My final goal is to find a multiple linear regression to predict actual evapotranspiration, ET (response variable) using remotely sensed land surface temperature and ground-observed air temperature (explanatory variable). All are raster (.tif) format. They are 5-year dataset with the same extent and temporal (16days) and spatial (500m by 500m) resolution.
I was trying to find some ways for spatial regression, raster regression, spatial statistics, etc. but I fail to one that can be applicable to my case. Most ways that I found didn't handle the time series as two explanatory variable at the same time.
Has there been anyone who tried or applied this type of approaches?
You can try this:
https://training.gismentors.eu/grass-gis-irsae-winter-course-2018/units/27.html
According to the above link "GRASS GIS comes with different modules performing regression analysis. These modules work with single raster map or with time series data too."
You can develop your own code using Python or R.
Many packages are available for multilinear regression such as "SCI-KIT Learn".
https://scikit-learn.org/stable/
This packages efficiently work for regression of time-series data also.
I would suggest you to use Python and do the following steps:
1. Read raster file in Python using Rasterio package
2. Convert it into Numpy array by keeping dimension as your time steps
3. Fit the regression model in Numpy Array data
I hope this would solve your problem.
- Badges
- Science topic
- Similar topics
- Correction