Remote sensing has been increasingly used to map and classify cropland, and the accuracy has been improving as remote sensing technology, computing power, and statistical methods continue to advance. To answer your question, I think it depends on how you define "sufficient" (funding is almost always insufficient, but that seems to be another topic). In my opinion, It will be difficult to show that remote sensing is "insufficient" to map global crop yield. That said, just like any remote sensing studies, there are limitations. For example, Massey et al. (2018) listed some limitations to their cropland classification study (doi: 10.1016/j.rse.2018.10.013) :

"The observed limitations in the North American cropland extent for the nominal year 2010 are largely due to the data availability across the North American continent (Fig. 3). Regions of significant cloud cover, such as the Central American countries, can have large variability in the Landsat 5 TM and Landsat 7 ETM+ image collection, which caused errors in the RF classification. Lastly, regions with relatively small field sizes, large amount of rainfall, irregular patterns of irrigation, and longer growing periods might have lower accuracies due to highly similar time-series spectral curves between crop and non-crop samples. Many countries in the Central American regions have significant time gaps between their agricultural census, leading to limited validation opportunities."

In short, even though there are remote sensing technologies and methods to accomplish the goal of mapping global crop yield, accuracy varies by regions and is sensitive to data quality. Hope that helps.

Md. Shahinoor Rahman has conducted significant research on this area and also published. You can go through his work and he might guide you to right path.

For me it is difficult to do that job since it depends on the resolution of the satellite sensor to provide the image of such a huge domain. If some one has done it that may be just a rough estimate.

Generally, remote sensing is an estimation technique; but the accuracy of the estimation depend on spatial and spectral resolution of utilized satellite and the comparison with field work.

Thanks

i agree with Laith Alsaadi, and adding this information :

depending of the choice of the type and quality of optical imagery, the choice of channels and their combinaison, the types of sensors and type of satellites (heliosynchon or orbital).

To mapping crop yield globally in general, :

Article Crop yield estimation by satellite remote sensing

and other one for understanding Satellite-Imagery-Based Crop Yield Predictions

http://cs231n.stanford.edu/reports/2017/pdfs/555.pdf

Through remote sensing land cover, crop density, crop pattern, can be determined, but how can the crop yield be estimated by remote sensing. I think crop density or crop growth is different from crop yield. As mentioned by Dr. Laith, the accuracy of the interpretation depends not only on the spectral resolution but also on the ground truth validation.

The regional estimation can be done to a greater extent by integrating the remote sensing and simulation models but attempting it at global scale is a myth. I did not find any such reference.

Thanks Ho Yi Wan for your feedback. this paper helped me a lot to construct my topic of debate. I will come back to you if anything more is needed. Thanks again.

Thanks Istiak A Bhuyan Som Pal Singh Laith Alsaadi Herve Parmentier for your valuable opinions. I am still looking for references on this issue. I would be glad to have your support.

From the conceptual point of view I agree with Ho Yi Wan. From the operation point of view ( that is my field), you need to standarize procedures at the level of the quality acepted product. You also need to automatize those procedures, have a good data exploration method and good computing infastructure. You need to much expertise at the level of remote sensing, processing, programming and computing.

Cheers

Thanks Gabriel Asato. I would be happy to have some reference study from you.

Answering the question you did 2 steps before

Conference Paper Atlas Technology, a Digital Framework for Efficient Manageme...

The product was the development of a satellite atlas of my country in a short time.

The aim of this paper do not focus directly in your problem, but can give you and idea about the complexity of handling a big amount of satellite images using standard procedures. For that work I am not only develped specific programs but also digital management techniques and procedures.

After the development of this project I had the idea to do the same for mineral spectral index maps for all my country. Despite I develop the technite to certain level I didn't have enough time and resources to do that.

For doing what do you want to do you need at least:

1) a complete coverage of satellite images of the world.

2) a place where you can put such amount of data

3) A decent algorithm for correcting atmospheric influence and ilumination conditions in a way that can be implemented automatically near without human intervention

4) A decent algortihm for calculating the yield croping

5) A method that helps you to evalute the results rapidly

6) A system that can process such amount of data in a expected time.

7) A method for presenting the data

8) A method for product validation

I hope this workflow will give a better idea of the problems involved in the task you want to do.

Best lucks

Gabriel

By the way, if you live in Germany you can also ask to my friend Prof. Peter Baumann (Bremen Univ). He developed a special system for the management of large raster (coverages is his sense) datasets. I think any high demanding data processing can be done under his system: Rasdaman gmbh.

Cheers

Gabriel

Today I found this Plugin for QGIS. I am not sure if it can work for your own purposes.

agknow for QGIS

Plugin for using the ag|knowledge REST API from geo|cledian

This plugin allows the usage of the remote sensing API ag|knowledge from geo|cledian.com. The API provides a field monitoring service that allows to monitor any agricultural field worldwide with a variety of vegetation indexes, crop parameters, times series analysis and comparison features based on satellite data.

Features of the plugin include the download and analysis of time series of visible and vegetation indexes rasters of registered fields. The data is based on Landsat-8 and Sentinel-2 currently. See https://sites.google.com/site/geocledian for a detailed product overview.

Please note that you will need a registered API key from geocledian.com to use this plugin for your area of interest. The plugin is delivered with a demo key that allows to access some demo parcels around the world for evaluation purposes.

Happy Christmas

It may be useful we need to verify weather it works at global level or not. ...Happy Christmas to all research gate friends. .