Forecasting vegetation condition with a Bayesian  auto-regressive distributed lags (BARDL) model

Salakpi, Edward E.; Hurley, Peter D.; Muthoka, James M.; Barrett, Adam B.; Bowell, Andrew; Oliver, Seb; Rowhani, Pedram

doi:https://doi.org/10.5194/nhess-22-2703-2022

Articles | Volume 22, issue 8

https://doi.org/10.5194/nhess-22-2703-2022

© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/nhess-22-2703-2022

© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 22, issue 8

Research article

|

23 Aug 2022

Research article |

| 23 Aug 2022

Forecasting vegetation condition with a Bayesian auto-regressive distributed lags (BARDL) model

Edward E. Salakpi, Peter D. Hurley, James M. Muthoka, Adam B. Barrett, Andrew Bowell, Seb Oliver, and Pedram Rowhani

Data sets

MCD43A4 MODIS/Terra+Aqua BRDF/Albedo Nadir BRDF Adjusted Ref Daily L3 Global - 500 m V006 C. Schaaf and Z. Wang https://doi.org/10.5067/MODIS/MCD43A4.006

Model code and software

edd3x/Bayesian-ARDL: Data and Code for Bayesian Auto-regressive Distributed lag models (v.1.0) E. E. Salakpi, A. B. Barrett, and A. Bowell https://doi.org/10.5281/zenodo.7005168

Short summary

The devastating effects of recurring drought conditions are mostly felt by pastoralists that rely on grass and shrubs as fodder for their animals. Using historical information from precipitation, soil moisture, and vegetation health data, we developed a model that can forecast vegetation condition and the probability of drought occurrence up till a 10-week lead time with an accuracy of 74 %. Our model can be adopted by policymakers and relief agencies for drought early warning and early action.