|
Descripción
|
|
|---|---|
| Desert locust (Schistocerca gregaria) has severely influenced crop production in northern Africa and Middle East since antiquity. To prevent or mitigate its effects on local communities, it is necessary to precisely locate those areas where they breed. Previous works have relied on precipitation and vegetation indices obtained by satellite remote sensing, however many authors (Browning et al., 1990; Gay et al., 2018) agree on the necessity to improve desert locust prevention systems. In this PhD thesis, we have explored 3 different novel approaches to locate desert locust breeding areas in Mauritania. (1) Firstly, the SWAT hydrological model was used to locate wadis that may host desert locust given their favourable ecological conditions. (2) Secondly, the influence of soil moisture (SM) using the European Space Agency Climate Change Initiative Soil Moisture (ESA CCI SM) product was assessed over desert locust breeding sites using Artificial Intelligence (AI) and more specifically machine learning techniques. (3) Finally, we have generated a multivariate ensemble model using a combination of the frequently used Species Distribution Models (SDMs) in ecology. The results in (2) showed a good correlation between general monthly soil moisture patterns and hopper presences. It was found that an area becomes suitable for breeding when the minimum SM values are over 0.07 m3/m3 during 6 days or more. On the other hand, the identified wadis by means of SWAT hydrological model (1) did not find significant influence on locust presences for the studied period. Many uncertainties in precipitation records, as well as poor river gauge data were encountered, what impeded adequate calibration and validation procedures. Longer and more accurate data records (precipitation and river gauge) may permit to further develop this approach in the near future. Furthermore, the third approach showed highly satisfactory model results (KAPPA & TSS = 0.901 and ROC = 0.986) to detect hopper desert locust in solitary phase, implying that our model can identify suitable environmental conditions for breeding. This study also confirms the potential of the SMAP satellite from NASA to retrieve critical temperatures due to its time pass, in addition to reinforcing the NDVI product from MODIS as a reliable environmental predictor (3). These results demonstrate the validity of the methodologies exposed in this PhD thesis to identify favourable breeding areas in Mauritania. Earth observation techniques can retrieve periodically important environmental variables such as soil moisture, surface temperature or vegetation status to be managed by machine learning algorithms over remote and large areas. Thus, our work may be of interest to authorities of affected countries or international organizations to complement or improve current ongoing monitoring techniques and warning systems. | |
|
Internacional
|
No |
|
ISBN
|
|
|
Tipo de Tesis
|
Doctoral |
|
Calificación
|
Sobresaliente cum laude |
|
Fecha
|
15/03/2019 |