I am Dr. Partha Roy from India. I am utilizing the NCEP/CPC L3 Half Hourly 4km Global (60S - 60N) merged IR V1 data for my scientific study. It is really very useful data.
But, I have one query about this data. In the original data file, so many NaN values are coming with real values. I could not get why it is there. I have attached one figure (white color represents NaN values). Please go through the figure.
It would be very helpful if you kindly let me know the reason for this.
National Atmospheric Research Laboratory
Dept. of Space, Govt. of India
- IR data plot over South Asia region
- plot.jpg (32.47 KiB) Not viewed yet
My colleague who would be the best to address this question is out of the office for a few more days. I have sent your message to him and I will post his reply after he has considered it.
The missing values in this dataset are declared as "_FillValue
= -9999." in the local attributes to the "Tb" variable, in accordance with the Climate and Forecast (CF) conventions. The CF-compliant tools will honor this attribute and will automatically place NaN where fill values are encountered.
In case you write your own program, you must carefully observe this local attribute and manage the pixels with fill value in the data, as declared in the attribute, in any suitable way that does not allow these pixels to enter your calculations. Missing values in this dataset result for different reasons, one possible reason being that corrections for erroneously cold pixels off-nadir fail for the satellites that are available for merging into the particular 30-minute interval.
https://www.cpc.ncep.noaa.gov/products/ ... _res.shtml
Please read through the documents to better understand the dataset.