--- title: 'SeisDiff-denoNIA: Diffusion Seismic Denoising' url: https://www.emergentmind.com/topics/seisdiff-denonia type: topic --- # SeisDiff-denoNIA: Diffusion Seismic Denoising SeisDiff-denoNIA is a diffusion-based seismic denoising designation used in several recent seismic-processing contexts built around denoising diffusion probabilistic models, time-conditioned U-Nets, and reverse-time sampling for noise suppression on seismic images, shot gathers, or VSP acquisitions. In its most specific 2025 VSP usage, it denotes the Seismic Diffusion Model for Denoising applied to tube-wave attenuation in vertical seismic profiling data, where a conditional DDPM learns to remove coherent tube-wave noise while preserving reflected body-wave events [2503.00637]. Earlier and parallel uses describe a seismic denoising variant trained on synthetic noisy-clean patches [2207.10451] and a later reformulation that trains directly on extracted field noise rather than clean signal labels [2509.03629]. ## 1. Terminology and scope In the supplied literature, the label “SeisDiff-denoNIA” does not denote a single immutable implementation. Rather, it refers to a closely related set of diffusion-based denoising formulations in seismic processing, all grounded in DDPM-style forward noising, learned reverse denoising, and U-Net backbones. The most coherent line of usage runs from the 2022 seismic denoising variant in “Deep Diffusion Models for Seismic Processing” through the 2025 VSP tube-wave attenuation framework “SeisDiff-deno,” and then to the 2025 “NoiseNoise is All You Need” formulation that rethinks denoising by modeling field noise directly [2207.10451, 2503.00637, 2509.03629]. | Source | Problem setting | Distinguishing feature | |---|---|---| | [2207.10451] | Seismic denoising on BP 1994 and Model94 patches | Standard image DDPM with time-conditional U-net | | [2503.00637] | Tube-wave attenuation in VSP and DAS-VSP shot-gathers | Conditional DDPM with contaminated gather conditioning and fast DDIM sampling | | [2509.03629] | DAS-VSP denoising trained directly on field noise | Noise-target training with denoised shot = noisy shot – predicted noise | The supplied summaries also reuse the label in descriptions of interpolation and Newtonian-noise contexts [2307.04226, 2512.13554, 2601.05489]. This suggests that “SeisDiff-denoNIA” is not a fully standardized bibliographic name across secondary summaries, whereas the denoising usage is the most consistent technical meaning within the seismic-processing material. ## 2. Seismic problem class and operational motivation The denoising problem addressed by SeisDiff-denoNIA depends on the acquisition setting. In the general seismic-denoising variant, the objective is attenuation of undesired incoherent or Gaussian-like noise on 2-D seismic images or gathers. In the VSP-specific formulation, the target is tube-wave attenuation in vertical seismic profiles, where tube waves are guided waves traveling along the fluid–solid interface of the borehole and may arrive long before reflected body waves with amplitudes that overwhelm the reflections of interest [2207.10451, 2503.00637]. The VSP motivation is operationally specific. When the well is under active production or injection, tube waves can mask shallow reflections, reduce signal-to-noise ratio, and force interpreters to use only late arrivals at large source-receiver offsets. Modern fiber-optic DAS-VSP runs in real time during production, so tube waves are particularly strong. Suppressing them without distorting genuine reflections is therefore critical for continuous reservoir monitoring, fracture characterization, and cost-effective seismic acquisition [2503.00637]. The 2025 field-noise formulation reframes the same broad denoising objective. Instead of treating clean seismic signal as the primary target distribution, it treats real field noise as the training target. The stated rationale is that real noise distributions, including tube waves, coupling noise, fading noise, and environmental noise, exhibit complex, field-specific statistics that synthetic labels cannot replicate. The corresponding denoising rule is explicit: denoised shot = noisy shot – predicted noise [2509.03629]. ## 3. Probabilistic formulation Across its denoising variants, SeisDiff-denoNIA adopts the canonical diffusion decomposition into a forward Gaussian noising process and a learned reverse denoising process. For clean seismic data \(x_0\), the forward process is written as $$ q(x_t \mid x_{t-1})=\mathcal{N}\bigl(x_t;\sqrt{1-\beta_t}\,x_{t-1},\beta_t \mathbf{I}\bigr), $$ with \(\alpha_t=1-\beta_t\) and \(\bar{\alpha}_t=\prod_{i=1}^t \alpha_i\). Marginalizing intermediate states yields $$ q(x_t\mid x_0)=\mathcal{N}\bigl(x_t;\sqrt{\bar\alpha_t}\,x_0,(1-\bar\alpha_t)\mathbf{I}\bigr), $$ so that $$ x_t=\sqrt{\bar\alpha_t}\,x_0+\sqrt{1-\bar\alpha_t}\,\varepsilon,\qquad \varepsilon\sim\mathcal N(0,\mathbf I). $$ In the VSP tube-wave framework, the reverse model is conditional: a neural network \(\varepsilon_\theta(x_t,c,t)\) predicts the noise component at timestep \(t\), conditioned on the contaminated shot-gather \(c\), and training minimizes the simple noise-prediction loss $$ \mathcal{L}_t=\mathbb{E}_{x_0,\varepsilon,c}\Bigl[\bigl\|\varepsilon-\varepsilon_\theta(\sqrt{\bar\alpha_t}x_0+\sqrt{1-\bar\alpha_t}\varepsilon,c,t)\bigr\|_2^2\Bigr]. $$ The full objective is \(\mathcal{L}=\sum_{t=1}^T \mathcal{L}_t\) [2503.00637]. The 2022 denoising variant uses the standard DDPM parameterization \(p_\theta(x_{t-1}\mid x_t)=\mathcal{N}(x_{t-1};\mu_\theta(x_t,t),\sigma_t^2 I)\), with learning formulated as expected \(L_2\) error in the predicted noise and a linear \(\beta\) schedule over \(T=2000\) steps [2207.10451]. The 2025 field-noise formulation preserves the same diffusion formalism but changes the data distribution: \(x_0\) is sampled from the field-noise distribution, and the model learns \(p_{\text{noise}}\) rather than a clean-seismic distribution [2509.03629]. Noise scheduling differentiates the variants. The 2022 formulation uses a linear schedule with \(\beta_1\approx 10^{-4}\), \(\beta_T\approx 0.02\), and \(T=2000\) [2207.10451]. The VSP tube-wave framework instead uses a cosine schedule \[ \bar\alpha_t=\frac{f(t)}{f(0)}, \qquad f(t)=\cos\!\bigl(\tfrac{t/T+s}{1+s}\,\tfrac{\pi}{2}\bigr)^2, \] with small offset \(s\approx 0.008\), explicitly to preserve more signal in early steps [2503.00637]. ## 4. Network architecture and inference mechanics The architectural core in the VSP formulation is a conditional DDPM with a 2D U-Net backbone enhanced by ResNet blocks and self-attention, together with a fast DDIM sampler at inference time. The input comprises two concatenated channels—the contaminated shot-gather \(c\) and the current noisy latent \(x_t\)—and the output is a single-channel prediction of \(\varepsilon_\theta(x_t,c,t)\). The U-Net has five resolutions in downsampling and upsampling paths; each resolution uses 2–4 ResNet blocks with GroupNorm, Swish, convolution, dropout, and convolution; self-attention layers are inserted at user-specified resolutions; and skip connections concatenate encoder feature maps to the decoder. Timestep conditioning is implemented through a sinusoidal positional encoding of \(t\), injected into each ResNet block via FiLM or addition [2503.00637]. The implementation summary for the VSP model reports the following key hyper-parameters: input/output channels \(=2/1\), base filters \(=96\) with multipliers \([1,1,2,2,4,4]\), number of ResBlocks per level \(=2\), attention at the lowest two resolutions, and dropout \(=0.1\) [2503.00637]. The earlier 2022 denoising variant uses the time-conditional U-net backbone of Ho et al. (2020): four down-sampling levels of residual blocks plus convolutions, a bottleneck with one self-attention layer, four up-sampling levels, skip connections, and sinusoidal or learned timestep embeddings. All convolutions are \(3\times 3\), and channel widths start at 128 and double on each downsample [2207.10451]. The later field-noise formulation preserves the same broad design logic but states the conditioning more explicitly: a noisy seismic shot with real field noise is concatenated or cross-injected at each U-Net level to guide the denoiser, with residual blocks and self-attention modules at selected resolutions [2509.03629]. Inference speed is a major design axis. In the VSP tube-wave model, a deterministic DDIM sampler with \(S \ll T\) is used, specifically \(S=50\) versus \(T=1000\), and \(\eta=0\) is chosen to produce identical outputs under identical conditions [2503.00637]. The field-noise formulation instead reports accelerated sampling with DPM-Solver, reducing steps from 1000 reverse steps to approximately 20–50 with negligible quality loss [2509.03629]. This suggests a progression from standard DDPM sampling, to DDIM acceleration, to ODE-based solver acceleration as the denoising line matures. ## 5. Data construction and training regimes The training regime changes substantially across the three denoising formulations, and those changes encode different assumptions about what should be learned. The 2022 variant trains on the BP 1994 dataset by extracting random \(64\times 64\) or \(128\times 128\) patches, ensuring that no patch has more than \(40\%\) zeros, and synthetically adding Gaussian noise whose variance is drawn so that the added noise energy is \(50\%\) of the patch’s energy. Training uses batch size 32, a linear \(\beta\) schedule, \(T=2000\), and 200,000 iterations, with standard per-patch zero mean/unit variance scaling and no further pre-conditioning or normalization [2207.10451]. The 2025 VSP tube-wave framework uses paired noisy-clean supervision. For synthetic VSP data, the SEAM Phase I elastic model generates noise-free shot-gathers, and linear noise is simulated and superposed as upward and downward events. For field DAS-VSP, clean “shut-in” recordings have no tube waves, and simulated linear tube-wave noise is added to form paired \((\mathrm{noisy},\mathrm{clean})\) data. Optimization uses AdamW, learning rate \(2\times 10^{-4}\), batch size 16, 20,000 total iterations, EMA rate 0.999, diffusion timesteps \(T=1000\), and DDIM sampling \(S=50\) [2503.00637]. The 2025 field-noise version eliminates clean labels. From each field DAS-VSP gather \(S_{\text{field}}(t)\), the first-arrival time \(t_{\text{first}}\) is identified, and a pure noise segment \(n_{\rm field}(t)=S_{\rm field}(t)\) for \(t