Maize mapper and explorer

[qwu-570]

• decide visualisation scheme by Monday
• QA for MODIS LAI time series plotting
• Aspatial data explorer
• Spatial VI (link with DigitalAfrica team?)
• Instruct GEE account approval with @MarcYin

Priorities:

• legend/palette issue
• apply-mask as default
• axis labels (year, lai, precp, && their units) for TS figures
• upload mofa csv file(s) to GitHub
• add scatterplots as subplots: mofa yield vs precp
• displaying district/county names on click

Targets set on 11-Mar-2022:

• generate histogram per QA level
• generate IO webpage
• Tidy up, fix all bugs
• change rain m to mm
• add 1-to-1 line for scatterplot

Additional feature only if have time:

• all districts, instead of northern (in progress)
• some demo of toolboxes : e.g. upload own vector / raster
• user-defined AOI
• drawing tools: TBD: remove?
• on-click: per-pixel TS explorer
• graph of ALL properties of pixel, on click
• yield estimate to be related: e.g average yield over the masked area
• add scatterplots as subplots: mofa yield vs empirical yield; and show stats of graph

[profLewis]

Hi
so Monday is over and the first of these should be ticked off, surely. Where is the output for this? how can we see the progress?

Also, what is the name of the notebook here?

[profLewis]

why does https://github.com/UCL-EO/Workshop2022/blob/main/notebooks/CAU_Interface_Ghana_V1.ipynb
have pip install geemap?? that should already be here.

[qwu-570]

why does https://github.com/UCL-EO/Workshop2022/blob/main/notebooks/CAU_Interface_Ghana_V1.ipynb have pip install geemap?? that should already be here.

This is CAU's original draft, which I haven't used yet. Ignore this notebook for now.

[qwu-570]

Hi so Monday is over and the first of these should be ticked off, surely. Where is the output for this? how can we see the progress?

Also, what is the name of the notebook here?

Q1_basic_plotting.ipynb is where I put my stuff. In progress.

[profLewis]

What decision did you come to on the graphing? Lewis Professor Lewis, NCEO/UCL

…

________________________________ From: Qingling Wu ***@***.***> Sent: Monday, February 28, 2022 8:23:37 PM To: UCL-EO/Workshop2022 ***@***.***> Cc: Professor Philip Lewis ***@***.***>; Comment ***@***.***> Subject: Re: [UCL-EO/Workshop2022] Maize mapper and explorer (Issue #17) Hi so Monday is over and the first of these should be ticked off, surely. Where is the output for this? how can we see the progress? Also, what is the name of the notebook here? Q1_basic_plotting.ipynb is where I put my stuff. In progress. — Reply to this email directly, view it on GitHub<#17 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAOC3GS7UYUIS46NVHUCKT3U5PKUTANCNFSM5PKJUQ5Q>. You are receiving this because you commented.Message ID: ***@***.***>

[qwu-570]

planning to plot LAI time series as like this exmaple

[profLewis]

Hi - that doesn’t really look like any of the lai trajectories we’ve been looking at before, so I’m not sure how easy it will be to relate to them. What exactly does it show? Some sort of average in space? What do you want people to get out of looking at this plot? What would a simple 2d colour plot look like as an alternative? Professor Lewis, NCEO/UCL

…

________________________________ From: Qingling Wu ***@***.***> Sent: Tuesday, March 1, 2022 8:24:06 PM To: UCL-EO/Workshop2022 ***@***.***> Cc: Professor Philip Lewis ***@***.***>; Comment ***@***.***> Subject: Re: [UCL-EO/Workshop2022] Maize mapper and explorer (Issue #17) [Q1_LAI-3D]<https://user-images.githubusercontent.com/2849020/156243172-8fa9ed98-3a83-4e28-a612-ed2dcc42fa02.png> planning to plot LAI time series as like this exmaple — Reply to this email directly, view it on GitHub<#17 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAOC3GXNK652HHMSK6YOK5TU5Z4GNANCNFSM5PKJUQ5Q>. You are receiving this because you commented.Message ID: ***@***.***>

[qwu-570]

For example, 2006 and 2015 median modis LAI over cropland pixels over Tamale district

[profLewis]

That doesn’t look like a very stable measure. Have you filtered the lai at all, by QA, for example? Lewis Professor Lewis, NCEO/UCL

…

________________________________ From: Qingling Wu ***@***.***> Sent: Tuesday, March 1, 2022 8:33:12 PM To: UCL-EO/Workshop2022 ***@***.***> Cc: Professor Philip Lewis ***@***.***>; Comment ***@***.***> Subject: Re: [UCL-EO/Workshop2022] Maize mapper and explorer (Issue #17) [image]<https://user-images.githubusercontent.com/2849020/156244284-53390309-9f3a-4893-b801-696002ec1ca0.png> [image]<https://user-images.githubusercontent.com/2849020/156244393-0b7cd379-1ac1-4193-8c10-2d553247091a.png> For example, 2006 and 2015 median modis LAI over cropland pixels over Tamale district — Reply to this email directly, view it on GitHub<#17 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAOC3GQFFCSK7OFABFADYNLU5Z5IRANCNFSM5PKJUQ5Q>. You are receiving this because you commented.Message ID: ***@***.***>

[qwu-570]

I can do it tomorrow.

[MarcYin]

Instruction has been sent to Kofi @asarefi

1. Apply for a google account if do not have one
   Google earth engine account are linked to Google account, so you will need to have a google account before being able to sign up for Google earth engine. You can apply for a new google account here: https://accounts.google.com/SignUp
2. Sign up for Google earth engine account here: http://earthengine.google.com/signup

[qwu-570]

added widget into GEE interface to plot basic data layers. Will add MOFA yield and yield estimates from maxlai next.

[profLewis]

Great! - don’t forget to put scales in so the colours can be interpreted Lewis Professor Lewis, NCEO/UCL

…

________________________________ From: Qingling Wu ***@***.***> Sent: Friday, March 4, 2022 7:00:41 PM To: UCL-EO/Workshop2022 ***@***.***> Cc: Professor Philip Lewis ***@***.***>; Comment ***@***.***> Subject: Re: [UCL-EO/Workshop2022] Maize mapper and explorer (Issue #17) added widget into GEE interface to plot basic data layers. Will add MOFA yield and yield estimates from maxlai next. [image]<https://user-images.githubusercontent.com/2849020/156825282-9e37b6fb-cd5a-4bb5-a249-fe560f4d93d4.png> — Reply to this email directly, view it on GitHub<#17 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAOC3GWHGD66QGJE2X6U43DU6JMVTANCNFSM5PKJUQ5Q>. You are receiving this because you commented.Message ID: ***@***.***>

[qwu-570]

Empirical yield estimation layer added:

Interactive maize area map added, with time series plotted:

[profLewis]

Looks great - can you get an average (mean or median) for a district for years since 2000 and plot that on your mofa yield graph? That would connect the two things you have here. You can pre-calculate that and store it. It might be more Informative than the max lai you have there. Since it’s a linear yield model, it should be the same as the mean (or median) max lai (whatever you use?) put through the yield estimate. Can you also have a scatterplot of the data? The time series is useful, but a scatterplot can inform us about strength of relationship Of course, what would be really good here would be to replace the empirical model with the DA model. That should be feasible IRT for a pixel, which would be really interesting to see over multiple years. You could always pre-calculate the fitting and store the result for all pixels. All you’d need to store would be some index into the ensemble members. Then work from there. I thought Hongyuan had generated some ensembles for that? But, get this empirical thing sorted like you have, and connected to the mofa plots (with scatter plot option) and that is the minimum requirement done for this, probably. If you can take me through it tomorrow we can decide on that and what else might be worthwhile in the time available. Lewis Professor Lewis, NCEO/UCL

…

________________________________ From: Qingling Wu ***@***.***> Sent: Sunday, March 6, 2022 7:28:14 PM To: UCL-EO/Workshop2022 ***@***.***> Cc: Professor Philip Lewis ***@***.***>; Comment ***@***.***> Subject: Re: [UCL-EO/Workshop2022] Maize mapper and explorer (Issue #17) Empirical yield estimation layer added: [image]<https://user-images.githubusercontent.com/2849020/156938932-0cd4709b-1ab4-471c-af88-fbb8d78cad9d.png> Interactive maize area map added, with time series plotted: [image]<https://user-images.githubusercontent.com/2849020/156938959-eea3f7be-610e-4914-8fbe-c598dbf2e2f0.png> — Reply to this email directly, view it on GitHub<#17 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAOC3GSNHYWZ72M5SODIS7TU6UBM5ANCNFSM5PKJUQ5Q>. You are receiving this because you commented.Message ID: ***@***.***>

[profLewis]

One important point - have you filtered the lai by qa information? That is *critical* or you will get nonsense out of the product. The best thing is to assign weights from the qa and run a temporal filter over the data - exactly the same as we do in the MSc practical. Professor Lewis, NCEO/UCL

…

________________________________ From: Philip Lewis ***@***.***> Sent: Sunday, March 6, 2022 8:48:07 PM To: UCL-EO/Workshop2022 ***@***.***>; UCL-EO/Workshop2022 ***@***.***> Cc: Comment ***@***.***> Subject: Re: [UCL-EO/Workshop2022] Maize mapper and explorer (Issue #17) Looks great - can you get an average (mean or median) for a district for years since 2000 and plot that on your mofa yield graph? That would connect the two things you have here. You can pre-calculate that and store it. It might be more Informative than the max lai you have there. Since it’s a linear yield model, it should be the same as the mean (or median) max lai (whatever you use?) put through the yield estimate. Can you also have a scatterplot of the data? The time series is useful, but a scatterplot can inform us about strength of relationship Of course, what would be really good here would be to replace the empirical model with the DA model. That should be feasible IRT for a pixel, which would be really interesting to see over multiple years. You could always pre-calculate the fitting and store the result for all pixels. All you’d need to store would be some index into the ensemble members. Then work from there. I thought Hongyuan had generated some ensembles for that? But, get this empirical thing sorted like you have, and connected to the mofa plots (with scatter plot option) and that is the minimum requirement done for this, probably. If you can take me through it tomorrow we can decide on that and what else might be worthwhile in the time available. Lewis Professor Lewis, NCEO/UCL

________________________________ From: Qingling Wu ***@***.***> Sent: Sunday, March 6, 2022 7:28:14 PM To: UCL-EO/Workshop2022 ***@***.***> Cc: Professor Philip Lewis ***@***.***>; Comment ***@***.***> Subject: Re: [UCL-EO/Workshop2022] Maize mapper and explorer (Issue #17) Empirical yield estimation layer added: [image]<https://user-images.githubusercontent.com/2849020/156938932-0cd4709b-1ab4-471c-af88-fbb8d78cad9d.png> Interactive maize area map added, with time series plotted: [image]<https://user-images.githubusercontent.com/2849020/156938959-eea3f7be-610e-4914-8fbe-c598dbf2e2f0.png> — Reply to this email directly, view it on GitHub<#17 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAOC3GSNHYWZ72M5SODIS7TU6UBM5ANCNFSM5PKJUQ5Q>. You are receiving this because you commented.Message ID: ***@***.***>

[profLewis]

Also, why do you start in 2006? Is that a limitation of the mofa data? If you can, go back to 2000 or 2001 or 2 as you can take the modus there. Lewis Professor Lewis, NCEO/UCL

…

________________________________ From: Philip Lewis ***@***.***> Sent: Sunday, March 6, 2022 8:52:00 PM To: UCL-EO/Workshop2022 ***@***.***>; UCL-EO/Workshop2022 ***@***.***> Cc: Comment ***@***.***> Subject: Re: [UCL-EO/Workshop2022] Maize mapper and explorer (Issue #17) One important point - have you filtered the lai by qa information? That is *critical* or you will get nonsense out of the product. The best thing is to assign weights from the qa and run a temporal filter over the data - exactly the same as we do in the MSc practical. Professor Lewis, NCEO/UCL

________________________________ From: Philip Lewis ***@***.***> Sent: Sunday, March 6, 2022 8:48:07 PM To: UCL-EO/Workshop2022 ***@***.***>; UCL-EO/Workshop2022 ***@***.***> Cc: Comment ***@***.***> Subject: Re: [UCL-EO/Workshop2022] Maize mapper and explorer (Issue #17) Looks great - can you get an average (mean or median) for a district for years since 2000 and plot that on your mofa yield graph? That would connect the two things you have here. You can pre-calculate that and store it. It might be more Informative than the max lai you have there. Since it’s a linear yield model, it should be the same as the mean (or median) max lai (whatever you use?) put through the yield estimate. Can you also have a scatterplot of the data? The time series is useful, but a scatterplot can inform us about strength of relationship Of course, what would be really good here would be to replace the empirical model with the DA model. That should be feasible IRT for a pixel, which would be really interesting to see over multiple years. You could always pre-calculate the fitting and store the result for all pixels. All you’d need to store would be some index into the ensemble members. Then work from there. I thought Hongyuan had generated some ensembles for that? But, get this empirical thing sorted like you have, and connected to the mofa plots (with scatter plot option) and that is the minimum requirement done for this, probably. If you can take me through it tomorrow we can decide on that and what else might be worthwhile in the time available. Lewis Professor Lewis, NCEO/UCL

________________________________ From: Qingling Wu ***@***.***> Sent: Sunday, March 6, 2022 7:28:14 PM To: UCL-EO/Workshop2022 ***@***.***> Cc: Professor Philip Lewis ***@***.***>; Comment ***@***.***> Subject: Re: [UCL-EO/Workshop2022] Maize mapper and explorer (Issue #17) Empirical yield estimation layer added: [image]<https://user-images.githubusercontent.com/2849020/156938932-0cd4709b-1ab4-471c-af88-fbb8d78cad9d.png> Interactive maize area map added, with time series plotted: [image]<https://user-images.githubusercontent.com/2849020/156938959-eea3f7be-610e-4914-8fbe-c598dbf2e2f0.png> — Reply to this email directly, view it on GitHub<#17 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAOC3GSNHYWZ72M5SODIS7TU6UBM5ANCNFSM5PKJUQ5Q>. You are receiving this because you commented.Message ID: ***@***.***>

[qwu-570]

Histograms per QA level for Northern districts (and Tamale) per year:

[profLewis]

Hi Thanks, but I don’t understand that plot — what is it showing? Only a small number of QA values have any real relevance - eg full inversion, backup … so what do all of these codes mean and what are we to notice from the plot? Is this just for the growing season? How is it filtered spatially (eg c4 grass?). Lewis Professor Lewis, NCEO/UCL

…

________________________________ From: Qingling Wu ***@***.***> Sent: Tuesday, March 15, 2022 11:40:46 PM To: UCL-EO/Workshop2022 ***@***.***> Cc: Professor Philip Lewis ***@***.***>; Comment ***@***.***> Subject: Re: [UCL-EO/Workshop2022] Maize mapper and explorer (Issue #17) Histograms per QA level for Northern districts (and Tamale) per year: [image]<https://user-images.githubusercontent.com/2849020/158489679-75dbbc4f-113b-4121-9d76-b7e9b9b799b5.png> — Reply to this email directly, view it on GitHub<#17 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAOC3GUMAIKU4IM3ZNWSTA3VAENX5ANCNFSM5PKJUQ5Q>. You are receiving this because you commented.Message ID: ***@***.***>

[qwu-570]

histogram bars with grey shading are the ones from backup RT algorithms, and non-shaded bars are the "good"/"acceptable" ones.