New Observational Tools and Datasources for Hydrology: Hydrological data Unlocked by Tinkering

Advances in scientific understanding follow a cyclic pattern where new observational techniques lead to novel theoretical insight. From novel theories, in turn, hypotheses are derived that cannot be tested using current observations and thus create a demand for even newer observational techniques. Hydrology has recently experienced a period of great advances in theoretical understanding with hydrological models gaining considerably in complexity. Currently, the state of the hydrological science is at a point where further increases in complexity of hydrological models can no longer be supported by available data. The information content of currently used data is too low to uniquely identify the numerous parameters in hydrological models. Therefore new, and above all, different observational methods are needed to unlock thus far hidden hydrological data. In this thesis it is argued that three developments can be greatly beneficial to help hydrologists unlock new data: \u95 The occurrence of sensors in nearly every (consumer) device and the associated drop in the cost of those sensors \u95 The rise of the “open source hardware” movement, which promotes sharing hardware designs. \u95 The online sharing of research results and large data sets To be able to use these developments successfully in hydrology, pragmatic researchers are needed. Scientists whom, using a “MacGyver” attitude towards science, can show proof-of-concept results of how these new developments can best benefit hydrology. In this thesis, four such proof of concepts are presented, summarised below. Using the Wiimote as a sensor in water research An example of how consumer electronics can be used for hydrology is given in chapter 2. The $40 “Wiimote” (an input device belonging with the Nintendo® Wii\u99game system) is a device that contains three accelerometers and an infrared camera with built-in source tracking. It communicates by Bluetooth®. Because of the efforts of the hacking community it is easy to let the Wiimote communicate with a standard personal computer. Using a floating evaporation pan as an example, it is shown that the Wiimote, although it may have potential drawbacks when used in field campaigns, is a good addition to the hydrologist’s bag of tools. This first proof-of-concept work can inspire other scientists to consider using consumer electronics based sensors in their work. A resonating rainfall and evaporation recorder In chapter 3 a novel method of measuring rainfall and evaporation is presented. The device is basically a collection vessel (bucket) placed on top of a slender rod that is securely fixed at its base. As the vessel is deflected, either by manual perturbation or ambient forcing (for example, wind), its oscillatory response is measured by a miniature accelerometer. This response can be modeled as a damped massspring system. As the mass of water within the collection vessel changes, through either the addition of precipitation or by evaporative loss, the resonant frequency experiences an inverse shift. This shift can be measured and used to estimate the change in mass of water. This concept was tested by creating a simple prototype which was used in field conditions for a period of one month. The instrument was able to detect changes in mass due to precipitation with an accuracy of approximately 1mm. Design, calibration and field evaluation of an acoustic disdrometer designed for distributed measurements Another novel raingauge is presented in chapter 4. The “Delft-disdrometer” is an acoustic disdrometer: a raingauge that measures the size of individual raindrops by recording the sound that impacting raindrops make when they hit the sensor. The Delft-disdrometer is specifically designed to be low maintenance, thus allowing the installation of dense networks without incurring large upkeep costs. In a field evaluation, the Delft-disdrometer was compared to existing rain gauges and disdrometers. Results show that the Delft-disdrometer is capable of measuring precipitation intensities with comparable uncertainties as those of the industry standard Thies LPM and Ott Parsivel. For drops bigger than 1.75mm, the Delft-disdrometer measured drop size distributions similar to the drop size distributions measured by the Thies LPM and Ott Parsivel. The inability of the Delft-disdrometers to measure drops smaller than 1mm is an issue that should be addressed in future iterations of the design. The experiments with this first prototype show that it is feasible to measure precipitation intensities with the low maintenance Delft-disdrometer. Medicinal Footprint of the population of the Rhine basin The final example is given in chapter 5. By using freely available demographic data the relation between pharmaceutical residues along the river Rhine and the demographic characteristics of the upstream population was studied. A sampling campaign was performed in which water samples from the Rhine taken at 42 locations. Measurements were compared to a two parameter model with regional demographic data as main input. For 12 out of the 21 studied pharmaceuticals, a significant dominant demographic group could be identified. For 3 out of these 12 pharmaceuticals male elderly were the most contributing demographic group. A Monte Carlo analysis showed a high level of significance for the results of this study (? < 0.01). By combining environmental water quality data and demographic data, better insight was gained in the interplay between humans and their environment, showing the medicinal footprint of the population of the Rhine basin. Conclusions and look into the future In the ongoing cycle between new observational methods and new theories, hydrology is due for a round of new observational methods. Because of the complexity of current day hydrological models, richer datasets are needed to test the predictions (hypotheses) that those complex models make. More spatially distributed measurements of well known hydrological variables, as well as novel measurements of previously thought unmeasurable, or unimportant, variables are needed. Sensors designed for use in consumer electronics are often cheaper than sensors specifically designed for scientific research, because of the economies of scale achieved in the production of consumer electronics. Using these sensors normally required both hydrological insight as well as skills with (micro)electronics. However, the advent of the open hardware movement, spearheaded by the Arduino platform, has significantly lowered the threshold for hydrologist to build their own sensors. And finally: not all new data need to come from sensors: by looking at data sources outside of hydrology, such as the demographic data used in chapter 5, additional hydrological insight can be obtained. However, before the benefits of low cost sensors, open source hardware and open data can be reaped by all hydrologists three steps need to be taken. Work on innovative measurement techniques should be accompanied by step by step instructions on how to recreate the measurement device. Producers of sensors and open source hardware need to provide plug and play functionality. Geoscientific data needs to be stored in standardised formats, whether online or on local files. The work presented in this thesis shows examples of how lower cost sensors based on consumer electronics, open source hardware and open data sources from outside of hydrology can help hydrology move into the next phase of its ever continuing cycle.Water ManagementCivil Engineering and Geoscience