Evaluating Probability Distribution Functions for the Standardized Precipitation Evapotranspiration Index over Ethiopia

Round 1

Reviewer 1 Report

This study evaluated the performances of three candidate probability distribution functions for calculating SPEI over Ethiopia, and compared the differences between SPEI and SPAEI. The differences of varied data inputs on SPEI values were evaluated. I have several questions regarding the methods and conclusions for evaluating the effects of probability distribution functions and data selection on SPEI, please see my specific comments below:

1. The three candidate probability distribution functions for calculating SPEI were PE3, GEV, and Genlog, why not include the log-logistic distribution that the original index adopted?
2. The temperature data used for calculating SPEI were from ERA5, have the authors evaluated the accuracy of this product, or are there any references on the performance evaluation of this product?
3. Likewise, the actual evapotranspiration data were from the GLEAM, and the accuracy of this product needs to be evaluated first before drawing the major conclusion.
4. Line 260 states that PE3 fails for 43% of the gauging stations at a 1-month time scale, which indicates that PE3 is not a well candidate probability function in this study area, and why choose the log-logistic distribution instead?
5. Line 309: the current images in Figure 3 did not show clearly the SPEI series derived from different probability functions.
6. Line 311: The reason for the varied performances of probability functions at different time scales needs to be discussed.
7. Line 3.2.3: The authors named the title of this chapter as “Comparison based on drought events”. The authors can select several typical drought events, and provide a spatial image to show the performance of SPEI series derived from different probability functions in monitoring drought events.
8. Likewise, the effects of time scale on NSE performances need to be discussed.
9. Since the authors use the Thornthwaite equation to estimate PET and therefore for SPEI computation, a number of previous studies claimed that the role of temperature would be overestimated, particular for dry areas, the influences of using Thornthwaite equation on the derived conclusions need to be discussed. Particularly, the authors stated that there are significant differences between SPAEI and SPEI, would this difference be reduced if using Penman equation for estimating PET?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Introduction: It is necessary to insert citation in the third paragraph in the sentence: Furthermore, drought indices are useful for identifying the intensity and duration of drought. Suggestion: Furthermore, drought indices are useful for identifying the intensity and duration of drought (da Rocha Júnior et al., 2020). Citation: https://www.mdpi.com/2076-3263/10/4/135

Correct the mentions made to references 8 and 10, it is the same, that is, in the place of reference [10] put [8].

The statement that 17 years of data are sufficient to assess droughts is very important, relative to reference [20].

In the caption of Figure 1, add: Figure 1. Elevation map for the study area together with the location of the 125 gauging stations used in this study.

The last two paragraphs of topic 2.2 need to be better connected. I believe that the penultimate paragraph aims to inform that the ERA5 data for temperature were used to estimate the ETP and thus, together with the precipitation, also obtain the real evapotranspiration, for later calculation of the SPAEI. Therefore, I realize that a seasonal precipitation dataset was used, with gaps filled in by data from the CHIRPS project, but the temperature was not obtained from stations, just from the ERA5 reanalysis. This needs to be better described, as the authors claim that this is the most recent reanalysis of the ECMWF, but we have the ERA5Land version, even more recent and with better spatial and temporal resolutions (time data since 1950):

https://cds .climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-land?tab=overview

What interpolation method is used to extract data from ERA5 to station points? Closest grid point? Simple bilinear?

Topic 2.3 cites methods of estimating potential evapotranspiration, but it is not clear to the reader which one they used, and this needs to be clear in this topic. Was it the Thornthwaite method?

Topic 2.4 makes an important statement, but it is very succinct. I suggest that the authors better detail and reference the microwave estimation of vegetation.

The results are well presented and discussed. It makes an important contribution to the assessment of droughts in shorter time series.

Figure 2 needs to have the resolution improved, it's clearly visible but it appears blurry.

When checking the supplementary material, which is not very extensive, do I question the authors whether such results should not be presented in the full article? I suggest this.

Finally, I congratulate the authors for the work, I would like to see it published after these small suggested revisions.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

I am satisfied with the authors' answers and corrections.

Author Response

Thank you for your review.