COMPOSITE INDEX-BASED EARLY WARNING SYSTEM FOR DROUGHT MONITORING IN THE NIGER RIVER BASIN OF WEST AFRICA

Abstract

This study seeks to develop a comprehensive and integrated Early Warning System (EWS) for concurrent monitoring of meteorological, agricultural, and hydrological droughts in the Niger Basin of West Africa. The specific objectives of the research are as stated below. First, to establish threshold(s) for defining and detecting moderate to severe drought events in the Niger River. To accomplish the task, Baseline Assessment Analysis (BAA) using combined tool of Standardized Precipitation Index (SPI 6-month) and Percentile Rank. Second, to determine the thresholds beyond which a dry spell changes and becomes actual drought in the Niger Basin. This was achieved through combination of baseline assessment analysis (BAA), the standardized precipitation index (SPI 2-month) and percentile rank analyses. The SPI-2month has weak correlation with the ecosystems, because dry spell is a precursor to drought and not real drought. Third, to develop a Niger Basin Drought Monitoring (NBDM) scheme or Drought Empirical Model (DREM). The task was accomplished through development of an Objective Blend of Drought Disaster Burden Index (OBDDBI), using as inputs, the Standardized Effective Precipitation Index (SEPI 6-month), Soil Moisture Index (SMI) and Streamflow Index (SFI) to form one big picture composite index with ‘all-in-one’ drought initiation thresholds. Then, characterize the basin’s moderate to severe drought events, using the established drought thresholds from the DREM. Fourth, to evaluate and validate the performance of NBDM to ascertain its reliability or usefulness. This task was achieved by first evaluating the Composite Drought Index (CDI) ability to reproduce other input indices using some statistical tools. This was followed by the validation of the obtained CDI outputs from the DREM against ENSO chronology and known drought induced famine impacts reports in the drought chronology of the region. Furthermore, all the analyses including the development of the DREM and establishment of all-in-one dry spells and drought thresholds were done using hydrometeorological reanalysis datasets from 1980-2016. Results based on the DREM composite drought index (CDI) all-in-one thresholds, revealed that the onset of drought of moderate intensity, either meteorological, agricultural or hydrological can be defined with thresholds of range -0.26 to -1.19 over three consecutive months depending on the location. In terms of percentiles, it corresponds to threshold of 20th percentile. With DREM, therefore, these are thresholds to trigger alert for the occurrence of different biophysical forms of drought events of moderate intensity in the Niger River Basin. However, with SPI-6months all-in-one thresholds, the droughts of moderate or worst intensity occur when the sum of cumulative precipitation deficits equal or falls below the critical thresholds of -0.37 to -1.08 or less over three consecutive months depending on the location. This corresponds to 20th percentiles monthly precipitation deficits; thereby, resulting in late drought onset detection relative to CDI. On the other hand, dry spells occur in the Niger River Basin when the sum of cumulative precipitation deficits of SPI 2-month equal or falls below the thresholds of range -0.22 to -0.45 over 2 or more consecutive months, depending on the location. This corresponds to 35th percentile monthly precipitation deficits. However, phase change from dry spell to actual drought condition, is initiated as the cumulative precipitation deficits increases with the SPI 2-month values falling below the critical dry spell threshold of - 1.20 (corresponding to 10th percentile) and becoming fully established drought at threshold of - 1.66 (corresponding to 5th percentile). The potential of the CDI thresholds being used operationally, as all-in-One thresholds for concurrent monitoring of all biophysical forms of drought was investigated. Additionally, evaluation of the results showed significant values of the index of agreement (d) of 0.953 and 0.978, 0.844 and 0.898, 0.914 and 0.932, 0.890 and 0.910 between CDI and Soil Moisture Index (SMI) and Streamflow Index (SFI) over Upper Niger, Inland Niger, Middle Niger and Lower Niger sub-basins respectively. Validation results further revealed success rate of range 67 to 100% based on the past records of drought disaster events captured by the DREM CDI, and 62 to 77% based on ENSO-induced drought records. In conclusion, DREM could make an effective and reliable drought monitoring and early warning decision support tool.