vital input to hydrological models, including those used for 39 predicting future behaviour (Hannah et al. 2010); gaps can have a deleterious impact on 40 estimates derived from prediction and forecasting tools. Complete records are therefore 41 critical to the sustainable management of water resources worldwide, and gaps in records 42 represent a loss of information which can potentially affect the interpretation of data, and the 43 scientific outcomes of analysis; Marsh (2002) argues that, in many cases, the inclusion of 44 suitably flagged flow estimates is preferable to leaving gaps in records. 45Within the UK, the National River Flow Archive (NRFA) acts as the main 46 hydrometric archive, collating data from different monitoring network operators. Daily mean 47 river flows are stored for over 1500 gauging stations and validated, analysed and 48 disseminated to a wide range of users (Dixon 2010). Whilst the majority of these flow records 49 have high overall percentage completeness (78% of stations have records that are at least 50 3 95% complete, Marsh & Hannaford 2008), closer inspection reveals a significant quantity of 51 both contemporary and historical gaps, ranging in length from a single day to several months. 52Such gaps are an inevitable consequence of factors such as essential gauging station 53 maintenance, equipment malfunction, changes in instrumentation, data processing issues and 54 human error. For some gaps, the data are likely to be unrecoverable; for example, an extreme 55 high flow event that destroys a gauging station may be difficult to estimate with any degree 56 of certainty. In most cases, however, gaps may be amenable to infilling, particularly where 57 hydrological conditions are relatively stable. 58A previously observed decline in the completeness of river flow data submitted to the 59 NRFA (Marsh 2002) can in part be attributed to a lack of standardised infilling guidance 60 which, in its absence, has discouraged the infilling of gaps. While there has been a 61 demonstrable improvement in completeness in recent years (Dixon 2010), historical data gaps 62 remain and short sequences of missing daily mean flows (which appear readily amenable to 63 infilling) still regularly occur in data submitted to the NRFA. This highlights a need for 64 informed guidance on the use of infilling techniques to promote a consistent, repeatable 65 approach towards such record gaps. Simple, quick-to-apply techniques that perform well 66 across an extensive range of catchments could find wide applicability, thus limiting the 67 investment of time and resources required to infill data to an appropriate degree of accuracy, 68 while also significantly enhancing the overall utility of time series. However, there are 69 currently no widely-accepted standard techniques for data infilling, either in the UK or 70 internationally. 71The aim of this paper is to evaluate the performance of a range of existing simple 72 methodologies for gap filling. A variety of catchment types in ...

Complete river flow time series are indispensable to the sustainable management of water resources and even very short gaps can severely compromise data utility. Suitably-flagged flow estimates, derived via judicious infilling, are thus highly beneficial. The UK National River Flow Archive provides stewardship of and access to daily river flow records from over 1500 gauging stations and, whilst the majority are sensibly complete, historical validation reveals a significant quantity of gaps. A full assessment of the suitability of existing techniques for infilling such gaps is lacking. This paper therefore presents an appraisal of various simple infilling techniques, including regression, scaling and equipercentile analysis, according to their ability to generate daily flow estimates for 25 representative UK gauging stations. All of the techniques rely upon data transfer from donor stations and results reveal that the equipercentile and multiple regression approaches perform best. Case studies offer further insight and an example of infilling is presented, along with areas of future study. The results demonstrate the potential for developing generic infilling methodologies to ensure a consistent and auditable approach towards infilling, which could find wider application both within the UK and internationally.