Social media platforms like Reddit, YouTube, and online news portals have become dominant in understanding public opinion because of their immediacy and accessibility. The sheer amount of content in these platforms conveys important messages regarding how people perceive and respond emotionally to environmental issues. This paper presents an examination of contemporary public sentiment and emotional expression for environmental topics using state-of-the-art transformer-based NLP models. Unlike the lexicon-based approaches previously suggested, the proposed system makes use of DistilBERT for contextual sentiment classification and DistilRoBERTa for multi-label emotion detection. This paper identifies emotions such as fear, anger, trust, and joy simultaneously. Based on real-time data gathered during the project period, Top2Vec is proposed for automatic identification of trending discussion themes across multiple platforms, including Reddit, YouTube, and NewsAPI, offering up-to-date reflections of public attitudes toward environmental sustainability and climate-related issues. Experimental evaluations show that transformer-based models outperform traditional methods such as PMI and VADER in terms of contextual accuracy and sarcasm detection. By integrating sentiment, emotion, and topic-level insights, the system provides a holistic and cross-platform understanding of discussions associated with the environment and uses this knowledge to support researchers, policymakers, and organizations in data-driven decision-making to create awareness and drive environmentally positive actions.