Temporal variability of the Buenos Aires, Argentina, urban heat island

Abstract

This paper describes the statistical characteristics and temporal variability of the urban heat island (UHI) intensity in Buenos Aires using 32-year surface meteorological data with 1-h time intervals. Seasonal analyses show that the UHI intensity is strongest during summer months and an “inverse” effect is found frequently during the afternoon hours of the same season. During winter, the UHI effect is in the minimal. The interannual trend and the seasonal variation of the UHI for the main synoptic hours for a longer record of 48 years are studied and associated to changes in meteorological factors as low-level circulation and cloud amount. Despite the population growth, it was found a negative trend in the nocturnal UHI intensity that could be explained by a decline of near clear-sky conditions, a negative trend in the calm frequencies and an increase in wind speed. Urban to rural temperature differences and rural temperatures are negatively correlated for diurnal and nocturnal hours both for annual and seasonal scales. This result is due to the lower interannual variability of urban temperatures in comparison to rural ones.

Appendix

The linear correlation coefficient (R _{UHI, r} ) between the yearly mean urban to rural temperature differences (UHI) and rural (r) temperature for N years can be expressed as

$$ {R_{{{\text{UHI}},{ }r}}} = \frac{{{\sigma_{{{\text{UHI}},{ }r}}}}}{{{\sigma_{\text{UHI}}}{\sigma_r}}} $$

(3)

where σ _{UHI, r} is the covariance between UHI and r, and σ _UHI and σ _r are the corresponding standard deviations. Considering that UHI is calculated as the difference between the annual mean urban (u) and rural (r) temperatures and $\overline u$and $\overline r$are the means of u _i and r _i with i = 1,……N, Eq. 3 becomes

$$ {R_{{{\text{UHI, }}r}}} = \frac{1}{N}\,\frac{{\sum\limits_{{i = 1}}^N {\left[ {\left( {{u_i} - {r_i}} \right) - \overline {\left( {u - r} \right)} } \right]\left( {{r_i} - \overline r } \right)} }}{{{\sigma_{\text{UHI}}}{\sigma_r}}} $$

(4)

so

$$ {R_{{{\text{UHI, }}r}}} = \frac{1}{N}\frac{{\sum\limits_{{i = 1}}^N {\left[ {\left( {{u_i} - {r_i}} \right) - \left( {\overline u - \overline r } \right)} \right]\left( {{r_i} - \overline r } \right)} }}{{{\sigma_{\text{UHI}}}{\sigma_r}}} $$

(5)

and

$$ {R_{{{\text{UHI, }}r}}} = \frac{1}{N}\frac{{\sum\limits_{{i = 1}}^N {\left[ {\left( {{u_i} - \overline u } \right) - \left( {{r_i} - \overline r } \right)} \right]\left( {{r_i} - \overline r } \right)} }}{{{\sigma_{\text{UHI}}}{\sigma_r}}} $$

(6)

$$ {R_{{{\text{UHI, }}r}}} = \frac{1}{{N{\sigma_{\text{UHI}}}{\sigma_r}}}\left[ {\sum\limits_{{i = 1}}^N {\left( {{u_i} - \overline u } \right)\left( {{r_i} - \overline r } \right)} - {{\sum\limits_{{i = 1}}^N {\left( {{r_i} - \overline r } \right)} }^2}} \right] $$

(7)

Multiplying and dividing by σ _u

$$ {R_{{{\text{UHI, }}r}}} = \frac{1}{{N{\sigma_{\text{UHI}}}{\sigma_r}}}\frac{{{\sigma_u}}}{{{\sigma_u}}}\left[ {\sum\limits_{{i = 1}}^N {\left( {{u_i} - \overline u } \right)\left( {{r_i} - \overline r } \right)} - {{\sum\limits_{{i = 1}}^N {\left( {{r_i} - \overline r } \right)} }^2}} \right] $$

(8)

$$ {R_{{{\text{UHI, }}r}}} = \frac{{{\sigma_u}}}{{{\sigma_{\text{UHI}}}}}\left[ {\frac{1}{N}\frac{{\sum\limits_{{i = 1}}^N {\left( {{u_i} - \overline u } \right)\left( {{r_i} - \overline r } \right)} }}{{{\sigma_u}{\sigma_r}}} - \frac{1}{N}\frac{{{{\sum\limits_{{i = 1}}^N {\left( {{r_i} - \overline r } \right)} }^2}}}{{{\sigma_u}{\sigma_r}}}} \right] $$

(9)

The linear correlation coefficient (R _{u, r} ) between the yearly mean urban and rural temperatures for N years can be expressed is

$$ {R_{{u,{ }r}}} = \frac{1}{N}\frac{{\sum\limits_{{i = 1}}^N {\left( {{u_i} - \overline u } \right)\left( {{r_i} - \overline r } \right)} }}{{{\sigma_u}{\sigma_r}}} $$

(10)

Thus, Eq. 9 becomes

$$ {R_{{{\text{UHI, }}r}}} = \frac{{{\sigma_u}}}{{{\sigma_{\text{UHI}}}}}\left[ {{R_{{u,{ }r}}} - \frac{{{\sigma_r}^2}}{{{\sigma_u}{\sigma_r}}}} \right] $$

(11)

Resulting in

$$ {R_{{{\text{UHI, }}r}}} = \frac{{{\sigma_u}}}{{{\sigma_{\text{UHI}}}}}\left[ {{R_{{u,{ }r}}} - \frac{{{\sigma_r}}}{{{\sigma_u}}}} \right] $$

(12)