Article 1

OpenHi.net: A synergistically built, national-scale infrastruc- 2

ture for monitoring the surface waters of Greece 3

Nikos Mamassis 1,*, Katerina Mazi 2, Elias Dimitriou 3, Demetris Kalogeras 4, Nikolaos Malamos 5, Spyridon 4
Lykoudis 2, Antonis Koukouvinos 1, Ioannis Tsirogiannis 5, Ino Papageorgaki 1 and Anastasios Papadopoulos 3 5

1 Department of Water Resources & Environmental Engineering, National Technical University of Athens 6
2 Institute for Environmental Research & Sustainable Development, National Observatory of Athens 7
3 Institute of Marine Biological Resources & Inland Waters, Hellenic Centre for Marine Research 8
4 Institute of Communication and Computer Systems, National Technical University of Athens 9
5 Department of Agriculture, University of Patras 10
* Correspondence: nikos@itia.ntua.gr 11

Abstract: The large-scale surface-water monitoring infrastructure for Greece Open Hydrosystem 12
Information Network (Openhi.net) is presented in this paper. Openhi.net provides free access to 13
water data, incorporating existing networks that manage their own databases. In its pilot phase, 14
Openhi.net operates three telemetric networks for monitoring the quantity and the quality of sur- 15
face waters, as well as meteorological and soil variables. New members can join up by offering their 16
data to a common, publicly accessible database. A web-platform was developed for on-line visu- 17
alization, processing and managing telemetric data. A notification system was also designed and 18
implemented for inspecting the current values of variables. The platform is built upon the web 2.0 19
technology that exploits the ever-increasing capabilities of browsers to handle dynamic data as 20
time-series. A GIS component offers web-services relevant to geo-information for water bodies. 21
Citation: Mamassis, N.; Mazi, K.;
Accessing, querying and downloading geographical data for watercourses (segment length, slope, 22
Dimitriou, E.; Kalogeras, D.; Mala-
name, stream order) and for water basins (area, mean elevation, mean slope, basin order, slope, 23
mos, N.; Lykoudis, S.; Koukouvinos,
mean CN-curve number) are provided by Web Map Services and Web Feature Services. A new 24
A.; Tsirogiannis, I.; Papageorgaki, I.;
Papadopoulos, A. OpenHi.net: A
method for estimating the streamflow from measurements of the surface velocity has been ad- 25

synergistically built, national-scale vanced, too, and, to reduce hardware expenditures, a low-cost ‘prototype’ hydro-telemetry system 26
infrastructure for monitoring the (at about half the cost of a comparable commercial system) was designed, constructed and installed 27
surface waters of Greece. 2021, 13, at six monitoring stations of Openhi.net. 28
x. https://doi.org/10.3390/xxxxx
Keywords: observation networks; low-cost solutions; surface water monitoring; discharge estima- 29
Academic Editor(s): tion; open access 30
31
Received: date
Accepted: date
Published: date
1. Introduction 32

Publisher’s Note: MDPI stays

Surface-water monitoring is essential not only for the protection of the environment 33
neutral with regard to jurisdictional but also for the design of hydraulic works and for the management of water resources. 34
claims in published maps and Since the 1950s, streamflow in Greece has been monitored by the Public Power Corpora- 35
institutional affiliations. tion, which has been systematically gauging the main rivers in mountainous areas for the 36
hydroelectric power plant design and operation, and by the Ministries of Public Works 37
and of Agriculture as a part of their water resources development, flood protection and 38
land reclamation activities. These entities have also established an extensive network of 39
Copyright: © 2021 by the authors.
meteorological stations, comprising mainly conventional rainfall gauges for daily ob- 40
Submitted for possible open access
servations. The National Meteorological Service and (recently) the National Observatory 41
publication under the terms and
of Athens collect data on the gamut of meteorological variables. Finally, the Institute of 42
conditions of the Creative Commons
Geology & Mineral Exploration of Greece (supervised by the Ministry for Development) 43
Attribution (CC BY) license
monitors the groundwater resources. 44
(http://creativecommons.org/licenses
/by/4.0/).
Up until the 1990s, the hydrological and meteorological data were collected by 45

Water 2021, 13, x. https://doi.org/10.3390/xxxxx www.mdpi.com/journal/water

Water 2021, 13, x FOR PEER REVIEW 2 of 14

conventional means. The continuously recording instruments used paper charts, hence 46
regular replacing (daily, weekly etc.) of that paper was required. In the 1990s, various 47
public entities (municipalities, prefectures, institutes) installed automatic telemetric sta- 48
tions, mainly for collecting meteorological data. During the same period, the evolution of 49
informatics led to digitizing older conventional data and providing them through the 50
ubiquitous internet. Hydroscope (Koutsoyiannis et al., 1995; Mamassis et al. 2010) 51
was the first attempt at unifying the fragmented hydro-meteorological information 52
landscape (Sakellariou et al., 1994). It created a national data base (DB), however, that 53
effort was only partially successful: today Hydroscope is simply maintained by the Spe- 54
cial Secretariat for Water, Ministry of Environment and Energy . The lesson learned was 55
that data from various sources that did not follow a common model could not be har- 56
monized, so common ground truth couldn’t be ascertained. Consequently, this was not 57
perceived as a success in collective data acquisition. Today, several telemetric networks 58
are in operation that collect surface-water information independently from each other, at 59
the local to the regional scale. 60
When ?
The Open Hydrosystem Information Network (Openhi.net) was created to remedy such 61
? pathogenies: it is a large-scale infrastructure for collecting, processing and storing data 62
on the surface waters of Greece, serving applied, research and operational needs (Efstra- 63
tiadis et al., 2020). The tenet underlying Openhi.net is openness, free access to data and 64
monitoring infrastructure. Paramount in this work are (a) that partners develop and 65
manage their own DBs, but according to common agreed-upon standards, and provide 66
data to a common, publicly accessible DB, and (b) that new members can join up by also 67
offering their data for public access. In its pilot phase Openhi.net operates a small number 68
of stations in few river basins, but it is envisioned as precursor of a national hydro-data 69
infrastructure for the surface waters of Greece (Efstratiadis et al., 2020). The observed 70
variables fall in four areas: 1) water quantity: river-lake stage and streamflow; 2) water 71
quality: temperature, pH, conductivity, and dissolved oxygen; (3) meteo-variables: tem- 72
perature, rainfall, wind speed, relative humidity and solar radiation and (4) soil variables 73
(temperature and soil moisture). 74
Openhi.net includes (a) a web-platform for visualizing, processing and managing 75
telemetric data, (b) three pilot telemetric networks, installed for monitoring quantity and 76
quality of surface waters, as well as meteorological and soil variables, and (c) a GIS 77
component offering web-services relevant to geo-information for water bodies (topolog- 78
ically consistent hydrographic network, lakes, reservoirs etc.). Openhi.net also (i) fosters 79
the development of smart, low-cost hydrometric and data transmission technologies, (ii) 80
evaluates the existing gauging infrastructure and (iii) elaborates a strategic plan for es- 81
tablishing a national monitoring network. In our connected world, pooling multiple data 82
sources allows researchers and practitioners to come together in creative ways, enhances 83
data understanding and results in broadly applicable insights. 84
Openhi.net aims to bring together existing networks and to provide reliable infor- 85
mation on the country's surface waters to various users, such as water resources profes- 86
sionals, environmental agencies, researchers and the public. The visibility of the project is 87
important, as: (a) historical measurements are critical for the management of water re- 88
sources, for the design of hydraulic and renewable energy projects, for climatic assess- 89
ments and for agricultural management; (b) real-time observations are crucial for the 90
operation of early warning systems; and (c) incorporation of existing networks into the 91
infrastructure gives the opportunity for more effective management and dissemination of 92
this important information. 93
?
It is also worth noting that the Openhi.net initiative is in line with the perspective on 94
hydrologic measurements and observations of the hydrological community, expressed in 95
a recent survey conducted by the International Association of Hydrological Sciences 96
(Tauro et al. 2018), that: (1) new measuring techniques and equipment, as well as more 97
how do results of the field data and monitoring, were needed to advance hydrological science; (2) maintaining 98
survey actually inform the monitoring networks was the main challenge; (3) traditional monitoring systems 99
the design of
Openhi.net ?
 Water 2021, 13, x FOR PEER REVIEW 3 of 14

suffer from major limitations, and observations through standard equipment still grasp 100
natural processes inadequately, also often tending to be expensive and to afford limited 101
spatial coverage; and (4) using advanced technology entails high maintenance costs and 102
access to trained staff and resources. Such difficulties have caused observations to de- 103
crease consistently since the 1980s, while most research groups cannot afford costly 104
monitoring equipment. Hydrologists have confronted the above challenges by designing, 105
developing and deploying their own sensors, taking advantage of open source control- 106
lers such as Arduino (https://www.arduino.cc). Scientists were thus inspired to build 107
their own or to modify off-the shelf equipment, developing low-cost, easily accessible, 108
References ?
and tailored sensors. 109
The OpenHi.net consortium consists currently of the following partner organizations: 110
• Department of Water Resources & Environmental Engineering (WREE), National 111
Need an intro sentence
Technical University of Athens (NTUA) 112
or two? • Institute for Environmental Research & Sustainable Development, National Obser- 113
vatory of Athens (NOA) 114
• Institute of Marine Biological Resources & Inland Waters, Hellenic Centre for Ma- 115
rine Research (HCMR) 116
• Institute of Communication & Computer Systems (ICCS) of NTUA 117
• Department of Agricultural Technology, University of Ioannina (UnIo) 118
NTUA (WREE and ICSS) designed the functionality of the system, developed the 119
web platform and database and performed its pilot operation; NOA created the wa- 120
ter-quantity monitoring infrastructure and collected streamflow data (instrumentation, 121
hydrometric campaigns, new discharge estimation method); HCMR created the wa- 122
ter-quality monitoring infrastructure and collected data on relevant variables; and UnIo 123
created the infrastructure for collecting meteorological data and soil variables. The web 124
platform, the GIS component and the telemetric networks are described next. 125

2. A Web Platform for Inland Waters 126

2.1. Services 127
Need an intro sentence The main services offered by the platform are: 128
or two? • A dynamic map showing in near-real-time observations for each selected variable 129
(Figure 1a) 130
• Retrieval of raw and processed time series (Figure 1b) 131
• Retrieval of geographical data and water basin processed parameters (area, slope, 132
time of concentration etc.) (Figure 1c) 133
• Integration of new stations in the system, which includes applications for station 134
All Figures are
management, data correction and time series process. 135
blurry and
hard to read

Figure
Legends
need a lot
of work
(accuracy,
sufficiency
of detail)

136
Figure 1. (a) Real-time observations, (b) retrieval of raw and processed data, (c) geographical data and water basin 137
parameters. 138

Furthermore, two distinct notification applications were designed and implemented 139
in the platform. The first application inspects whether the current value of a variable is 140
Water 2021, 13, x FOR PEER REVIEW 4 of 14

greater/lower than a given threshold. The second application concerns the evolution of 141
the variable and examines the differences of current value from values at previous time 142
steps. Specifically, for the current value xt are calculated: 143
(a) the differences of the current value from previous values for multiple (1…n) 144
time lags Δ1t = xt – xt-1, Δ2t = xt – xt -2, …Δnt = xt – xt-n (Figure 2a) and 145
Too much detail here? (b) the probability distributions (empirical or theoretical for a particular model) of 146
differences for various time lags Δ1t, Δ2t, …Δnt. The thresholds for each time lag are calcu- 147
lated according to a defined exceedance probably (Figure 2b). The thresholds are stored 148
for each time series and, in case they are exceeded, the system triggers a notification to 149
predefined recipients. 150

151
Figure 2. Examination of time evolution of a parameter: (a) Calculation of the differences of the current value from pre- 152
vious values and (b) definition of thresholds. 153

Are these
stations Streamflow monitoring can warn about extreme conditions when critical thresholds 154

referenced are exceeded. For instance, information on a rapidly rising stage upstream can provide 155

on a figure useful flood mitigation services by increasing the reaction time downstream. For exam- 156
ple, the analysis of recorded stages in Pinios river during major flood events reveal con- 157
map?
siderable travel times of the flood peak from the upstream station Nomi to the interme- 158
What are diate station Giannouli and the downstream station Tempi. The travel time between the 159
the lag two upstream stations is one to two days and one-half to one day between the two 160
times? downstream stations, depending on the magnitude and spatio-temporal pattern of the 161
extreme event occurring across the basin. These time lags are long enough to provide 162
Show data
downstream communities with the time needed for adequate preparation. 163
in a small
table?
2.2. Platform Design 164
The proposed platform is built upon the successful web 2.0 technology that prevails 165
on all the online platforms, mainly due to the ever-increasing capabilities of browsers 166
Refs for that can consume not only plane static pages but also dynamic data, and, in our case, 167
web time-series. The proposed platform is built around the well-known architectural Mod- 168
2.0, el-View-Controller (MVC) pattern for web application, Figure 3. 169
MVC ? 170
Water 2021, 13, x FOR PEER REVIEW 5 of 14

More explanation of figure needed 171

Figure 3. Typical MVC development pattern for web applications. 172

This pattern places a separation between the User Interface, i.e. the View component, 173
and the Controller component that receives updates from the View and notifies updated 174
Define Terms information to the View. The Model component encompasses the defined data structures 175
Hard to follow for handling the required information of the necessary objects. The platform considers 176
stations, instruments, administrators and time-series as objects (classes, in the modelling 177
terminology) with associated attributes. The Controller implements the required business 178
logic for implementing actions imposed by end-users or administrators. For instance, a 179
typical workflow action is the elimination of ‘improper’ time-series values extending 180
beyond the rational dynamic range of a value. 181
As this platform constitutes a national infrastructure with multiple stakeholders, it is 182
important to implement the sovereign relations originating from an institution and 183
spanning across their own set of stations. In a sense, there are multiple administrators, 184
each one corresponding to an affiliated member organization. In this initial phase of the 185
project, NOA, HCMR, NTUA and UoI are the founding members of this common infra- 186
structure. A contributing member organization performs initially certain boot-strapping 187
actions to instantiate a new station, with its equipment, within the web platform. Fur- 188
Too much jargon. thermore, the administrator may choose to initiate data quality control or an aggregation 189
Refs? process across selected time-series data. This means that the platform can generate de- 190
rivative quality-checked time-series, in contrast to the original “dirty” raw data. 191
The previous system implementation incorporated a file-based approach, where 192
time-series data were stored in flat files and only metadata were stored in a relational DB. 193
Such a scheme did not allow for value-related queries across an area, as, for instance, 194
what was the one-hour aggregated rainfall depth in a river basin? The proposed system ac- 195
commodates a data persistence layer that incorporates a time series and a GIS DB. This 196
allows time-interval related data handling of stored values as well as related spatial and 197
geographic data. The implemented system utilizes TimeScaleDB, a well-known open 198
source Postgress extension for column-oriented data, where the main row information is 199
time and columns represent the different time series values. 200
201
Water 2021, 13, x FOR PEER REVIEW 6 of 14

3. GIS component 202

Geographic data are essential for linking streamflow to hydrological characteristics 203
of upstream basins. The OpenhiGis is the GIS component of Openhi.net. Its main purpose 204
is to collect and analyze data, to query and to offer web-services related to hydrologi- 205
cal-geographical information for watercourses and hydrographic network, lakes and 206
reservoirs, stream drainage basins and basins upstream of monitoring stations at the na- 207
tional level. Web Map Services (WMS) and Web Feature Services (WFS) are provided to 208
access, query and download the geographic data through the Openhi.net platform, Figure 209
4. A geodata search capability is also provided, reading the geographical names stored in 210
the database and zooming to the relevant boundary. 211

212
Figure 4. Openhi.net platform with geodata. 213

Explain more
The geographical database design and implementation adopts the “Data Specifica- 214
about INSPIRE
tion on Hydrography – Technical Guidelines” of the Directive 2007/2/EC (INSPIRE Di- 215
document, how it
relates here, etc. rective) (https://inspire.ec.europa.eu/id/document/tg/hy). Data stored in the database 216
are following the INSPIRE scheme. 217
Elevations are extracted from the European Digital Elevation Model (Copernicus 218
EU-DEM v. 1.1) at 25-m resolution. Extraction (before editing) of a primary watercourse 219
line is done through the EU-DEM, by applying an upslope 10-km2 contributing-area 220
threshold, following Directive 2000/60/ECC. Other main data sources used are: (a) hy- 221
drographic network, lakes and reservoirs from implementation of the Directive 222
2000/60/ECC, (b) hydrographic network from the implementation of the Directive 223
2007/60/EC (c) hydrographic network from OpenStreetMap (publicly available) and (d) 224
hydrographic networks from scanned maps at the scale ~1:50000. Several models are de- 225
signed and created to run the geographical processes and to perform attribute calcula- 226
tions. Watercourse extraction from the filled EU-DEM, hydrographic network definition 227
and river basin delineation are some of the primary processes in the geomorphologic at- 228
tributes’ calculation. 229
Stored watercourses attributes are segment length and slope, geographical name 230
and stream order. Geographical names are collected using the above-mentioned four 231
sources, applying a buffer of 150 m; the stream order follows the Strahler method. At- 232
tributes stored for lakes and reservoirs are area, elevation and geographical name. Area is 233
calculated at the elevation given by the EU-DEM. Attributes stored for basins (river ba- 234
sins, drainage basins, upstream basins of monitoring stations) are area, mean elevation, 235
mean slope, basin order, main watercourse length and slope, mean CN. The average 236
filled EU-DEM value and average slope value (in percent rise) are stored as mean eleva- 237
tion and mean slope of the basin. CN is computed according to an empirical formula 238
Water 2021, 13, x FOR PEER REVIEW 7 of 14

developed by NTUA (Savvidou et. al. 2018) and stored as a basin’s mean CN. Permea- 239
bility, vegetation and slope are the inputs required by the method, so the related data 240
were collected from hydro-lithological maps at scale 1:50000, the CORINE Land Cover 241
dataset (2018) and the EU-DEM slope. Strahler order is used to define the watercourse 242
order based on a hierarchy of tributaries. A basin’s order is equal to the Strahler order of 243
the main tributary (main watercourse). The group of segments that constitutes the main 244
watercourse is used to calculate the length and slope of the main watercourse as sum of 245
lengths and mean slope, respectively. A hydrographic network has been also created 246
from the watercourse’s links and nodes, as a way to quantify the stream network con- 247
nectivity. For each watercourse link the start and end nodes are recorded in the water- 248
course link attribute table. The whole procedure is based on the relevant INSPIRE design 249
specifications shown in Figure 5. 250

251
Figure 5. Elements, database and relations for the hydro network model (source: INSPIRE Directive). 252

Spatial relationships between features (watercourses, basins, nodes) are defined 253
using relevant and validated topological rules of hydrology. A toolbox with several 254
models was created to perform the geographical processes and to run calculations of 255
spatial and geomorphological attributes, Figure 6. Such models are: watercourse ex- 256
traction, hydrographic network definition, river basin delineation, creation of an IN- 257
SPIRE based geo-DB, data import from various sources and data export to the geo-DB, a 258
naming model, to give name to each segment of the watercourse using the names from 259
the four afore-mentioned hydrographic networks, and a basin’s mean-CN calculation. 260
All the modeling procedures and output datasets are considered at the scale of 1:50000.
CN = Curve Number? 261
The ArcGIS Model Builder environment is used as the main spatial analysis tool. Final 262
editing is performed with the QGIS software by using base-maps of OSM, ESRI, Google 263
and Bing. Furthermore, QGIS is used to make the connection to the ArcGIS geodatabase 264
and to transfer the data to PostGIS (Postgres). Finally, the MapServer software is used for 265
publishing the data to the web through WMS and WFS, to access, query and download 266
the geographic data. 267
Water 2021, 13, x FOR PEER REVIEW 8 of 14

268
Figure 6. OpenhiGis toolbox and models. 269
Modules?
4. Monitoring Networks 270
Three pilot networks were installed in the framework of the project. The positions of 271
the stations are depicted in Figure 7. HCMR operates the water-quality network (Figure 272
7a), NOA operates the water-quantity network (Figure 7b), and UnIo operates the net- 273
work for collecting meteorological data and soil related variables. 274

275
Figure 7. Pilot networks: (a) HCMR, (b) NOA, (c) UnIo. 276

4.1. Water Quality Pilot Network 277

4.1.1. Installation - Site Selection and Monitoring 278
The Institute of Marine Biological Resources and Inland Waters of HCMR has sur- 279
veyed the existing network of automatic water quality stations in Greek rivers and, by 280
applying a GIS-based multi-criteria decision analysis, has identified the monitoring 281
needs, along with the priority areas, for the installation and operation of new automatic 282
telemetric stations (Mentzafou et al., 2019). So far, eight stations have been established in 283
the framework of the HIMIOFoTS project (Figure 7a) that measure water stage and four 284
physicochemical characteristics of water: pH, Temperature (T), Electrical Con- 285
ductivity (EC), and Dissolved Oxygen (DO). These characteristics are directly 286
measurable through a single instrument assembly, with the respective sensors attached 287
 Water 2021, 13, x FOR PEER REVIEW 9 of 14

The stations are installed under bridges; power for the instru-
to it (In-situ, 2020). 288
mentation and the telemetry, via GSM networks, is supplied through a solar panel 289
(Panagopoulos et al., 2021). 290
On the other hand, these four physicochemical quantities are all important for 291
the biogeochemical characteristics of surface water ecosystems (Wagner et al., 292
2006). Water temperature controls the rate of chemical reactions and the biological ac- 293
tivity affecting fish growth and reproduction (Patil et al., 2012). The pH of water is con- 294
trolled by chemical reactions that produce or consume hydrogen ions; there are pH 295
standards for drinking water (WHO, 2017), while deviations from the allowable range 296
due to pollution can damage animals and plants that live in water. The pH is measured 297
with a hydrogen-ion electrode. The EC of water, expressed in microsiemens per centi- 298
meter (μS/cm), is a measure of the ability of electric current to pass through water, in- 299
creases with the concentrations of dissolved ions and is considered a key variable to de- 300
termine the suitability of water for irrigation. Finally, DO, typically ranging from 2 to 10 301
mg/L, is important for the survival of aquatic organisms, bacterial activity, photosynthe- 302
sis and availability of nutrients (Wagner et al., 2006). 303
The real-time monitoring of the above variables can thus reveal the starting 304
point of a pollution event and can subsequently affect (a) the decisions of water man- 305
agement authorities regarding drinking and irrigation water, (b) the required adjust- 306
ments of water treatment strategies, and (c) the suitability of a water body for recrea- 307
tional use, or possible actions against adverse effects on aquatic life. Moreover, these 308
common physicochemical characteristics can be used as surrogates for many other con- 309
stituents in water including salinity, sediment, bacteria and nutrients or can be associated 310
with them through regression analyses to further facilitate the evaluation of water quality 311
(Christensen et al., 2002, Wagner et al., 2006). 312

4.1.2. Data Quality Control Algorithms and Assessment of the Operation of Stations 313
Ensuring good data quality of the automatic monitoring network of surface waters 314
begins before the data are recorded, through proper station site survey, systematic site 315
maintenance, and rigorous and regular sensor calibration. Using similar instruments and 316
instrument configurations throughout the network also facilitates comparing data values 317
from sensor to sensor objectively, allowing for efficient troubleshooting. Site visits at each 318
station take place at least twice per year, when onsite measurements with a field refer- 319
ence sonde are compared directly with a station’s measurements. Thus far, deviations of 320
Field Ref Sonde? the latter from reference standards have been lowest for T and highest for DO, since T is 321
measured by simple and very stable sensors with accurate recording under any type of 322
environment while DO sensors are affected strongly by biofouling and suspended sedi- 323
ment deposition. From ~50 comparisons of in-situ measurements and station records in 324
2019 and 2020, the mean deviations for all variables have indicated that the stations 325
tended to underestimate slightly the true values in the rivers. Mean deviations from all 326
sites were negative, with the lowest deviations (< 1%) observed for T and the highest 327
(20%) for DO. A seasonal analysis has also shown that the variability of percentage de- 328
viations for all variables among the four seasons did not differ considerably (Panagop- 329
oulos et al., 2021). 330
The raw timeseries data from the sensors are offloaded to a server at intervals 331
ranging from 10 min to 1 h, and subsequently uploaded in the Openhi.net time-series da- 332
tabase. As raw data might incur error-prone sensor readings due to various reasons, 333
subsequent reliability checks are applied, thus leading to a quality-checked derivative 334
time series. Initially, algorithmic plausibility tests are applied to observations, with 335
pass/fail criteria based on allowable value ranges and variability to detect possibly erro- 336
neous values. In a second stage, suspicious values are specially flagged and examined 337
graphically to determine whether they represent extreme natural phenomena and should 338
be kept in the data series without labels or are of poor quality and must remain flagged. 339
Water 2021, 13, x FOR PEER REVIEW 10 of 14

A range-based test verifies that an observation is within a predetermined range, i.e., 340
that it falls within certain minimum and maximum values. Variability is examined by 341
step and persistence tests. Step tests typically evaluate the change in magnitude between 342
sequential observations and flag values as implausible if this change exceeds a preset 343
maximum value or is lower than a preset minimum one. A persistence test assesses 344
whether multiple consecutive observations vary minimally with time, indicating a pos- 345
sible technical problem. The quality control checks implemented in the water-quality 346
monitoring network in rivers are summarized in the table 1 below. 347

Table 1. Quality tests applied on the data recorded by the HCMR automatic stations. 348

Problem Reliability Check Type of data Definition

empty record
multiple empty records
implausible values
extreme values
(within plausible range of
observation)
extreme differences of values
(within plausible range of
observations)
persistent values
(within plausible range of
observations)
Null test missing data or long period
leave empty records
Gap test with missing data
Range test extreme values min-max preset limits
2.5% smallest & 2.5% largest
Extreme value test extreme values
observations
2.5% smallest & 2.5% largest
differences (absolute) of
Extreme difference test consecutive absolute differences
consecutive pairs of values
of the observations
consecutive differences
Zero-change of the last 48 1-h or
Stuck value test (absolute) of two consecutive
96 (half-hour) recorded values
values

Our experience with the stations operation and data received so far provides evi- 349
dence that in most of the installed instruments the pH and DO sensors require increased 350
maintenance and quality checks compared to the more reliable sensors of water stage, T 351
and EC. It seems that, for periods of several months without sensor calibration, the spe- 352
cific conditions at the installation sites, biofouling and a river’s sedimentation regime 353
largely determine the level of a sensor’s deviation from proper functioning. 354

4.2. The Hydrometric Network 355

4.2.1. Installation of Pilot Network 356
Hydronet is a pilot streamflow monitoring network established and operated by the 357
Institute for Environmental Research and Sustainable Development of the National Ob- 358
servatory of Athens (NOA) within the “Hydronet” project (2018-2021) Hydro-Telemetric 359
Network of Surface Waters: Gauging instruments, smart technologies, installation and operation 360
(Mazi et al. 2021). Hydronet aspires to establish time-series of high-quality data for water 361
resources management and for risk assessment of hydrological extremes, as well as for 362
water-use planning and hydraulic infrastructures design. Sixteen stations (Figure 7b) are 363
currently in operation (10 commercial and 6 designed and constructed by NOA), 6 in 364
Attica and 10 in the Peloponnese, of which 7 in Messinia, two in Laconia and one in Ar- 365
cadia. 366
In the past, NOA has installed and operated two hydro-telemetric networks in four 367
of the above-mentioned basins (TELEFLEUR, 1998-2001; DEUCALION, 2011-2014). That 368
acquired experience has been useful in the design, installation, operation and mainte- 369
nance of Hydronet. The selection of basins has been prioritized based on relevant so- 370
cio-economic and environmental assets and land use (inter alia, the safety of the popula- 371
tion in flood-prone areas). Population density and economic activities (industri- 372
al/tourism/transportation) are relevant for selecting the basins of Kifissos (Athens) and 373
Sarantapotamos in Attica and the basins of Nedon, Selas, and Pamissos-Mavrozoumaina 374
in Messinia. Agricultural activities are important mainly in Evrotas basin, Laconia, but 375
also in the upstream reaches of the Nedon, Selas, and Pamissos-Mavrozoumaina streams 376
in Messinia; the pristine river Loussios, Arcadia is monitored mainly for ecological rea- 377
Water 2021, 13, x FOR PEER REVIEW 11 of 14

sons. Severe floods (also flash floods) have occurred in Attica, particularly in Kifissos 378
basin (Koussis et al., 2001; Mazi and Koussis, 2006), and parts of the city of Kalamata, 379
Messinia were inundated by the overflows of Nedon in the past. Criteria considered in 380
designing Hydronet, to ensure its economically viable, long-term operation, included also 381
the available financial resources, the stations’ proximity to NOA (reduced travel time 382
and costs) and their deployment in geographically adjacent areas (efficient field cam- 383
paigns and station maintenance). 384
Hydraulic suitability criteria were applied in the selection of monitoring sites. The 385
cross-sections, natural or constructed, had to be well-formed –stable and relatively nar- 386
row, ideally at bridges– and easily accessible for observations in field campaigns (set-up 387
of rating curves; see “Discharge determination from surface-velocities”), with conventional 388
current meters and/or modern flow metering devices (ADCP or hand-held radar). The 389
monitoring stations are equipped with water level sensor, modem, thermometer (air) 390
and rain gauge (in some). Measurements of the water stage are taken every 10 min and 391
the data are sent to NOA’s Hydronet server for automatic processing (quality control) 392
and storage and then forwarded to Openhi.net. The operating status of the stations is 393
monitored automatically from Hydronet’s server, twice a day. Data quality control is 394
along similar lines and procedures used in HCMR’s water quality network; therefore, it 395
will not be elaborated here. To ascertain resilience against vandalism, the equipment is 396
installed out of sight, or in difficult-to-reach places, preferably under a bridge, but far 397
from the water and debris, to avoid destruction during floods. Solar panels, receiving 398
enough sunlight, keep a station’s battery recharged and telecommunications working, 399
thus ensuring continuous operation of an unattended hydro-telemetric station. 400

4.2.2. Development of an Inexpensive Hydro-Telemetric Station 401

To reduce expenditures for hydro-telemetric equipment, a ‘prototype’ hy- 402
dro-telemetry system was designed and constructed. That system combines cus- 403
tom-built firmware and intelligent sensing technologies with telecommunication at low 404
cost, about 50% of the price of a comparable commercial station. This prototype, Figure 405
8, is equipped with an ultrasonic sensor for measuring stage, an air thermometer, a 406
GPRS modem, a camera and a data logger (it can also receive input from a rain gauge), 407
and is powered by a solar panel; data and photos are transmitted to NOA’s server via 408
mobile internet. The system’s additional advantages are flexibility in programming, low 409
maintenance cost, and the ability to extend its capabilities with additional sensors (e.g., it 410
is already compatible with water quality sensors and a video camera). Six stations have 411
been produced and have been installed at stream-sites in Attica and in Evrotas, Laconia. 412
413
Water 2021, 13, x FOR PEER REVIEW 12 of 14

414
Figure 8. Hydrometric monitoring station, equipped with camera, at Evrotas river. 415

4.3. The Agrometeorological Network 416

Modern irrigation management techniques require extensive data concerning the 417
soil, plant, and atmosphere continuum in fine spatial and temporal scales. The objective 418
of those techniques is to estimate the water available for the irrigation of plants. Drivers 419
are rainfall and evapotranspiration, representing the water supply and water removal, 420
respectively. Malamos et al. (2016) presented a method for estimating the irrigation water 421
requirements and for irrigation scheduling based on agro-meteorological data, with 422
site-specific crop and soil-water data for the plain of Arta in Epirus region 423
(https://irmasys.eu/, released in 2015 and since then operational). This system utilizes in- 424
formation from a network of 18 telemetric agro-meteorological stations that were in- 425
stalled during the last 6 years by the University of Ioannina in 8 plains in the Region of 426
Epirus with significant agricultural productivity, i.e. Ioannina, Arta, Preveza, Thespro- 427
tiko, Kanalaki, Kestrini, Kalpaki and Konitsa. These stations were installed according to 428
WMO requirements and FAO paper 56 (Allen et al., 1998). 429
The network of stations across the region of Epirus is depicted in Figure 7c. The ob- 430
jective was to cover as many Land Reclamation Organizations as possible. All stations 431
have modular design, to support different types of sensors. At each station are installed 432
sensors for wind speed and wind direction, rainfall, air temperature, relative humidity 433
and solar radiation. Six stations are equipped with additional sensors for soil moisture 434
and soil temperature at three different depths. Data are transmitted by UHF or GPRS to 435
the communications center and are freely provided in near-real time by the Openhi.net 436
online time-series database: (https://system.openhi.net). This network of agrometeoro- 437
logical stations constitutes valuable infrastructure for the Region of Epirus, providing 438
near-real time data for most of the cultivated areas. 439

5. Innovations 440
5.1. Discharge Estimation in Open Channels From Surface Velocity Measurements 441
The discharge through a cross-section of a stream is estimated indirectly for opera- 442
tional purposes, by converting observed stages h to flows Q via rating curves Q(h). Rating 443
curves at monitoring stations are established, in principle, by measuring and integrating 444
the velocity field over a cross-section of known bathymetry for a range of flows. How- 445
ever, there are practical issues involved. Point-sampling of the velocity field with a cur- 446
rent meter is tedious and often dangerous (prohibitively so in high flows). On the other 447
hand, an Acoustic Doppler Current Profiler (ADCP), which measures velocities based on 448
the frequency shift of a transmitted signal scattered back to the transceiver from particles 449
Water 2021, 13, x FOR PEER REVIEW 13 of 14

in the water, is too expensive for routine use in small streams that may be additionally 450
difficult to access. Velocities can also be measured with a portable ADCP by wading in 451
the stream, but the hydrographer is again exposed to risks. 452
What does The alternative of estimating the mean velocity V as a fraction fv of the surface ve- 453
literature say
locity surf, V = fv surf, has thus gained attention, due to its safe and low-cost operations; 454
about efficacy
image velocimetry (e.g. Rozos et al., 2020) and hand-held radar allow carrying out such 455
of deriving
observations safely and rapidly. Discharge estimation from surface velocities would be 456
mean velocities
particularly useful in the study of small basins often encountered in places along the 457
from the
northern rim of the Mediterranean, e.g. in parts of Greece or in Liguria region, Italy, 458
surface
where the steep terrain and short response times cause dangerous floods. However, ap- 459
velocity?
plying the rule of thumb V = 0.86 surf enables estimating the flow only roughly, because, 460
by hydraulic reasoning, a constant fv is an approximation. The study of the variation of 461
the fv-ratio ascertains that fv varies with the distribution of the velocity, expressed by the 462
momentum distribution factor = /V2 (Koussis et al., submitted). The small varia- 463
tion of fv ( varies little in turbulent profiles, 1 < 1.15) contains the errors in estimating 464
the flow rate, so a method was developed in which fv( ), with estimated based on the 465
macro-geometry and the bed roughness. Testing that method in two shallow streams 466
(depth/width ≈15) in the Peloponnese, Greece it was found capable of determining the 467
discharge within 5%. To further support a hydrographer’s judgement, we are currently 468
studying the underpinnings of the fv-ratio hydromechanically; the results of that study 469
will be reported in a separate publication. 470

These don’t sound

like conclusions.. 6. Conclusions 471

seem to be a rehash Various public and private entities are involved in surface-water monitoring in 472
of some of the Greece; however, data from various sources do not follow common acquisition and pro- 473
above info. cess procedures. A pioneer effort to unify the fragmented hydro-meteorological infor- 474
Conclusions should mation was attempted during 1990s with the creation of Hydroscope, a national hy- 475
be “emergent” dro-meteorological database. Today, Hydroscope is maintained by the Special Secretariat 476
thoughts based on for Water but does not support real-time data and includes a small portion of the availa- 477
the above (is this ble information in Greece. On the other hand, several telemetric networks collect sur- 478
feasible? how useful face-water data in Greece operating independently from each other. 479
is it according to all The Openhi.net infrastructure was created for hosting existing telemetric networks 480

the different users? and databases, in order to provide free access to real-time data and monitoring facilities. 481

Etc..) A notification system was also developed, and geographical data are provided. Addi- 482
tionally, a discharge estimation method was advanced and low-cost telemetric equip- 483
ment was produced. In Openhi.net, partners develop and manage their own DBs, ac- 484
cording to common agreed-upon standards, and provide the information to a common, 485
publicly accessible DB. 486
A national monitoring infrastructure is a multidisciplinary, synergistic effort by 487
various stakeholders. Universities, research centers, public services, the private sector, 488
municipalities, even the public are involved in the production and dissemination of sur- 489
face-water information. Today, the Openhi.net web platform includes three pilot net- 490
works that monitor water quantity-quality parameters and meteorological variables. The 491
applications for station and data management permit the incorporation of new networks, 492
with main purpose the open access to real-time data. The synergy between existing 493
networks, through the infrastructure, provides more effective management and distri- 494
bution of the information but also more efficient design of the future network. 495

Funding: This research was funded by the Open Hydrosystem Information Network (OpenHi.net) 496
research project, which received funding from the Special Secretary of Operational Programme 497
“Competitiveness, entrepreneurship and innovation” 498

499
Water 2021, 13, x FOR PEER REVIEW 14 of 14

References 500
1. Allen, R.G., L.S. Pereira, D. Raes, M. Smith et al. 1998. Crop evapotranspiration-Guidelines for computing crop water 501
requirements-FAO Irrigation and drainage paper 56. FAO, Rome 300, D05109. 502
2. Christensen, V.G.; P.P. Rasmussen, and A.C. Ziegler. 2002. Real-time water quality monitoring and regression analysis to 503
estimate nutrient and bacteria concentrations in Kansas streams. Water Sci. Technol. 45, 205–219, doi:10.2166/wst.2002.0240. 504
3. Efstratiadis, A., N. Mamassis, A. Koukouvinos, K. Mazi, E. Dimitriou, and D. Koutsoyiannis, Strategic plan for establishing a 505
national-scale hydrometric network in Greece: challenges and perspectives, European Geosciences Union General Assembly 506
2019, Geophysical Research Abstracts, Vol. 21, Vienna, EGU2019-16714, European Geosciences Union, 2019. 507
4. Efstratiadis, A., N. Mamassis, . Koukouvinos, D. Koutsoyiannis, K. Mazi, A. D. Koussis, S. Lykoudis, E. Demetriou, N. 508
Malamos, A. Christofides, and D. Kalogeras, Open Hydrosystem Information Network: Greece’s new research infrastructure 509
for water, European Geosciences Union General Assembly 2020, Geophysical Research Abstracts, Vol. 22, Vienna, 510
EGU2020-4164, doi:10.5194/egusphere-egu2020-4164, 2020. 511
5. In-Situ Inc. In-Situ Aqua TROLL 400 multiparameter probe spec sheet Jan. 2020; https://in-situ.com/ (accessed November 1, 2020). 512
6. Koussis, A. D., K. Lagouvardos, K. Mazi, V. Kotroni, D. Sitzmann, J. Lang, H. Zaiss, A. Buzzi and P. Malguzzi. 2003. Flood 513
forecasts for an urban basin with integrated hydro-meteorological model, J. Hydrologic Engineering, 8(1): 1-11. 514
7. Koussis A.D. et al. (submitted). Discharge determination from surface velocities: an assessment of maximum entropy-based 515
methods. 516
8. Koutsoyiannis D., G. Tsakalias, A. Christofides, A. Manetas, A. Sakellariou, R. Mavrodimou, N. Papakostas, N. Mamassis, I. 517
Nalbantis, and Th. Xanthopoulos, HYDROSCOPE: Creation of a national data bank of hydrological and meteorological 518
information, Research and Technology Days '95, National Technical University of Athens, 1995. http://www.itia.ntua.gr/94/ 519
9. Malamos, N., I.L. Tsirogiannis, and A. Christofides. 2016. Modelling irrigation management services: the IRMA_SYS case. Int. 520
J. Sustain. Agric. Manag. Informatics 2, 1–18. https://doi.org/10.1504/IJSAMI.2016.077264 521
10. Mamassis N., E. Tiligadas, D. Koutsoyiannis, M. Salahoris, G. Karavokiros, S. Mihas, K. Noutsopoulos, A. Christofides, S. 522
Kozanis, A. Efstratiadis, E. Rozos, and L. Bensasson. 2010. HYDROSCOPE: National Databank for Hydrological, 523
Meteorological and Geographical Information, Towards a rational handling of current water resource problems: Utilizing 524
Data and Informatics for Information, Hilton Hotel, Athens, http://www.itia.ntua.gr/968/ 525
11. Mazi, K. and A.D. Koussis. 2006. The July 8, 2002 storm over Athens: Analysis of the Kifissos River/Canal overflows, 526
Advances in Geosciences, 7: 301-306. 527
12. Mazi, K., A.D. Koussis, S. Lykoudis, G. Vitantzakis, P. Dimitriadis, N. Kappos, B. Psiloglou, D. Katsanos, I. Koletsis, E.Rozos 528
and T. Kopania, HYDRO-NET: Hydro-telemetric Network for surface waters – Innovations and Prospects, EGU General 529
Assembly 2021, online, 19–30 April 2021, EGU21-14956, https://doi.org/10.5194/egusphere-egu21-14956, 2021. 530
13. Mentzafou, A., Υ. Panagopoulos, Ε. Dimitriou, 2019. Designing the National Network for Automatic Monitoring of Water 531
Quality Parameters in Greece. Water 11 (6), 1310 https://doi.org/10.3390/w11061310. 532
14. Panagopoulos, Y., A. Konstantinidou, K. Lazogiannis, A. Papadopoulos, E. Dimitriou, 2021. A New Automatic Monitoring 533
Network of Surface Waters in Greece: Preliminary Data Quality Checks and Visualization. Hydrology 8, 33 534
https://doi.org/10.3390/hydrology8010033. 535
15. Patil, P.N., D.V. Sawant and R.N. Deshmukh. 2012. Physico-chemical parameters for testing of water – A review. Int. J. Environ. 536
Sci. 3, 1194–1207. 537
16. Rozos E., P. Dimitriadis, K. Mazi, S. Lykoudis and A. Koussis. 2020. On the Uncertainty of the Image Velocimetry Method 538
Parameters, Hydrology. Details ? 539
17. Savvidou E., A. Efstratiadis, A. D. Koussis, A. Koukouvinos, and D. Skarlatos, The curve number concept as a driver for de- 540
lineating hydrological response units, Water, 10 (2), 194, doi:10.3390/w10020194, 2018. 541
18. Tauro, F., J. Selker, N. van de Giesen, T. Abrate, R. Uijlenhoet, M. Porfiri, S. Manfreda, K. Caylor, T. Moramarco, J. Benveniste, 542
G. Ciraolo, L. Estes, A. Domeneghetti, M.T. Perks, C. Corbari, E. Rabiei, G. Ravazzani, H. Bogena, A. Harfouche, L. Brocca, A. 543
Maltese, A. Wickert, A. Tarpanelli, S. Good, J. M. Lopez Alcala, A. Petroselli, C. Cudennec, T. Blume, R. Hut and S. Grimaldi. 544
2018. Measurements and Observations in the XXI century (MOXXI): innovation and multidisciplinarity to sense the hydro- 545
logical cycle, Hydrological Sciences Journal, 63:2, 169-196, DOI: 10.1080/02626667.2017.1420191 546
19. Wagner, R.J., R.W. Boulger, Jr., C.J. Oblinger and B.A. Smith. 2006. Guidelines and standard procedures for continuous 547
water-quality monitors-Station operation, record computation and data reporting: U.S. Geological Survey. Tech. Methods, 1–D3, 548
51, doi:10.3133/tm1D3. 549
20. WHO. 2017. World Health Organization Guidelines for Drinking-Water Quality, Incorporating the First Addendum, Licence: CC 550
BY-NC-SA 3.0 IGO, 4th ed.; WHO: Geneva, Switzerland. 551
552