Search

Sao Tome and Principe is a small insular African country extremely vulnerable to rising sea levels and impacts such as inundation, shore line change, and salt water intrusion into underground aquifers. Projections of climate change have considered coarse model resolutions. The objective of this work is to dynamically downscale the global model projections to 4-km resolution and to assess the climate change in the Sao Tome and Principe islands. The global climate projections are provided by the Canadian Earth System Model under two Representative Concentration Pathways greenhouse gas scenarios, RCP4.5 and RCP8.5. The downscaling is produced by the Eta regional climate model. The baseline period is taken between 1971 and 2000, and the future climate period is taken between 2041 and 2070. The 2-m temperature simulations show good agreement with station data. The model simulates temperature more accurately than precipitation. The precipitation simulations systematically show underestimation and delay of the rainy and the dry seasons by about 1 month, a feature inherited from the global climate model. In the middle of the 21st century, projections show the strongest warming in the elevated parts of the Sao Tome Island, especially in February under RCP8.5. Warmer nights and warmer days become more frequent in the islands when compared with those in the present. While under RCP4.5, precipitation increases in the islands; under RCP8.5, it decreases everywhere in both islands. Heavy precipitation rates should increase, especially in the south-southwestern parts of the Sao Tome islands. Detailed spatial variability of the temperature and precipitation changes in the islands can only be revealed at very high spatial model resolution. Implications for the potential energy production from two major river basins are assessed in this work.

This study evaluates the use of tiles to estimate the effects of surface heterogeneities in simulations of local climate in a landscape-fragmented region of Amazonia. Three experiments using the Eta/Noah-MP model were designed as follows: two experiments with a spatial resolution of 5 km with and without the use of the tile approach. The third experiment used a finer spatial resolution of 2 km without the use of the tile approach (high-resolution experiment). Simulations were carried out for the rainy and dry seasons, which correspond to the months of March and September 2003, respectively. In general, the magnitude of surface fluxes is significantly affected by the introduction of tiles. In the dry season, the use of tiles in the simulation increases the magnitude of precipitation (about 7 mm day−1), skin temperature (about 4 °C), sensible heat flux (about 80 W m−2) and surface albedo (about 0.7). It reduces the latent heat flux (about 30 W m−2) and net radiation (about 40 W m−2). On the other hand, in the rainy season, the model produces excessive precipitation. This may have partly masked the signs of the land fragmentation impact. The tile experiment generates a warmer and drier boundary layer during the dry season. The impact of the surface heterogeneity representation is higher in the dry season than in the rainy season. The results suggest that the degree of impact of the subgrid process representation on the local climate is related to the spatial scale of the fragmentation. The use of the tile approach improves the representation of the effects of landscape heterogeneity on the spatial distribution of surface flux variability in fragmented areas. Comparison against measured data from flux towers in the region show that the model can simulate diurnal and seasonal variations in the local fluxes, despite the biases.

This study evaluates the simulations of the Noah-MP surface processes over a crop and a forest sites in Amazonia using tower observations. Soil moisture simulations agree with the observations in both land covers, mainly during the rainy season. However, simulations show cold biases in the soil temperature at both sites. The magnitude and seasonal cycle of the surface energy fluxes are better simulated at the crop site, although the model significantly underestimates the sensible heat flux at this site. The model reproduces the seasonal pattern of surface runoff at both sites. The Noah-MP model does not adequately simulate the base flow at the crop site, while the simulated total runoff at the forest site is closer to the observation than at the crop site. The results show that, in general, the Noah-MP model simulations for the two sites in Amazonia exhibit fairly realistic performance, particularly over the crop site. However, there are cold biases in soil temperature simulations, which could be related with the parameterization of the equilibrium relationship between soil moisture and soil temperature.