Limitations of Remote Sensing for Assessing Damage Caused by Desert Locusts during the East Africa Upsurge

Abstract:

Desert locusts (Schistocerca gregaria) (DL) are considered one of the most dangerous pests on the planet. When rainfall increases in the arid regions across Africa and the Middle East, locusts can phase change from solitary individuals to gregarious swarms, devouring vegetation at massive scales. Remote sensing has been used to predict the likely spread of locust swarms for targeted interventions, but the spatial scale of that data have not always provided actionable information; identifying tens to hundreds of square kilometers for intervention for small teams with limited resources and often in very remote or dangerous regions. The spatial scale of damage from DL swarms has yet to be quantified using remote sensing techniques. This is in part due to: 1) infrequent time scales of population upsurges, 2) the sporadic nature of locust damage, 3) sparsity of ground data required for validation, 4) difficulty identify identifying damage due to natural vegetation senescence, and 5) anomalously high regional greenness conditions due to the increased rainfall that supported the preconditions of upsurges. However, an assessment of damage can be valuable information for ground control operations, planning response programs such as food aid distribution, and assessments of the effectiveness of interventions to reduce locust populations. The objective of this research was to explore methods of using remotely sensed data to determine damage from locusts from the most recent event in East Africa, deemed a once in 70 year event. The methodologies included an exploration of several optical and radar remotely sensed data sources with various spatial and temporal scales in order to try to capture rapid declines in vegetation due to DL activity. The data were compared to DL observations collected by ground surveys by FAO and PlantVillage. The methods explored could not differentiate random declines in greenness from DL activity and may represent a limitation of remotely sensed data. There are also several confounding factors associated with this research including widespread regional flooding, civil unrest, and COVID19 which made the attribution of vegetation decline due to DL unattainable.