VEGETATION - 2000

Lake Maggiore - Italy, 3-6 April 2000

VEGETATION contribution to the desert locust habitat monitoring

Cherlet Michael* , Mathoux Pierre**, Bartholomé Etienne*** and Defourny Pierre**
*FAO Locust Group, Plant Protection Service
** Department of Environmental Sciences - Université catholique de Louvain
Croix du Sud, 2 bte 16 B-1348 Louvain-la-Neuve
***JRC-SAI - Global Vegetation Monitoring Unit I-21020 Ispra

Paper (pdf file, 299 k)

The Desert Locust, Schistocerca gregaria, is a continuous threat to agriculture, subsistence farming and pastures in the arid areas of northern Africa, the Middle East and south-west Asia. Control of the Desert Locust is an important part of the general effort in ensuring food security. To further improve the combating of this migrant pest and to improve its routine global locust monitoring and forecasting activities, the FAO is implementing a regional Emergency Prevention programme, EMPRES. One aspect of the programme focuses on increasing national early warning capacities in view of optimising monitoring through more efficient survey and control planning .

Early detection and control of initial locust populations in the recession area are critical in order to prevent the development of outbreaks and plagues. Traditionally, Plant Protection Services organise field surveys to try to obtain an idea of the locust populations and the condition of the locust habitats in terms of soil moisture and vegetation. Low resolution satellite remote sensing, integrated with other field information, offers a cheap means to obtain a synoptic overview on the conditions of the habitats in near real time. Rainfall is the first indicator of potentially good habitat conditions, but estimates based on satellite data are often not adequate over desert areas . However, the important secondary effects of rainfall such as soil moisture and vegetation growth, can be more reliably observed and monitored by low resolution satellites. Past projects developed methodologies to increase the reliability of the AVHRR NDVI time series. However, the limits were reached for using NOAA to detect very sparse vegetation due to poor image positioning, poor calibration and high sensitivity of the NDVI-MVC compositing technique to the directional perturbing factors.

This research aims to provide demonstrative results for Desert Locust habitat monitoring using the SPOT VEGETATION that can be readily implemented into actual operations. The current work focuses on three main technical issues.

1. The sensitivity of the VEGETATION signal to soil moisture change and to very low levels of green biomass. The detection of soil humidity is based on the SWIR band. A feasibility study investigates the impact of actual rainfall events on the observed SWIR signal over bare soil areas. The detection of the light greening of the desert is essential and requires selection of a sensitive vegetation index which is not disturbed by soil variability. To improve the consistency of the selected index, ‘correction’ of bi-directional effect is performed. These analyses are carried out using daily S1 VEGETATION product.
2. However, handling daily synthesis is a major constraint for operational monitoring. Therefore the S10 product is also investigated with regard to optimising interpretation reliability by looking at: the influence of the temporal sampling on the monitoring performances, the impact of the MVC-NDVI compositing strategy on these performances and the index calibration using frequent and targeted field observations provided by national locust teams.
3. The VEGETATION central processing and distribution principle offers a sound alternative to solve the existing problems of timely data delivery provided some specific work is done on transformation of data. This project will consider developing dedicated methods for the transformation of data into information appropriate for fast delivery. "Intelligent" compressing methods, i. e. transfer of only the significant information from the thematic point of view, will be considered in order to keep data flow as low as possible. Optimised and reliable final information, e.g. a "habitat risk-map", can then be efficiently transferred to the final users which are the national service in charge of desert locust control in affected countries.