Hi,
The aim of the study is to calibrate hydrological model (SWAT model in this case) using remotely sensed actual evapotranspiration (ET) at a daily basis. Consequently, autocalibration requires daily remotely sensed ET. Unfortunately, the MOD16 estimates of ET are at a level of total 8-day, and cannot be directly used in autocalibration.
In this paper, for a similar problem, authors used NLDAS estimates of daily actual ET to disaggregate MODIS 8-day ET using a temporal scaling factor. According to the authors this method can be valid since the NLDAS estimates capture the daily variation of ET well. https://www.sciencedirect.com/science/article/pii/S0022169418307856
In my region, such data are not available freely. I have only MODIS 8-day ET, and daily observed meteorological data (precipitation and temperature, wind speed, humidity, solar radiation). I don't know if MODIS 8-day ET can be disaggregated to daily using other information such as precipitation and temperature?
Thank you very much
Suppose you can, it will add more uncertainties which for sure limits achieving your target.
I agree. It will add more uncertainties. GLEAM 3.5b provides remotely sensed actual evapotranspiration at daily scale. It has been evaluated and tested all over the world but still needs a verification step against measurement before its use.
You can try Daily PET from USGS (https://earlywarning.usgs.gov/fews/product/81) and convert it to weekly estimates and statistically model relating them to weekly values of MODIS. This function can be used to interpolate to get daily estimates. PET of USGS uses the Pennman-Montieth equation as well.
You can use your daily meteorological data to estimate actual ET (=Kc*ET0), and compare it (after aggregating at 8 day) with MODIS ET. You can see if any statistical relationship exists and disaggregate accordingly if you need to use the MODIS ET product only.
Dear Soufiane Taia
Smaranika Mahapatra and Timothy O Randhir have provided, I believe, the best options for you. You start doing it. Once any issues about uncertainties arise, we can discuss them further here.
Hi Soufiane,
Have you tried EEFlux" (Earth Engine Evapotranspiration Flux)? EEFlux is based on the operational surface energy balance model "METRIC" (Mapping ET at high Resolution with Internalized Calibration) and can provide you Landsat based daily Actual ET obtained from surface energy balance approach. There is obviously difference in spatial resolution of Landsat and MODIS. I am attaching the relevant publication for reference.
Kindly check on this paper
https://www.mdpi.com/2072-4292/12/9/1433/htm
If you want to accurately quantitative comparison, it is recommended to average the daily value data into every eight days, and then compare.
Using a long time series with an interval of 8 days, it does contain spectral components of 1 to 8 days.
Which requires careful harmonic, fourier, or wavelet analysis, to generate a fine set of time-frequency structure parameters, and then reconstruct the daily time series.
Even so, there is no significant difference between the above two comparison results.
I suggest you to calibrate the SWAT model at monthly time scale. Because to disaggregate or aggregate data and then use it for calibration that will add more uncertainty to the calibrated parameters and consequently increase the error in the output of the model
- Badges
- Science topic
- Similar topics
- Geoscience
- Atmospheric Sciences
- Climate Science
- Climate Change