ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
Area of improvementPossible interventionSuggested by/derived from?Covered by someone else?Thoughts on costFirst pass on tractability*
2
Data assimilationFund long-term maintenance of stationsYouds et al. (2021), Rogers et al. (2019)For new SOFF facilities, some provision in the budget. For others, authors suggest this is neglected10%-15% of cost of equipment. Equipment cost estimates are available, so rough calc possibleCould help to extend lifetime of existing stations, may require skilled engineers to maintain
3
Add more observation stationsKull et al. (2021)SOFF (adding a limited number) USD 200 million over three yearsCould pledge additional funds to SOFF
4
Ensure global NWPs share raw data with NMHSsYouds et al. (2021)UnknownOne expert estimated that the costs for the basic data fields needed for a rainfall projection may be $3,000 per yearLikely possible to do, e.g. could extend the S2S program; may be most effective if accompanied by capacity building
5
Ensure countries have ICT infrastructure [equipment or internet; including trained staff for upkeep] to handle dataRogers et al. (2019, p. 32), could be a private-public partnershipUnknownUnknown
6
Digitise historical observations so that they can be used for model evaluation/developmentYouds et al. (2021)International Data Rescue (I-DARE)Likely very low if using volunteers, similar to this in the UKLimited by ability to get access from LMICs of interest; may be perceived as sovereign/ security data
7
Improve sharing of observations, particularly African rainfall dataYouds et al. (2021)UnknownLikely lowUnclear why this is not already shared, likely similar concerns to above (i.e. security concerns)
8
Modelling/forecasting and post-processingResearch to better understand how to forecast weather systems in SSA and improve accuracyYouds et al. (2021)Unknown, but likely somebody working on this in academiaVery scalable, e.g. could fund data access + 1-2 FTEsImpression is that forecasting for the tropics is particularly difficult, timescales to meaningful improvements unclear
9
Evaluation/verification of existing models against observational data - to identify areas/regions/timescales that already have sufficient accuracyDouglas Parker (in an interview)Not systematicallyVery scalable, e.g. could fund data access + 1-2 FTEsSimilar evaluations have been done piecemeal, suggesting this is very feasible; impact would come from using the results to direct implementation
10
Improve downscaling by LMICs if global NWP resolution cannot be improved Tim Palmer @ Oxford, see here
Possibilities for AI and private-public partnerships		YrUnknownThough this feels technically possible, concerns about "false accuracy". Would these forecasts actually help farmers?
11
Train AI models using existing publicly available weather dataAbhijit Basu (in conversation)UnknownUnknownFeels tractable, and is likely already being done outside of LMICs.
12
Develop NWP in Africa, primarily as a capacity building exerciseYouds et al. (2021)UnknownUnknownUnlikely to be the most efficient way to improve modelling
13
Aim for much lower resolution in initial forecasts (e.g., global NWP)Tim Palmer @ Oxford, see here
Nils Wedi @ ECWMF, see here		ECWMF trialled this globally, see hereLikely very high, because of the supercomputing power required to run a forecast at this resolutionSignificant supercomputing power needed and this is a limited resource; extent to which this will improve forecasting for tropics may depend on how well underlying science is understood
14
15
*Note: these thoughts on tractability are very initial, and could change significantly with < 1/2 day of additional research
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100