Carbon sinks in the tropical rainforests are restricting the global warming to attain unprecedented heights. However, deforestation and climate change is switching them to a net carbon source at some of the deforested patches. Using machine learning algorithm we predict that more than 50% of the tropical rainforests will undergo rapid “Savannisation”/transformation by the end of 21^st century under high emission scenarios. Climate change projects ‘El Niño-like’ warming condition, which decreases precipitation in the rainforests and favors atmospheric dryness. In Central Amazonia vegetation degradation saturates the carbon sink and more than 25% of the rainforests will transform into a net carbon source due to increase in soil microbial respiration. This transition will accelerate if Eastern Pacific/Global temperature warms beyond 1.5^◦K/2.3^◦K (by 2050’s) and will undergo a steeper transit by ~2075 (2.45^◦K/3.8^◦K warming). This alteration will exacerbate global warming and has consequences for policies that are intended to stabilize Earth’s climate.

热带雨林中的碳汇正在限制全球变暖达到前所未有的高度。然而，森林砍伐和气候变化正在将一些森林砍伐的地区转变为净碳源。使用机器学习算法，我们预测到²¹世纪末，在高排放情景下，超过 50% 的热带雨林将经历快速“稀树草原化”/转变。气候变化预计将导致“类似厄尔尼诺现象”的变暖状况，从而减少雨林的降水并有利于大气干燥。在亚马逊流域中部，植被退化使碳汇饱和，由于土壤微生物呼吸的增加，超过 25% 的雨林将转化为净碳源。如果东太平洋/全球温度变暖超过 1.5 ^° K/2.3 ^° K（到 2050 年），这种转变将会加速，并且到 2075 年将经历更陡峭的转变（2.45 ^° K/3.8 ^° K 变暖）。这种变化将加剧全球变暖，并对旨在稳定地球气候的政策产生影响。