Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Etats Unis → 2010 → Simulations and predictions of mosquito populations in rural Africa using rainfall inputs from satellites and forecasts

Massachusetts Institute of Technology (2010)

Simulations and predictions of mosquito populations in rural Africa using rainfall inputs from satellites and forecasts

Yamana, Teresa K.

Titre : Simulations and predictions of mosquito populations in rural Africa using rainfall inputs from satellites and forecasts

Auteur : Yamana, Teresa K. (Teresa Keiko)

Université de soutenance : Massachusetts Institute of Technology

Grade : Master of Science in Political Science (SM) 2010

Résumé
This thesis describes studies on the use of the Hydrology, Entomology and Malaria Transmission Simulator (HYDREMATS) developed and tested against field data by Bomblies et al. (2008) in simulating and predicting the potential for malaria transmission in rural Africa. The first study examined the temporal resolution of rainfall input required by HYDREMATS. Simulations conducted over Banizoumbou village in Niger showed that for reasonably accurate simulation of mosquito populations, the model requires rainfall data with at least 1 hour resolution. The second study investigated whether HYDREMATS could be effectively forced by satellite based estimates of rainfall instead of ground based observations. The CPC Morphing technique (CMORPH) (Joyce et al., 2004) precipitation estimates distributed by NOAA are available at a 30-minute temporal resolution and 8 km spatial resolution. We compared mosquito populations simulated by HYDREMATS when the model is forced by adjusted CMORPH estimates and by ground observations. The results indicate that adjusted CMORPH rainfall estimates can be used with HYDREMATS to simulate the dynamics of mosquito populations and malaria transmission with accuracy similar to that obtained when using ground observations of rainfall. The third study tested the ability of HYDREMATS to make short term predictions about mosquito populations. A method was developed by which the rainfall forcing for HYDREMATS is constructed to suit a prediction mode. Observed rainfall is used up until the date of the prediction. The rainfall for the following two weeks (or four weeks) is assumed to be the seasonal mean for that period. HYDREMATS predictions using this method were not significantly different from simulations using observed data.This thesis describes studies on the use of the Hydrology, Entomology and Malaria Transmission Simulator (HYDREMATS) developed and tested against field data by Bomblies et al. (2008) in simulating and predicting the potential for malaria transmission in rural Africa. The first study examined the temporal resolution of rainfall input required by HYDREMATS. Simulations conducted over Banizoumbou village in Niger showed that for reasonably accurate simulation of mosquito populations, the model requires rainfall data with at least 1 hour resolution. The second study investigated whether HYDREMATS could be effectively forced by satellite based estimates of rainfall instead of ground based observations. The CPC Morphing technique (CMORPH) (Joyce et al., 2004) precipitation estimates distributed by NOAA are available at a 30-minute temporal resolution and 8 km spatial resolution. We compared mosquito populations simulated by HYDREMATS when the model is forced by adjusted CMORPH estimates and by ground observations. The results indicate that adjusted CMORPH rainfall estimates can be used with HYDREMATS to simulate the dynamics of mosquito populations and malaria transmission with accuracy similar to that obtained when using ground observations of rainfall. The third study tested the ability of HYDREMATS to make short term predictions about mosquito populations. A method was developed by which the rainfall forcing for HYDREMATS is constructed to suit a prediction mode. Observed rainfall is used up until the date of the prediction. The rainfall for the following two weeks (or four weeks) is assumed to be the seasonal mean for that period. HYDREMATS predictions using this method were not significantly different from simulations using observed data.

Présentation

Version intégrale (10,2 Mb)

Page publiée le 7 décembre 2011, mise à jour le 28 mars 2019