TemperatureGAN: Generative Modeling of Regional Atmospheric Temperatures

TL;DR

TemperatureGAN is introduced, a Generative Adversarial Network conditioned on months, regions, and time periods, to generate 2 m above ground atmospheric temperatures at an hourly resolution and it is shown that TemperatureGAN produces high-fidelity examples with good spatial representation and temporal dynamics consistent with known diurnal cycles.

Abstract

Stochastic generators are useful for estimating climate impacts on various sectors. Projecting climate risk in various sectors, e.g. energy systems, requires generators that are accurate (statistical resemblance to ground-truth), reliable (do not produce erroneous examples), and efficient. Leveraging data from the North American Land Data Assimilation System, we introduce TemperatureGAN, a Generative Adversarial Network conditioned on months, locations, and time periods, to generate 2m above ground atmospheric temperatures at an hourly resolution. We propose evaluation methods and metrics to measure the quality of generated samples. We show that TemperatureGAN produces high-fidelity examples with good spatial representation and temporal dynamics consistent with known diurnal cycles.

Figures (25)

• Figure 1: TemperatureGAN model framework. Regional temperature maps are passed as input to discriminator during the training, however the generator never sees the training data
• Figure 2: Image depiction of data aggregation scheme for a specific region $R$ and for the chosen month $M$ of January. For a 4-year period $k_i$, all the 24-hour time-series (31 for each year because January has 31 days) within all four months of January are grouped into the same bucket (making it a total of 124 examples) and have the same labels
• Figure 3: Generator $G$ architecture. The sampled noise is concatenated with the learned label embedding and passed through a dense, fully-connected (FC) linear block with Rectified Linear Unit (ReLU) activation functions. The output of the linear block is sent through series of convolution layers with batch normalization to obtain the desired ($8\times8\times24$) output shape
• Figure 4: Spatial discriminator $D_{s}$ architecture. The inputs to the spatial discriminator $D_s$, are the $8\times8\times24$ temperature maps. The input is then passed through a series 2D convolution layers with the output of the prior layer serving as input into the following layer. The final convolution layer outputs a 2-dimensional $24\times1$ vector, which is flattened into a one-dimensional vector before it is passed through a dense FC linear block to produce a score
• Figure 5: Temporal discriminator $D_{t}$ architecture. The inputs to the temporal discriminator $D_t$, are the $8\times8\times24$ temperature maps and then a gradient computation is followed, to compute $\frac{\partial{T}}{\partial{t}}$. The output from the convolution layers is flattened and passed through a series of fully-connected (FC) linear layers to produce a score