Statistical relationship between onset dekad of raining season and normalized difference vegetation index: an indicator of acute food insecurity levels in districts of Zambia.
Date
2024
Authors
Hamakala, Derick
Journal Title
Journal ISSN
Volume Title
Publisher
The University of Zambia
Abstract
Agricultural drought conditions driven by delayed onset of the rainy season is one of the main contributing factors to Acute Food Insecurity (AFI). A number of studies on AFI severity in Zambia have been done at national level but no assessment of severity of AFI at district level has been done. This study investigated the statistical relationship between the delays in the Onset of the rainy season and peak Normalized Difference Vegetation Index (NDVI) and associated it with the AFI levels in the districts of Zambia using AFI classification of the Disaster Management and Mitigation Unit (DMMU), Zambia. The NDVI data for districts of Zambia were downloaded from USGS EROS’s FEWS NET’s Data warehouse. The rainfall data for districts of Zambia were downloaded from the Climate Hazards Group InfraRed Precipitation (CHIRPS) wbsite, Climate Hazards Centre, University of California, Santa Barbara. CHIRPS data has been assessed against the Zambia Meteorological Department’s station observations for its capability in reproducing
rainfall climatology. The rainy season onset was determined for seasons from 2005/06 to 2020/21 using prevailing criteria for determining onset in the Southern African region. Peak NDVI for districts of Zambia for each of the mentioned seasons was determined at the end of the raining season. The Pearson Correlation Coefficient was employed to investigate the relationship between onset dekad of a rainy season and NDVI. The Districts of Zambia were put in 3 bins based on AFI modes (AFI-2 bin 1, AFI-3 bin 2 and AFI-4 bin 3). For each bin, an empirical Pearson correlation was computed. The proportion of districts in each bin of AFI levels were utilized to partition correlations range from -1 to +1 into three subintervals. If correlation between seasonal onset dekad anomalies and standardized peak NDVI for a district falls in subinterval 1, the district is likely to experience AFI level 4, if correlation falls in the subinterval 2, the district is likely to experience AFI of level 3 and if correlation falls in subinterval 3 then district is likely to experience AFI level 2. The results showed that CHIRPS data captured the monthly climatology of the rainy season for the districts of Zambia reasonably well. It was found that there is a negative relationship between onset occurrence (early or delayed onset) and peak NDVI. Each district in the study was assigned an AFI severity level using DMMU’s AFI data, It was noted that districts of Zambia generally experienced only three levels of AFI severity namely 2 (stressed), 3 (crisis) and 4 (emergency) out of five classified levels where level 5 refers to Catastrophe or Famine. Districts
were binned together according to their AFI levels. Pearson correlation coefficient between anomaly in onset dekad and peak NDVI were computed using data for districts in each of the three bins for AFI levels 2, 3 and 4. The correlations were -0.265 for level 4, -0.09 for level 3 and 0.01
for level 2. The empirical correlations confirmed that delay in onset of rainy season is negatively correlated with low vegetation index leading to AFI of different levels. Subinterval 1 is [-1, -0.96) subinterval 2 is [-0.96, 0.1) and subinterval 3 is [0.1, 1]. This early warning system for assessing the AFI levels in the districts of Zambia for a delayed onset season can be improved as more DMMU disaster related data, longer climate and NDVI daily time series data are available in future. It can enable government to plan and organize resources required to provide AFI stricken communities a relief aide according to the severity way before a disaster situation becomes grave due to delay in onset of rainy season.
Key words: Onset Dekad, CHIRPS, NDVI and AFI.
Description
Thesis of Master of Science in Statistics.