EnsoPrMap
Computes the spatial correlation and root mean square error (RMSE) of global (60°S-60°N) precipitation anomalies (PRA) during boreal winter (December value smoothed with a 5-month triangular-weighted moving average) between model and observations
TropFlux and GPCPv2.3 1979-2018 (main)
ERA-Interim 1979-2018, HadISST and GPCPv2.3 1979-2018
Niño3.4, global60
model and observations regridded toward a generic 1°x1° grid (using cdms esmf linear method)
- seasonal cycle removed
- detrending (if applicable)
- smoothed with time running average
- spatial average
- seasonal cycle removed
- detrending (if applicable)
- smoothed with time running average
- regridding (if applicable)
- December Niño3.4 SSTA regressed onto December PRA global (60°S-60°N)
- correlation computation
- RMSE computation
monthly
mm/day/°C
precipitation (PR) sea surface temperature (SST)
The first level shows the diagnostic used to compute the metric and highlight the main differences between the model and the reference.
Figure 1: structure of precipitation anomalies (PRA) on Earth (between 60°S-60°N), showing the location of PRA associated with ENSO. Usually the teleconnection pattern is ok - corr ~ 0.5, rmse ~ 0.25mm/day/°C - but it is too weak in southern Africa, North and South Pacific, USA and South America, and too strong in the Indian Ocean. The left and right maps show respectively the reference and the model. The metrics derived are the spatial correlation and spatial RMSE between the model map and the reference map.
The second level focusses on teleconnections over key land regions: southern half of Africa, North America, South America, South East Asia, Australia.
Figure 2: precipitation anomalies (PRA) over the southern half of Africa. The reference shows a negative teleconnection (up to 0.8mm/day/°C) over southern Africa (South of 15°S) and a positive teleconnection (around 0.6mm/day/°C) over eastern Africa (Kenya & Tanzania). Usually the models simulate a weak negative teleconnection (around -0.1 or -0.2mm/day/°C) South of 10°S and a weak positive teleconnection (around 0.1 or 0.2mm/day/°C) North of 10°S. Here, the spatial teleconnection pattern is quite well simulated but the negative teleconnection over southern Africa is too weak (around -0.2mm/day/°C). The left and right maps show respectively the reference and the model.
Figure 3: precipitation anomalies (PRA) over North America. The reference shows a positive teleconnection (up to 0.6mm/day/°C) over southeast USA (around the Gulf of Mexico). Usually the models simulate this pattern but weaker (0.3 to 0.4mm/day/°C) and extends too far west in Mexico, in addition, they simulate a negative teleconnection (around -0.3mm/day/°C) in the Southwest Canada-Northwest USA. The left and right maps show respectively the reference and the model.
Figure 4: precipitation anomalies (PRA) over South America. The reference shows a strong negative teleconnection around the Amazon river and in Colombia (both up to -1mm/day/°C), and a strong positive teleconnection (around 1mm/day/°C) over southeastern South America (Uruguay and South Brazil). Usually the models simulate this pattern but its intensity is weaker (divided by two). The left and right maps show respectively the reference and the model.
Figure 5: precipitation anomalies (PRA) over South East Asia. The reference shows a negative teleconnection in Indochinese peninsula, Indonesia and Philippines (minimum around -1mm/day/°C), and weaker positive teleconnections over Southeast China and West Borneo (both around 0.5mm/day/°C). Usually the models simulate this pattern but its intensity is weaker (divided by two). Here the model simulates a pattern close to the "usual" simulated pattern but the positive teleconnection over Borneo is very weak and the one over Southeast China extends too far West. The left and right maps show respectively the reference and the model.
Figure 6: precipitation anomalies (PRA) over Australia. The reference shows negative teleconnections over this region with minima over New Guinea (around -1mm/day/°C), northwestern and northeastern Australia (both around -0.6mm/day/°C). Usually the models simulate a quite well the reference pattern but the negative teleconnection over New Guinea is too weak and even positive over the northern half of New Guinea (up to 0.8mm/day/°C). Here the model simulates quite well the reference reference pattern but does not simulate the negative teleconnection northwestern Australia and simulate a positive teleconnection over the South Island of New Zealand (up to 0.3mm/day/°C). The left and right maps show respectively the reference and the model.