← Back

Icarus

Hyperspectral CNN for soil chemistry prediction

Use hyperspectral drones to predict soil chemistry instantly. This data determines what gets grown, contracted, and turned into industrial materials. The Icarus Model processes drone-captured spectral data to estimate soil composition and is deployed in Enotrium's pipeline to contract farms and optimize cash crop economics.

Drone + Hyperspectral Imaging (HSI) — Remote sensing & verification layer.

Eyes Down. Soil Analytes.

Autonomous hyperspectral remote sensing & AI verification layer for regenerative agriculture and national-security-grade feedstock assurance.

Core Capabilities

See field_results/ for actual drone flight results with ground truth validation (79 samples, MD + DE pilots). Includes real vs predicted scatter plots for N, SOC, moisture, and confusion matrices.

Abstract

This pipeline converts raw spectral data into actionable soil intelligence.

Instead of sampling soil manually, we:

The output is a layer over land use and input economics.

Soil chemistry determines crop yield and input costs. By predicting soil conditions at scale, we can contract farms based on expected output, optimize crop selection and downstream material properties, and control downstream material tensile strength.

System Architecture

┌──────────────────────────────┐ │ UAV Drone (HSI Sensor) │ │ 400–2500 nm Spectral Capture │ └──────────────┬───────────────┘ │ ▼ ┌──────────────────────────────┐ │ Hyperspectral Data Cube │ │ (x, y, λ) ~ 200+ bands │ └──────────────┬───────────────┘ │ ▼ ┌──────────────────────────────┐ │ 3D CNN Model │ │ Spectral-Spatial Learning │ └──────────────┬───────────────┘ │ ▼ ┌──────────────────────────────┐ │ Soil Prediction Maps │ │ - DNN: Nutrients (N, C) │ │ - Contaminants (PFAS, MPs) │ │ - Moisture / Structure │ └──────────────┬───────────────┘ │ ▼ ┌──────────────────────────────┐ │ Field-Level Decisions │ │ - Land valuation │ │ - Crop selection │ │ - Remediation targeting │ └──────────────────────────────┘

Theoretical Framework

Spectral Signatures of Soil Health Stages

Soil systems progress through distinct health stages characterized by shifting reflectance patterns across visible, near-infrared (NIR), and short-wave infrared (SWIR) bands. These transitions are particularly evident in regions affected by legacy agrochemicals, heavy metals, PFAs, glyphosate residues, and microbial imbalances.

HSI captures these stages through high-dimensional reflectance vectors, providing an indirect but reliable measure of chemical and biological activity. Research has demonstrated strong correlations between specific spectral features and soil contaminants, nutrient profiles, and ecological restoration potential.

Microplastics Detection

Microplastics (MPs, particles <5 mm) are among the fastest-growing and most insidious soil contaminants worldwide. They originate from plastic mulch degradation, biosolids application, tire wear, atmospheric deposition, and irrigation water. Once in soil they alter microbial communities, reduce water retention, impair root growth, and enter the food chain, directly threatening regenerative agriculture and long-term soil health.

Our 400–2500 nm hyperspectral pipeline is exceptionally well-suited for non-invasive MP detection from UAV platforms. While visible and near-infrared (VNIR, 400–1000 nm) can only provide limited discrimination, the short-wave infrared (SWIR, 1000–2500 nm) region captures the diagnostic overtone and combination absorption bands of synthetic polymers that are absent in natural soil matrices.

Mathematical Framework

The Hyperspectral Pixel

A single pixel in a hyperspectral image is not merely RGB values—it is a complete spectrum spanning hundreds of contiguous wavelength bands.

$$\text{Pixel}_{\text{HSI}} = [R_{400}, R_{405}, R_{410}, \dots, R_{2500}]$$

Where $R_\lambda$ denotes reflectance at wavelength $\lambda$ (in nanometers). Our sensors capture 210 bands at 10 nm intervals, yielding a 210-dimensional vector per pixel.

$$\text{Pixel}_{\text{HSI}} = [0.14, 0.15, 0.05, 0.11, 0.18, 0.22, \dots, 0.10]$$

3D Convolution Kernels

Traditional 2D convolutions process spatial neighborhoods only. Icarus employs 3D kernels that simultaneously convolve across spatial dimensions (height, width) and the spectral dimension. A $3 \times 3 \times 200$ kernel captures local spatial texture while integrating 200 contiguous spectral bands.

Kernel visualization ($3 \times 3$ spatial slice at band $b$):

$$K_{:, :, b} = \begin{bmatrix} w_{0,0,b} & w_{0,1,b} & w_{0,2,b} \\ w_{1,0,b} & w_{1,1,b} & w_{1,2,b} \\ w_{2,0,b} & w_{2,1,b} & w_{2,2,b} \end{bmatrix}$$

Flattened kernel vector:

$$\mathbf{w} = [w_0, w_1, w_2, \dots, w_{1799}]^\top$$

Dimensions: $3 \times 3 \times 200 = 1800$ weights per kernel.

Convolution Operation

At each position $(x, y, b)$ in the input cube, we extract a 3D patch and compute the dot product with our learned kernel:

$$z_{x,y,b} = \sum_{i=0}^{2} \sum_{j=0}^{2} \sum_{k=0}^{199} w_{i,j,k} \cdot \mathbf{x}_{x+i, y+j, b+k} + b$$

Feature Map Generation

$$\mathbf{a}^{(l)} = \text{ReLU}\left(\text{BN}\left(\mathbf{W}^{(l)} * \mathbf{a}^{(l-1)} + \mathbf{b}^{(l)}\right)\right)$$

Here $*$ denotes 3D convolution, BN is batch normalization, and $\mathbf{a}^{(l)}$ is the activation of layer $l$.

Training Objective

$$C = \sum_{j=0}^{n_{L-1}} (a_j^{(L)} - y_j)^2$$

Mean squared error between predicted soil chemistry values and laboratory-validated ground truth from portable X-ray fluorescence (pXRF) spectroscopy.

Back-Propagation

$$\frac{\partial C}{\partial w_{i,j,k}^{(l)}} = \frac{\partial C}{\partial z^{(l)}} \cdot \frac{\partial z^{(l)}}{\partial w_{i,j,k}^{(l)}} = \delta^{(l)} \cdot a^{(l-1)}_{i,j,k}$$
$$\frac{\partial C}{\partial b^{(l)}} = \delta^{(l)} = \frac{\partial C}{\partial z^{(l)}}$$
$$\delta^{(l-1)} = \left(\mathbf{W}^{(l)}\right)^\top \delta^{(l)} \odot \sigma'(z^{(l-1)})$$

Spectral Indices

NDVI — Normalized Difference Vegetation Index

Indexes density and health of vegetation. Healthy foliage strongly absorbs red light.

$$\text{NDVI} = \frac{R_{\text{NIR}} - R_{\text{Red}}}{R_{\text{NIR}} + R_{\text{Red}}}$$

NDWI — Normalized Difference Water Index

$$\text{NDWI} = \frac{R_{860} - R_{1240}}{R_{860} + R_{1240}}$$

Higher NDWI values indicate greater surface water content. It also serves as a proxy for electrical conductivity, contaminant transport potential, and redox-sensitive metal mobility.

Microplastics Detection

Our 400–2500 nm hyperspectral pipeline detects microplastics via SWIR absorption features. Common polymers exhibit distinct C–H, C–O, and C=O vibrational features:

Polyethylene (PE)

$$\lambda_{\text{PE}} \approx 1210\,\text{nm} \quad (\text{C-H 2nd overtone}), \quad 1725{-}1760\,\text{nm} \quad (\text{C-H 1st overtone})$$

Polypropylene (PP)

$$\lambda_{\text{PP}} \approx 1155{-}1200\,\text{nm}, \quad 1700{-}1735\,\text{nm}$$

Real-Field Performance

From actual drone flights with pXRF + lab chemistry ground truth (79 samples, MD + DE pilots):

Target R² RMSE Sites
Nitrogen 0.70 0.29 % MD + DE
SOC 0.76 0.39 % MD + DE
Moisture 0.88 3.5 % MD + DE
Microplastics ~0.50 F1 — DE

Preprocessing Pipeline

$$\text{Preprocess}(I) = N\left(R\left(S(I)\right)\right)$$

where $S$ performs dimension validation, $R$ applies spatial resizing, and $N$ implements spectral normalization:

$$I_{\text{norm}}(b) = \frac{I(b) - \mu_b}{\sigma_b + \epsilon}$$

Convolution Operation

Each pixel in the HSI cube is treated as a high-dimensional reflectance vector $\mathbf{r} \in \mathbb{R}^{B}$ (typically $B \approx 200$ bands). A 3D kernel $K$ of size $3 \times 3 \times B$ is convolved across the cube:

$$f(i,j) = \sum_{m=-1}^{1} \sum_{n=-1}^{1} \sum_{b=1}^{B} I(i+m, j+n, b) \cdot K(m, n, b)$$

Spatial Dropout

For a feature map tensor $F \in \mathbb{R}^{h \times w \times C}$:

$$F_{\text{drop}}(i,j,c) = F(i,j,c) \cdot M_c \quad \text{where} \quad M_c \sim \text{Bernoulli}(p)$$

Summary and Open Request

UAV HSI data from varying flight conditions is volatile to work with, and reliance on multi-source datasets without perfectly standardized calibration certainly doesn't help. Nonetheless the strong empirical performance in simulation suggests a strong correlation worth scaling, indicating that integrating hyperspectral reflectance data into EnotriumAI provides essential insights into soil biology and chemistry for regenerative economics.

Future work should expand beyond single-drone HSI to incorporate the full spectrum of ground-truth signals identified in the whitepaper (portable XRF, soil microbiome sequencing, IoT sensor arrays). This multi-modal approach, combined with the CNN backbone and zero-knowledge proofs, could enable Enotrium's platform to deliver privacy-preserving, verifiable land valuation and supply-chain transparency at continental scale — turning degraded soil into a self-generating, data-rich economic asset.

References and Results