Hi,

if you have only the aggregated day data (that is the daily mean temperature), you have to work with them. From my opinion it is unaccurate (and impossible) to try to "generate" the hourly temepratures from the daily mean, if you hav NO INFORMATION about the real hourly measurements. To model something based on such "converted daily value to hourly value" would not be correct.

See:

daily mean = 10

possible hourly data = value (5) measured 12-times and value (+15) measured 12-times

possible hourly data = value (-5) measured 12-times and value (+25) measured 12-times

possible hourly data = value (0) measured 14-times and value (+24) measured 10-times

and many other combinations...

I concur with Veronika.... I dont think thats possible

You don't do that. If daily data is what you have, then daily data is what you use. Converting daily data to hourly is not a valid practice.

AMIGO Verhaegen Yoni :

ES VIRTUALMENTE IMPOSIBLE REALIZAR LA OBTENCIÓN DE DATOS HORARIOS A PARTIR DE DATOS DIARIOS, BIEN SEA DE TEMPERATURAS O DE CUALQUIER OTRO ELEMENTO METEOROLÓGICO.

EN CUAQUIER CASO, UD. PODRÍA OBTENER ECUACIONES DE CORRELACIÓN PARA UN PUNTO ESPECÍFICO SI TIENE TANTO LOS DATOS HORARIOS COMO DIARIOS, PERO LA MISMA LE SERVIRÍA SOLAMENTE PARA, EN CASO DE EXISTIR ALGÚN DATOS HORARIO FALTANTE ESTIMARLO.

Y TAL COMO LE DIJO LA DRA. Veronika Bojdova, YO LE RECOMENDARÍA TRABAJAR CON LOS DATOS MEDIOS DIARIOS, CON LOS CUALES PUEDE OBTENER BUENAS CONCLUSIONES SI TRABAJA CORRECTAMENTE CON ELLOS.

GRACIAS POR SU TIEMPO.

It is a difficult question. I do not think if there is a solution.

Do not do it! There is no reviewer, who will pass thi. Work with daily means. This is the information you have. No more. More will be your imagination only.s

Silly question!

There are no silly questions...

Dear Verhegena;

Although there is no detailed method for your goal, but this methods could be useful; We have two different kind of methods:

i) Temperature Based Methods; and ii) Complex Methods which need different kinds of meteorological data.

For the first group it could be named:

1. Sanders, C. G. (1975). Comments on the model for estimating the completion of rest for “Red Haven” and “Elberta” peach trees. Hortic Sci, 10, 560-561.

2. De Wit, C. T. (1978). Simulation of assimilation, respiration and transpiration of crops.

3. Porter, C. H., Pickering, N. B., Jones, J. W., & Hoogenboom, G. (2000). Weather module in DSSAT v. 4.0 documentation and source code listing. Agricultural and Biological Engineering Department Research Report, 1203.

4. Parton, W. J., & Logan, J. A. (1981). A model for diurnal variation in soil and air temperature. Agricultural Meteorology, 23, 205-216.

Chicago 5. Wilkerson, G. G., Jones, J. W., Boote, K. J., Ingram, K. T., & Mishoe, J. W. (1983). Modeling soybean growth for crop management. Transactions of the ASAE, 26(1), 63-0073.

6. Campbell, G. S. (1985). Soil physics with BASIC: transport models for soil-plant systems (Vol. 14). Elsevier.

7. Goudriaan, J., & Van Laar, H. H. (2012). Modelling potential crop growth processes: textbook with exercises (Vol. 2). Springer Science & Business Media.

8. Ephrath, J. E., Goudriaan, J., & Marani, A. (1996). Modelling diurnal patterns of air temperature, radiation wind speed and relative humidity by equations from daily characteristics. Agricultural systems, 51(4), 377-393.

and for the second group:

1. Donatelli, M., Bellocchi, G., Habyarimana, E., Bregaglio, S., & Baruth, B. (2010). AirTemperature: extensible software library to generate air temperature data. SRX Computer Science, 2010.

Also some software could generate the data for you.

Now, may I ask you for where do you want the climate data, and for which period of time?!

If you have any more question, I am here.

hope it helps.

Best

@ Gordon R Miller: It's not a silly question. Parametrization of the diurnal cycle of temperature has been proven to work in literature before. It's silly from you to give such an answer, especially when not substantiated.

@ MuhammadReza Tabatabaei: Thank you. I use these data to create a surface energy balance model for a glacier, thus diurnal cycles of e.g. temperature and solar insolation are very important. I have datasets from a weather station close to the glacier with daily averaged temperature data for 1951-2016.

Dear Verhaegen;

Fortunately now I am working on the subject, and I suggest you to have look at The Climate Modelling Primer, by: Kendal McGuffie, Ann Henderson-Sellers.

If you don't have the book, I could send you the pdf.

Hope it helps.

Do you have other information from this area?

Perhaps daily min and max T from other period of time, or from another station close to the study area, that can give you an idea of the diurnal oscilation. From that you can model the curve.

Dear Verhaegen:

The problem are very complex. I recommend you adjust a known mathematical function to the 8 observations and then estimate the value.

Best Regards,

Axel.

@ MuhammadReza Tabatabaei: I don't have the book, so a pdf would be nice. Anyway, thanks for your help!

Dear Verhaegen;

I'm not sure that new researchgate terms of service let me to attach the book or not, but I just attach chapter 3, Energy Balance Model (EBM), and I will send you the book if only I have your email.

Hope it helps.

Best

I agree with all the comments that say this is an impossible task. But if for some reason you are compelled to do it, I offer the following.

First, you need to get from somewhere the minimum and maximum daily temperatures. You need to get the daily range as a percentage of the mean temperature. You may have to assume that the daily range is lognormal. My reason for this is that I have seen datasets where the mean was not in the middle of the range, indicating that the hourly distribution throughout the day is skewed. (The median does not coincide with the mean. If I recall correctly the median was less than the mean.)

The shape of the daily hourly distribution (skewness) may vary during the year owing to the variation in inclination of the Sun and cloudiness. You could search for data from a nearby site at similar latitude and elevation to work out the shape of the daily distribution by time of year.

There are published equations for relating most of the relevant variables.

Somewhere you should be able to find hours of insolation and inclination of the Sun. You may be able to find degree-days from meteorological and agricultural services. These more general variables you can use as controls to ensure that your calculations do not depart from other datasets.

What you will end up with is a reasonable model of daily heating and cooling by time of year based on known relationships with external controls by datasets that constrain your calculations to fit parameters that cannot be derived from your data but have been measured elsewhere.

Without external parameters that define skewness and external controls, there is not much you can do except spend at least one year recording hourly temperatures.

You could try researching the lag in response between insolation and heating / cooling during the day. Problem is that the lag is site specific. It will be different in the desert compared to the seaside. It will be different in a city compared to a rural area with cropland or grassland.

I agree with the comments, this is complex and if you manage to do it the results will be questionable and not suitable for scientific consumption

Hi Yoni,

I think that no mathematical function will yield reliable results. Temperature is influenced by a number of climatic and non-climatic factors that cannot be accounted for correctly in such types of parameterizations.

This a typical interpolation with re-sampling problem. Its solution requires a hypothesis about the temperature variation pattern on higher than sampled frequencies. The simplest solution is a linear interpolation (no information available). The smart solution is a kriging with external drift by observations. If you have some observational record in the relevant conditions then you should go to kriging.

You also need daily maximum and daily minimum temperature data as inputs into some of the existing models for conversion to hourly values. You may need to test many of them to chose the best model for your region of consideration. The available models include: Ephrath model (Ephrath et al., 1996); Parton model; de Wit model; Hirotaka model and Fourier series model.

Dear Verhaegen Yoni

It is already a mean data. That means it is averaged from a set of observations. By averaging, its hourly (even less or greater) behavior is lost. Like all others I am also saying that it is impossible to redistribute again. Please recall the mathematics of averaging. It is summed up!. But many of these AWS will have a raw data file and if it is still there in your case, you can retrieve temperature in original sampling interval. Please check for the availability of raw file. People generally have a tendency to average the atmospheric parameters into daily or hourly. It will kill the minute useful information. I think it is good to let AWS sample air temperature data in 5 minute or 10 minute. After recording by keeping original data some where you can duplicate it and average it whatever scale you need.

Coming back to your need, you cal also check the hourly availability of other derived and other parameters from temperature and other variables like (RH, dew point, solar radiation etc). from this you can estimate temperature values.For this also you need hourly data. If full data set is averaged, no way!.

Dear Verhaegen Yoni

Daily data conversion to hour is not valid.

It's always advisable to convert from higher time-resolution dataset to lower-resolution. Not advisable to go otherwise scientifically.

Conversion from daily to hourly data may not give true picture of what the real observed hourly data was.

If you know Tmax and Tmin, and a profile type

Hubert Mallard Could you give more information? I have max and min and also a hourly TMY data i am trying to convert future daily data to hourly

Daily temperatures profiles have mainly the same shape between max et min temperature of the day, because of the thermal and hygral properties of the place you are... If you have some hourly data, if possible during one year you can identify the profile type of a day of the place between the max and min temperature with a Weibull density function which have 2 parameters, taking account of skewness and kurtosis of the profile due to sun. Snow, clouds change the profile but they are secondary parameters.

Hubert Mallard thank you for the information is there any code and name of the method Which i can immediately apply that?

These methods have been create and explain for example in the study :

https://hal.science/hal-03130349v1

See page 160 to 171 for years and 173 to 177 for days... for your problem

some pages of the report :

- Tome IX : Connaitre et comprendre pour protéger les espèces animales et végétales face au changement climatique

Fanny Mallard

https://www.linkedin.com/in/fanny-mallard-04a4ab49/?originalSubdomain=fr

All the complete reports (biodiversity and climate) are here :

https://technique.sentinelles-climat.org/les-articles-scientifiques/

Hubert Mallard thank you ill check them

but i couldn't find an accurate scientific python code for daily future weather scenario data to convert hourly data using typical hourly weather data (TMY), do you know any useful Python code?

You use Climate Data Operator software. Te first step, you convert data in netcdf, te next step use comand interpoltime

Thank you, ı ll check this file conversion too, I need epw file at the end. Now ı have csv files.