Browsing by Author "Sulungu, Emmanuel D."

Now showing 1 - 5 of 5
  • Thumbnail Image
    Item
    Application of neural network in prediction of temperature: a review
    (Springer Nature, 2021) Johnstone, Charles; Sulungu, Emmanuel D.
    The aim of this study was to review different literatures to assess the applicability of artificial neural network in predicting temperature. Temperature prediction as part of weather prediction involves the application of science and technology to predict the state of temperature for a future period in a specific location. Artificial neural network (ANN) has been found to be a promising tool to be used in temperature prediction because it is able to handle complex and nonlinear physical variables of the atmosphere. The use of ANN for prediction of weather elements has shown significant improvements in prediction and accuracy. The performance of the ANN model varies depending on the nature and number of input data used in training the network, the number of neurons in the hidden layer, architecture of a network, transfer function and on the training algorithms. The choice of ANN architecture and the type of data depend on the nature of the problem to be addressed. ANN is therefore found to be a powerful tool in predicting temperature of a specific place, provided input parameters of the model are well chosen.
  • Thumbnail Image
    Item
    Comparison of GPS derived TEC with the TEC predicted by IRI 2012 model in the southern Equatorial Ionization Anomaly crest within the Eastern Africa region
    (Elsevier, 2018) Sulungu, Emmanuel D.; Sibanda, Patrick; Uiso, Christian B. S.
    We have compared the TEC obtained from the IRI-2012 model with the GPS derived TEC data recorded within southern crest of the EIA in the Eastern Africa region using the monthly means of the 5 international quiet days for equinoxes and solstices months for the period of 2012 – 2013. GPS-derived TEC data have been obtained from the Africa array and IGS network of ground based dual-frequency GPS receivers from four stations (Kigali (1.95°S, 30.09°E; Geom. Lat. 11.63°S), Malindi (2.99°S, 40.19°E; Geom. Lat. 12.42°S), Mbarara (0.60°S, 30.74°E; Geom. Lat. 10.22°S) and Nairobi (1.22°S, 36.89°E; Geom. Lat. 10.69°S)) located within the EIA crest in this region. All the three options for topside Ne of IRI-2012 model and ABT-2009 for bottomside thickness have been used to compute the IRI TEC. Also URSI coefficients were considered in this study. These results are compared with the TEC estimated from GPS measurements. Correlation Coefficients between the two sets of data, the Root-Mean Square Errors (RMSE) of the IRI-TEC from the GPS-TEC, and the percentage RMSE of the IRI-TEC from the GPS-TEC have been computed. Our general results show that IRI-2012 model with all three options overestimates the GPS-TEC for all seasons and at all stations, and IRI-2001 overestimates GPS-TEC more compared with other options. IRI-Neq and IRI-01-corr are closely matching in most of the time. The observation also shows that, GPS TEC are underestimated by TEC from IRI model during noon hours, especially during equinoctial months. Further, GPS-TEC values and IRI-TEC values using all the three topside Ne options show very good correlation (above 0.8). On the other hand, the TEC using IRI-Neq and IRI-01- corr had smaller deviations from the GPS-TEC compared to the IRI-2001.
  • Thumbnail Image
    Item
    Performance of IRI 2016 model in predicting total electron content (TEC) compared with GPS-TEC over East Africa during 2019–2021
    (Springer Science and Business Media LLC, 2024) Sulungu, Emmanuel D.
    This study evaluated the applicability of IRI-2016 model in predicting GPS TEC using the monthly means of the fve (5) quiet days for equinoxes and solstices months. GPS-derived TEC data were obtained from the IGS network of ground based dual frequency GPS receivers from three stations [(KYN3 0.53° S, 38.53° E; Geom. Lat. 3.91.63° S), (MBAR 0.60° S, 30.74° E; Geom. Lat. 2.76° S) and HOID 1.45° S, 31.34° E; Geom. Lat. 3.71° S]. All the three options for topside Ne of IRI-2016 model and ABT-2009 for bottomside thickness have been used to compute the IRI TEC. The results were compared with the GPS TEC measurements. Correlation Coefcients between the two sets of data, the Root-Mean Square Errors of the IRI-TEC from the GPS-TEC, and the percentage RMSE of the IRI-TEC from the GPS-TEC have been computed. In general, the IRI-2016 model underestimated GPS-TEC during the nighttime, whereas the model overestimated GPS-TEC values during the daytime. At most of the stations and during all seasons where data were available, correlation coefcient was above 0.9, which is quite strong. The variation of O/N2 ratio may potentially be the cause of the IRI TEC deviation from the GPS TEC. This variation arises from lower thermosphere plasma drift that moves upward.
  • Thumbnail Image
    Item
    Total electron content derived from global positioning system during solar maximum of 2012-2013 over the eastern part of the African sector
    (University of Dar es Salaam, 2018) Sulungu, Emmanuel D.; Uiso, Christian B. S.; Sibanda, Patrick
    This work presents results of diurnal, seasonal and latitudinal variations of vertical Total Electron Content (TECv) derived from GPS receivers at four locations, [Dodoma (6.19oS, 35.75oE), Mzuzu (11.43oS, 34.01oE), Zomba (15.38oS, 35.33oE) and Tete (16.15oS, 33.58oE)] during the solar maximum period of 2012 – 2013. The receivers are located directly below the EIA and at approximately the same longitude, ~ (33 – 3 oE) within the eastern part of the African sector. Diurnal and latitudinal variations of TECv are presented for an average of the five (5) quietest days of each of the four seasons: March equinox, June solstice, September equinox and December solstice; for the seasonal variations all months in a year were considered. Results showed that TECv is characterized by consistent minimum diurnal variations during presunrise hours, rises steeply during the sunrise period to the maximum peak during the daytime, followed by a decrease to a minimum during nighttime. The values of TECv from all stations used and for both years (2012 and 2013) showed semiannual variations. Our study also showed that, the day maximum value of the TECv decreased significantly with the increase in latitude.
  • Thumbnail Image
    Item
    Total electron content prediction model using the artificial neural networks over the Eastern Africa Region
    (College of Natural and Applied Sciences, University of Dar es Salaam, 2019) Sulungu, Emmanuel D.; Uiso, Christian BS
    In this paper, development of a model using NN technique for prediction of GPS TEC over the Eastern Africa region is presented. TEC data was obtained from the Africa array and IGS network of ground based dual-frequency GPS receivers from 18 stations within the East African region. It covers approximately the area from ~2.6°N to ~26.9°S in magnetic latitudes and from ~95°E to ~112oE in magnetic longitudes. The input layer of the developed model consisted of seven neurons which were selected by considering the parameters that are known to affect the TECv data. The results showed that when the number of hidden layer neurons surpassed about 18, the RMSEs were noted to continuously increase indicating poor predictions beyond this number. The RMSE at this point was observed to be about 5.2 TECU which was lowest of all. The errors and relative errors were fairly small. Developed NN model estimated GPS TECv very well compared to IRI model. It is established in this study that, the IRI electron density at F2 peak (NmF2) gives good GPS TECv prediction when added as an input neuron to the NN.