I doubt there is any value in using the data you have described in the way you intend.

1) the impact of water deficit is usually dependent of the season in which it occurs.

2) Plants would commonly have roots deeper then 5 cm.

3)35 years is not a long time.

4) wind humidity and temperature also have a influence on the crops response to water deficit.

If you have a good weather data set then why not run a crop model and use modelled yield as the basis of your climate assessment?    

Dear Muhammad,

The answer to your question will depend on the ultimate purpose of your study: for instance, if the main motivation is to explore the impact of drought on vegetation, then it would be much more appropriate to analyze data on time scales and with a time resolution that match the growth and development of plants. That is the main reason for focusing on monthly, or (even better) weekly data.

In any case, it should be clear from the outset that with yearly data, you will not be in a position to reliably document any drought event of less than about 2 years in duration: See the following Wikipedia pages for an introduction to this phenomenon:

https://en.wikipedia.org/wiki/Aliasing

https://en.wikipedia.org/wiki/Nyquist_frequency

Also, since your records are 36 years long, you won't be able to assess slow processes that may take place over periods of 20 or more years (eventual drought cycles on such long periods would appear as trends): the number of events you might actually detect on the basis of yearly data is therefore going to be quite limited.

If you have access to the original satellite data, you might be able to remedy the first restriction, at least in part, as those instruments typically acquire observations at a much higher frequency. Hence, if your application might benefit form a information on a finer time resolution, you should aim at analyzing the original data rather than some annual summaries.

Cheers, Michel.

Dear Michel

Thank you for your valuable information and comments.

I am sorry for not mentioning the purpose of my study, which is to study the impact of drought on vegetation. Moreover, I thought about choosing the annual summaries because the original dataset (ECV CCI soil moisture product) has a lot of missing values, which also limits the application for drought analysis for vegetation. Any further suggestion will be appreciable. Thanks

Regards

Zohaib

Hi Muhammad,

Thanks for clarifying your ultimate purpose. As Jeffrey and I said earlier, you will really need to develop a methodology based on a much higher sampling frequency to address your problem. Of course, each technique or instrument has its own constraints. And you don't control the performance of the satellites... or any other data source, for that matter, unless you acquire your own observations.

The spatial extent of the area of interest to you is also critical: are you monitoring a single field, a region, a nation or the globe? Different instruments yield data with different scales of representativity. For instance, data from the space-based gravity missions can be interpreted in terms of soil water content, but on the scale of sub-continents. At the other extreme, a field measurement of soil moisture may be representative only of a very small area around that site (e.g., a few square meters).

The solution, then, is to acquire multiple data streams, from different instruments, and not only from space. Merging and analyzing data originating from diverse sources is a scientific and technical challenge, of course, and the most appropriate approach would likely be an assimilation model, in this case. You will need a soil map (or even better: maps of soil properties such as thermal and hydraulic conductivity as a function of depth), a description of the spatial distribution of plant species, weather data (temperature, precipitations, wind, etc.), as well as information on agricultural practices such as seed type, planting dates, use of fertilizers, and irrigation scheduling, for instance, as functions of space and time over the domain of interest.

When empirical data are not available, you need to estimate their most likely values, using models. Hence, if the roots of your plants are deeper than 5 cm (as also suggested by Jeffrey), and you only have reliable satellite data for those top few cm, you may need to simulate the likely soil moisture profile over a much larger soil depth, using other--proxy--data sources or calculations based on precipitation, runoff, evaporation, and soil properties, for instance.

Last but not least, you may need to think carefully about your accuracy expectations: your efforts, investments and expenditures should be commensurate with the value of the expected outcomes. There is no point in developing a very complex and costly infrastructure to assimilate many data streams if the goal is to conduct an exploratory study. On the other hand, if the commercial or national benefits of information on drought impact are substantial, then such an approach may be justified, and it should be accompanied by measures, procedures and processes to verify the validity and worthiness of the information delivered: in a most formal context, you may need to demonstrate that the effective accuracy of your method meets or exceeds the expectations of the ultimate user of this information.

Cheers, Michel.

Dear Michel

Thank you so much for such a valuable and comprehensive comments. Can I contact you in future for some more useful discussion for my studies?

Regards

Zohaib

Dear Zohaib,

Sure, you can continue to ask questions... I will likely not be the only one willing to support you. Anyway, I'll help as I can, and remember that I may not have answers to all or even most your questions! ;-) . But I'll also point you to other experts as needed. Cheers, Michel.

Dear Zohaib,

I doubt if annual mean of soil moisture has any logical meaning for drought analysis. You need to define the soil moisture? Moisture available in the surface/ root zone soil is important for growth and yield of crops/ vegetation. Moisture present in deeper soil layer, lying beyond root zone, does not have any use to crop plants/ vegetation. Then drought occurs as an event in the summer. You may encounter severe drought as well as flood in the same year at a location. By averaging of soil moisture over the whole year you may arrive at a normal mean value, not indicating two extreme events (drought and flood) that happened at the place during the year.

Thus I have got doubt about practical meaning/value of using annual mean of soil moisture. I hope you find my input useful.