=+regression function (a graph in the cited paper suggests that the authors did this):

Predictor Coef SE Coef T P Constant 111.277 2.100 52.98 0.000 temp 0.32845 0.03303 9.94 0.010 tempsqd -0.0010050 0.0001213 -8.29 0.014 S=3.44398 R–Sq=98.1% R–Sq(adj)=96.3%

Analysis of Variance Source DF SS MS F P Regression 2 1245.39 622.69 52.50 0.019 Residual Error 2 23.72 11.86 Total 4 1269.11