Okay, good morning to everybody I'm the nearest rigor. Oh, and this is a project coordinated by sub C and with other two partners University of Morata and Colombo Sri Lanka and IST in Pakistan the project

Would like to help in finding a solution a Solution to the left to a lack of monitoring system in developing a low-income countries Due to high cost of software hardware due to in accessibility to local support replacement parts and also due to

the missing of interoperability between data caused by protocols and so on and In this field we have some open Stuff that we that we can use we have the open technology open protocols open standards and this is in here

comes out the research We are developing and integrating open monitoring system based on open hardware like Arduino We are using the sense observation service as standard to retrieve and share data Thanks to the IST to the source software, which is open source and then we also use

Open data portal like second to share reports and Data and so on with this Integrating monitoring system. We would like to know if it could be a solution to this problem. So we

Will we are evaluating? The prototypes that we have developed we have we are deploying we are deploying the network to see if in a real case study these

things works and this is the system we have Mainly three layers. We have the hardware layer, which is basically the weather station Then we have the communication layer, which is about this transmission of the data between the hardware and the service layer

and this last layer is The one who manage the data who collects data or store data and share the information Starting from the hardware layer. We have developed two prototypes The forums mode and the forums PCB

The first one is a modular solution based on common materials and So it's easy to replace parts and things but the negative aspect of this solution is that it needs a lot of quite time to be built and

also the connection on my mobile wires and Also, it needs more space the other solution is the Forest PCB which is based on a PCB the schema is open and can be Can be downloaded and

it is easy to install but also in this case, we have some negative aspects because the components are soldered so it's harder to change them and Also, the PCB is not widely available in the market and this is also a problem and also

It's a little bit more expensive These two solution share the same internal components So in the main box, we have the Arduino as main controller We have the RTC to get the daytime The open log to store the data to have a cup of the data that we are sending

We are using the C major hundred to Transmit the data and then the DHT 11 to monitor the in the internal temperature because we are Deploying the system Sri Lanka where there are high temperatures and then we have another

component which is the step up down to shift up and down the voltages because they are doing it's Seven volts the other components and it's a five volts about the source of the of the

Energy is a battery of 12 volts And also these two solution are we sharing the same external sensors. We are using the MBME 280 for pressure and humidity We are using the BH 1750 to measure light in terms of looks and

Also the DS DS 18 the 20 to measure temperature and then we have other free Sensors, they are engaged. Of course. It's only more sure and the wind speed and direction and Now there is the service layer. We are using East saucer because it's open source

it's a actor it's currently used used by the eye to Manage the hydro meteorological data of the canton Tichino And also because it's SOS OGT compliant

but I would like to invite you to the representation of tomorrow because we have Some news. I guess East sauce freeze Out and so if you want to know more about this software you can go to this presentation and

Finally, there is the communication layer Which is the parts with? transmitted information We need some we have some requirements we want would like some something that With low data consumes because we are using the GPRS connection

We would like something that is fast and we would like something with a sender response pattern to be sure that the data are sent correctly, so we Integrated in the IE saucer a new feature which is called fast insert to response to these needs

this is how we sense we send the data basically, we read the data from a data logger then the GPRS is activated and Then the data will be transmitted for three times

It means that if the first time is successful then the other two are not need but in Sri Lanka There are a lack of connection also and so it's useful to have this kind of Way to transmit data because if the data are successfully sent they are storing a data sensor folder

Otherwise, you know that are not the center and the data to be sent folder And in this way, you can keep track of keep track of the data that we have sent or not and This is the whole system that we are deploying in

Sri Lanka in the Daturaya basin this basin has a catchment area of 2600 square kilometer These are there are More than 2006 millimeters per year and basically the 50% of these total amount of rain

Is received during the monsoon periods so it is an area with heavy rainfall events We have selected

22 locations basically 18 schools and Other and temples Because in this way we can Be sure about also the safety of the of what we are installing the weather station These are some images about the installation process

many students of the University of monitor are involved in and The possibility to install the weather station in in schools elementary primary school Is also useful to make a sort of the education

activity Of course when the system is deployed there is the need to maintain it. So To be something that is easy to use We are using the OD kappa the open data kit which has an OD kappa collect application where there is a form to be filled and to

guide the maintainer on the maintenance stuff that is needed to to the station and Also, there is the OD kappa aggregate which is server side and in this way we can collect all the

activities and made some Perform some statistical analysis on the way that are On the issues that are present in the weather stations

basically every Station has a QR code and the form with the OD kappa call it has six section the first one is to give information about The maintainer the second to evaluate the general status of the station. Then the third section is about the

To check the box if there is some holes That can cause some water inside Then there is the fourth one To perform maintenance activity on the internal box component So you can check if the wires are good if there is some corrosion and so on and then the fifth

Section to check external sensor status. In fact, we are we are having some problem in the corrosion also and The Last one is about the rain gauge calibration that has to be done to keep the data quality

in a certain level but how the prototype works and to be sure that this is something that Is good enough we compare the in Switzerland

The for once mode model with an official weather station of the Canton Stichino The it is called Trevano and it has a cost of around the 4050 euros and the for once mode solution is has a cost of

450 euros and We Are evaluating we have evaluated the temperature the pressure on the humidity in the rain Because the official weather station has only these four parameters

These are the tools that we used to make this comparison for of course We are using a saucer to retrieve data through the SOS standard We are using ORT Python library because it is well integrated to

the SOS Standard and you can and there is also in some statistical analysis some statistical function that can be used and also everything is automatized through Python and some scripts

How we do the comparison we Have 10 minutes aggregate time series then we make a sort of visual comparison and we use the coefficient of a determination by Pearson to check the

goodness of fit between these two time series and then we analyze also Aggregating values daily mean max minimum values through the needs of scatter plots and the probability density functions

And this is the first result about the the air temperature here there are only the Daily mean values of the temperature in the top left There is the graph of the two time series and you can see by a visual comparison that is very well over the overlap

and also the Coefficient of Pearson is very high. So it's it's really good. The mean error is below 0.3 degree of temperature and this is

also And you can see these also through the scatter plot and density probability function it's very interesting the graph in the bottom right because we plot also there is it the The trend of the residuals and we can see that the promo goes to February. We have an increase

Value of the residuals so in a year error and Probably this is due to a different response of the sensor in the Cold or warm season, but we are not sure in these and only

the next Winter will will answer to this Issue Because probably could be also a sensor drift so decreased quality of the measurement of this answer

This is air pressure and also here is the results are very Very good because the mean error is basically under 0.4 Ecto Pascal also the standard deviation is quite good and also here in the

Both on the right you can see that there is this behavior in in this increasing of the error In winter but maybe with the time Here there is the humidity which is the worst

Parameter that we have evaluated but still good because From the scatter plot. You can see that there is this dual behavior. So when values are over 70% the

Humidity sensor of the prototype Under overestimates value Otherwise when the humidity is below the 70% the sensor Under estimate value, so this is there is this dual behavior and also for this parameter

The residual seems to increase During the winter or during time and here there is the most interesting parameter, which is the

Precipitation and here we are using the rain gauge by Davis. So it's appropriate a resolution Because we don't have Other solution open solution we are Building something that is open, but we have to test it

but the results from this rain gauge, which is The most expensive essential that we are using but is quite cheaper because it is $100 are very good Because the error if you compare the total amount of rain Is more or less a 6% and also it's very interesting how the forums

prototype Through the analysis of the data we have we have selected the 14 rain events Including some snow events has the official weather station

so also this kind of comparison in my opinion is Is good to see that It can be a Sensor that that can be used in a real case study so in conclusion the the results are

very good of the comparison of the data The temperature at the pressure on the best parameter that we are measuring But also the humidity but here there is

There are more issues, but probably because of the humidity parameter. It is quite not so easy to measure and Currently we have deployed the 18th station in the Detroit basin. So we are near to start the Real case study. So we are starting where we start to collect the data and

For this For this reason in the future We would like to have some extreme event notification system we would like to send automatically reports that can

Give information about the quality of the data and if the station are working So maybe for daily or weekly weekly reports and also we would like to use The data for hydrological model because in the do right they draw amazing there are

there is a tank system to laminate the flats, but There are no There is no integration with the real-time data with weather data and so on so with the with the detail Which we are collecting it could be possible to use

Hydrological model to to better manage the tank system and also we have to check the These be these behavior of the residual to be sure that there is something related to the sensor or to the

season to the season Yeah, yes because

We have a now to now to build at the station online Yes, if you visit the website you can find it and

Yeah, you can build on your own the station it's Easier the mode solution because the PCB you need the tool to send the schema to A service that can produce to you the PCB any further questions

Then I will have another question so you already produced 18 of these stations and it seems like it's much cheaper than the selected reference

Do you see any business opportunity in mass producing such weather stations in the future? Well, I don't know because The our solution we need to see if it's something that Could can a bet I can can have a life in time

Okay But I don't know. This is an open source project. And so I Don't know You can build a startup on an open source project, yeah, so maybe consider it later

Then thanks again for the presentation The authors of our last presentation will be Alan manga, Fitch Nikka Messner and Mihaela cicada and

The topic will be grassland recognition with the usage of thermal aids