Abstract
AbstractAbstract — Air Temperature (Ta) is a fundamental measure of the Earth’s climate but is only measured at fixed locations (weather stations) and might not be always available in some areas. In contrast, land surface temperature (Ts) can be measured widely using satellites all over the globe.
Finding methods to estimate air temperature from surface temperature is an ongoing research area. in this research Five models with unique sets of inputs were tested using five machine learning algorithms in Kilimanjaro forest and desert zone data sets.
The RMSE for each model was compared with a benchmark model. Models and algorithms were ranked according to their RMSE (Root Mean Square Error) The models and algorithms reliability and consistency ranking were calculated. The best model and algorithm were determined.
Model-1 with three inputs (TsD, TsN, TaN) in forest zone and model-3 with two inputs (TaN,
∆Ts) in desert zone were the best models. ANFIS was the best algorithm in both zones.
Novel models results were compared with the benchmark model. %50 of models in both zones outperformed the benchmark model in reliability ranking. In consistency ranking comparison,
%75 of models outperformed the benchmark model in desert zone where %25 outperformed in forest zone.
Thus, machine learning improves the estimation of air temperature using surface temperature with few variables.
| Date of Award | 16 Sept 2024 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Alexander Gegov (Supervisor), Mo Adda (Supervisor) & Nick Pepin (Supervisor) |