Simulating urban expansion with interpretable cycle recurrent neural networks

Kurzfassung

Recent advances in deep learning have brought new opportunities for analyzing land dynamics,and Recurrent Neural Networks (RNNs) presented great potential in predicting land-use and land-cover (LULC) changes by learning the transition rules from time series data. However, implement-ing RNNs for LULC prediction can be challenging due to the relatively short sequence length ofmulti-temporal LULC data, as well as a general lack of interpretability of deep learning models. Toaddress these issues, we introduce a novel deep learning-based framework tailored for forecastingLULC changes. The proposed framework uniquely implements a cycle-consistent learning schemeon RNNs to enhance their capability of representation learning based on time-series LULC data.Moreover, a local surrogate approach is adopted to interpret the results of predicted instances. Wetested the method in a LULC prediction task based on time-series Landsat data of Shenzhen, China.The experiment results indicate that the cycle-consistent learning scheme can bring substantialperformance gains to RNN methods in terms of processing short-length sequence data. Also, thetests of interpretation methods confirmed the feasibility and effectiveness of adopting localsurrogate models for identifying the influence of predictor variables on predicted urban expansion instances.