Correction Note on “Understanding the Sensitivity of a GCM Simulation of Amazonian Deforestation to the Specification of Vegetation and Soil Characteristics”

J. Lean Hadley Centre for Climate Prediction and Research, Meteorological Office, Bracknell, United Kingdom

Search for other papers by J. Lean in
Current site
Google Scholar
PubMed
Close
and
P. R. Rowntree Hadley Centre for Climate Prediction and Research, Meteorological Office, Bracknell, United Kingdom

Search for other papers by P. R. Rowntree in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

No abstract available.

Corresponding author address: Dr. Peter R. Rowntree, Hadley Centre for Climate Prediction and Research, Meteorological Office, Bracknell, Berkshire RG12 2SY, United Kingdom.

Email: prrowntree@meto.gov.uk

Abstract

No abstract available.

Corresponding author address: Dr. Peter R. Rowntree, Hadley Centre for Climate Prediction and Research, Meteorological Office, Bracknell, Berkshire RG12 2SY, United Kingdom.

Email: prrowntree@meto.gov.uk

1. Introduction

It has been found that some of the experiments reported in Lean and Rowntree (1997) were affected by erroneous code. The error was such that in nighttime conditions in the presence of a full vegetative cover, the stomatal resistance used in the calculation of evaporation was set to zero, producing a small, spurious enhancement of the evaporation. As full vegetative cover was only specified for the Amazonian forest, only those experiments in which the forest was not removed were affected; these were the FOREST control experiment, and the experiments REDROUGH (with reduced roughness length) and INCALB (with increased albedo). The experiment DEFOREST, with which FOREST is compared, was also rerun, because minor code changes were made in correcting the error. The durations of the experiments were as for the originals (10 yr, 3 months for FOREST and DEFOREST; 5 yr, 3 months for the other experiments), though the analysis here does not use the final 3 months.

The main results of the original experiments that need to be confirmed are as follows. Total deforestation reduced precipitation, P, but reduced evaporation, E, more, so that the moisture convergence (PE) was increased. Surface temperature was also increased. The increases in surface albedo reduced moisture convergence, so that the decreases in P were much greater than those in E. The decreases in roughness increased P and slightly reduced E and contributed over half the increase in temperature, the rest being due to changes in soil permeability. Large regional variations were found; in particular, the largest decreases in precipitation with total deforestation occurred over the northeast of the deforested area, near the mouth of the Amazon, with increases near the west coast of northern South America. This pattern could be attributed mainly to the decrease in roughness that generated similar spatial variations.

2. Results for FOREST and DEFOREST

Table 1 shows the annual mean differences between FOREST and DEFOREST for a few relevant variables, together with Student’s t-statistics for the differences between the new and old experiments and for those between FOREST and DEFOREST. The averages are for 9 yr, omitting the first year of the experiments as a spinup period.

The results show that there are highly significant changes in the soil evaporation (i.e., the surface moisture flux apart from evaporation of intercepted rain) and surface temperature in the FOREST experiment. However, the differences due to deforestation remain highly significant; the only noticeable reduction in significance is in moisture convergence (PE), which remains easily significant at the 1% level. Note that the changes due to the correction are not significant for precipitation and (PE). Both these and the changes in the DEFOREST experiment all fall below the level expected to occur by chance 40% of the time.

The significant changes are in the direction expected, with reduced soil evaporation and higher surface temperature in the rerun FOREST experiment. The evaporation is also reduced, though by less, evaporation of intercepted precipitation increasing slightly. Consequently, the warming and decreases in evaporation with deforestation are reduced by 25% for temperature and soil evaporation. The regional variations noted earlier appeared in broadly similar form in the rerun experiments. The greater decrease in P was evident mainly over the northwest of the deforested area, with only a weak increase remaining over western Colombia.

3. Results for REDROUGH and INCALB

Lean and Rowntree (1997) separated out the effects of increasing albedo and decreasing roughness length over Amazonia. Because these experiments were shorter, we may expect it to be more difficult to identify the effects of correcting the evaporation error on these experiments and on their differences from the control experiment. This is particularly so because the code error affected these experiments as well as FOREST.

The additional data indicate that the experiments are not so long that we can afford to omit data from the analysis; we have therefore analyzed differences between means for years 2–10 of the 10-yr experiments and years 2–5 of the 5-yr experiments. For this to be valid, we need to confirm that spinup effects after the first year do not affect the assessment. Comparisons of mean data from the four available 10-yr experiments (the original and rerun versions of FOREST and DEFOREST (each with and without the error) were made to confirm that the means for years 2–5 and 6–10 were no more different than expected by chance. The magnitudes of the differences were 0.02 mm day−1 for precipitation, 0.01 mm day−1 for evaporation, and 0.04 K for surface temperature.

As found for the main experiments, the statistical significance of the results (Table 2) is not much altered by the correction. The effects of the corrections on P and (PE) are again not significant. Because of this and because precipitation differences, even from experiments of 5–10 years, are subject to noise due to interannual variability, the original experiments should probably be regarded in the context of precipitation as independent realizations of the effects of deforestation. Thus, particularly for the shorter experiments on the effects of roughness and albedo changes, we should take some account of results from the original experiments, rather than ignoring them completely. The best estimate of the true effect is likely to lie between the two.

It will be noticed that the reduction in soil evaporation is less in INCALB than in FOREST. This is linked with reduced cloudiness and increased solar radiation, which limit the decline in evaporation in INCALB from that of the original run; these are associated with (insignificant) rainfall decreases. The reduced significance of surface warming due to reduced roughness may be a robust result, arising from the larger effect on evaporation (and evaporative cooling) of a change in roughness with low than with high surface resistance (Rowntree 1991). The broad patterns of rainfall differences are similar to those obtained originally; for REDROUGH there is some erosion of the increases over western and southern Brazil, while for INCALB the decreases over northwestern Amazonia are intensified while the insignificant decreases over northeastern South America east of the deforested region largely disappear.

4. Conclusions

The main conclusions from the rerun experiments are the following.

  1. The previous results are largely confirmed. However, the error caused near-surface warming and the decrease in evaporation to be overestimated in the full deforestation experiment. The magnitude of this overestimation is about 25% for surface temperature and soil evaporation.

  2. Compared with the results originally reported, the true effects of the changes appear to be mostly less for roughness and slightly larger for albedo.

REFERENCES

  • Lean, J., and P. R. Rowntree, 1997: Understanding the sensitivity of a GCM simulation of Amazonian deforestation to the specification of vegetation and soil characteristics. J. Climate,10, 1216–1235.

  • Rowntree, P. R., 1991: Atmospheric parametrization schemes for evaporation over land: Basic concepts and climate modelling aspects. Land Surface Evaporation: Measurement and Parametrization, T. J. Schmugge and J.-C. Andre, Eds., Springer-Verlag, 5–29.

Table 1.

The 9-yr means for the original and rerun experiments FOREST and DEFOREST, differences, and t statistics for the differences.Averages are over all deforested points. Data are soil evaporation (Esoil), total evaporation (E), precipitation (P), and surface temperature (T*).

Table 1.
Table 2.

Differences between 9-yr (years 2–10) means for FOREST experiments and 4-yr (years 2–5) means for experiments REDROUGH and INCALB, differences, and t-statistics for the differences. Averages are over all deforested points. Notation for data as in Table 1.

Table 2.
Save
  • Lean, J., and P. R. Rowntree, 1997: Understanding the sensitivity of a GCM simulation of Amazonian deforestation to the specification of vegetation and soil characteristics. J. Climate,10, 1216–1235.

  • Rowntree, P. R., 1991: Atmospheric parametrization schemes for evaporation over land: Basic concepts and climate modelling aspects. Land Surface Evaporation: Measurement and Parametrization, T. J. Schmugge and J.-C. Andre, Eds., Springer-Verlag, 5–29.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 177 22 2
PDF Downloads 18 7 1