Myanmar EM | EVI Analysis

[SahitiSarva]

[SahitiSarva]

Notes from 23rd October call

• Real agriculture places have no nighttime lights whatsoever
• If NTL is expanding, it can be a hypothesis that other sectors are coming in and replacing agriculture
• Type of crops: what is growing in which areas
• Draw a line for what we can carry forward to the December EM
• Conflict-affected and non-conflict-affected EVI could be an addition suggested by Nay
• Exports have an element of lag. It would be interesting to establish a relationship with EVI and vegetable exports
• Mid Oct-Dec is harvest in some periods

Actions Items

• Correlate the breakdown of GDP for vegetation alone and see if it works.
• Correlate agriculture exports
• Drop evi lag 2 and see if it still shows the same results, and is robust
• Aka to send us the coordinates of which crops are grown in which admin regions.
• Shan EVI could be Opium, compare with UNDC. Thi Da to send the details on this.
• Less priority: Aka to send areas where conflicts have happened [regions that have been outside for a long time vs new regions of conflict]
• Compare EVI 2024 to 2025, stack the months on top of each other
• Control for the season by adding a dummy variable

Questions

• What are the other variables we want to add to this?
• national level imported fertiliser

[SahitiSarva]

Notes from 30th October calls

• Corn is a little different crop. Between Feb and May --> is non-corn season in Thailand so Myanmar exports to Thailand only go during Feb and August.
• More sophisticated modelling for rice because most of the rice sent from Myanmar is broken rice.
• There were better rains this year, but there was an expected earthquake impact

Action Items

• Add a dummy variable to control for corn calendar (not priority)
• Ag GDP and Lag EVI chart at a national level and extend the EVI till the latest available date (quarterly + annual)
• Setup a call with the ag team separately (@SahitiSarva )
• Have a map showing the harvested area and the crop area (not priority)

[SahitiSarva]

Notes from the call with agri team

• Even the agri team doesn't know how the govt calculated crop area numbers are calculated
• Estimate of the error of the regression line
• If the ranking looks similar between satellite and govt numbers then its a sufficient robustness check.
• We need to add rainfall data
• EVI closely mirror total harvested area of the country

[SahitiSarva]

Note should include:

• We didn't find consistency in the actual crop area.
• add rainfall and improve results
• the cyclical nature of EVI and GDP agriculture is matching with each other
• add a footnote on what EVI is and where it has been used.
• Look at the crop season this year vs the crop season last year. [October is yet to come]
• median EVI from July-September and what is the change in median EVI

[SahitiSarva]

• Sensitivity Tests
• Try to predict the agricultural GDP for 2024 and 2023
• Add rainfall into the model and see how the results change
• Work with the national model and the subnational model

[SahitiSarva]

Notebook structure

1. What is EVI?
2. Phenology process matching the crop season in Myanmar
3. Crop area comparison chart with ranking
4. National level regression
5. Admin level regression
6. Findings
7. Potential prediction for 2025