Wheat Yield Estimation from NDVI and Regional Climate Models in Latvia
, Anne Gobin 1,2, Lola Kotova 3, Sara Top 4, Lesley De Cruz 5, Andris Vīksna 6, Svetlana Aniskevich 6, Leonid Bobylev 7, Lars Buntemeyer 3, Steven Caluwaerts 4, Rozemien De Troch 5, Natalia Gnatiuk 7, Rafiq Hamdi 5, Armelle Reca Remedio 3, Abdulla Sakalli 8, Hans Van De Vyver 5, Bert Van Schaeybroeck 5 and Piet Termonia 4,5Round 1
Reviewer 1 Report
Dear editor:
I have reviewed that paper and have several major concerns:
Generally, I think the paper contains two major parts, (1) crop yield estimation and (2) effect of temperature on crop yield, but neither part is fully analyzed. I suggest focusing on the crop yield estimation part because this part is more related to the remote sensing community.
1. section 2.2: it seems the aNDVI is an important predictor for crop yield calculation, and how was the aNDVI calculated?
2. I suggest comparing the aNDVI model crop yield result with other models, analyze the model performances, and then find a model with the best performance for crop yield estimation.
3. section 2.4.3: I think several factors affect the crop yield, such as precipitation. While, if the paper focuses on the effect of temperature on the crop yield, how to control the other factors.
4. The current manuscript has not analyzed the uncertainty of remote sensing derived crop yield and the temperature. But the analysis of uncertainty is essential.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The quality of presentation is high and the present study has merit for publication. However, I have a few questions about the methodological process.
1) What were the reasons for choosing the NDVI threshold of 0.2 and at least ten continuous observations? I believe that fundamentals are important to support the reproduction of the work.
2) The crop yield statistics data are available by month or year?
3) The 10-daily NDVI images by PROBA-V consider the maximum value between the period. Considering that the NDVI may saturate when the leaf area index increase, what do you think about this effect compared with the crop yield statistics data?
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
It is the study about wheat crop yield estimation through NDVI and climatic data from ALARO-0 and REMO Regional Climate Models (RCM). Although it is a good attempt and study conducted at regional scale and many regions are considered to see the relationship of VI with yield.
Good data set collected and analyzed. Although article is also written in a good way but methodology especially the statistical analysis like regression and correlation analysis are too old while there are latest and innovative methods/approaches are being used and available in literature since last decade. Although authors write about dynamic processed based crop models but did not consider as these models has potential to predict the overall system while only one equation of regression is unable to predict the accurate results.
One most importation issue with this study is about the selection of studied parameters, although different regions were considered but soil is totally ignored in this study. Although, soil has major role in yield like weather, soil determines the yield on the basis of water holding capacity and fertility like nitrogen and other nutrients.
VI, like NDVI is also related with wheat crop stages, also need to explain which wheat crop stage for both spring and winter wheat have positive relationship with yield.
What about the accuracy and reliability of the used climate data for test regions is also question and need to add details.
While there are many other specific comments can also bee seen in attached reviewed copy in different section.
Comments for author File: Comments.pdf
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
I am satisfied with the revision.
Author Response
Dear reviewer,
Thanks for your time to review this manuscript.
Best regards,
The authors.
Reviewer 3 Report
Revised version looks improved.
Can authors give more detailed like equation 1 ( linear regression model) for as random forest regression? Although authors wrote three lines (168-170) but i think these are insufficient.
Author Response
Dear,
We added further details on the random forest model applied in this manuscript. See lines 169-173 and lines 200-201.
Best regards,
The authors.
