Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/maxiricalde/ProfeLedesma/llms.txt

Use this file to discover all available pages before exploring further.

After removing duplicates, apply rigorous physics-based filters. Unlike simple range checks, these filters use the Solar Zenith Angle (SZA) and the theoretical top-of-atmosphere irradiance (TOA) — computed for each timestamp — as the reference frame for what is physically possible.

​
Computing solar geometry with the Geo helper

Before running QC, you need to enrich the DataFrame with solar geometry columns. The Geo class computes these from the site’s geographic coordinates and the timestamps: 
from helpers.Geo import Geo

dfGeo = Geo(
    range_dates=df.datetime,
    lat=site.lat,
    long=site.long,
    alt=site.alt,
    gmt=0,
    beta=0
).df

df['SZA'] = dfGeo.SZA.values   # Solar Zenith Angle in degrees
df['TOA'] = dfGeo.TOA.values   # Top-of-atmosphere irradiance W/m²
df['CTZ'] = dfGeo.CTZ.values   # cos(SZA)
df['TZ']  = dfGeo.TZ.values    # SZA in radians

​
The QC filters

The QC class (helpers/QualityControl.py) applies three independent, physically-motivated filters:
FilterConditionRule
filtro1SZA < 90° (daytime)GHI < 1.5 × 1361.7 × CTZ^1.2 + 100 (upper physical bound)
filtro2SZA > 90° (nighttime)GHI > (6.5331 − 0.065502×TZ + 1.8312e-4×TZ²) / (1 + 0.01113×TZ) (nighttime minimum)
filtro3All timesClearness index kt = GHI/TOA must be in (0, 1.4)
AceptedAll timesTrue only when all three filters pass

​
Applying QC and masking rejected values

from helpers.QualityControl import QC

QC(df)  # adds filtro1, filtro2, filtro3, kt, Acepted columns in-place

import numpy as np
df['ghi'] = np.where(df.Acepted, df.ghi, np.nan)
Readings that fail any filter are replaced with NaN, preserving the timestamp index for later resampling.
QC modifies df in place, adding five new columns: filtro1, filtro2, filtro3, kt, and Acepted.
The QC filters require the SZA, TOA, CTZ, and TZ columns — computed by the Geo helper — to already be present in the DataFrame before calling QC(df).

Build docs developers (and LLMs) love

Get started for freeTalk to us