Pemantauan Gas Rumah Kaca (GRK) di area agroindustri memerlukan data konsentrasi, waktu, dan lokasi yang dapat diakses secara bersamaan. Penelitian ini mengembangkan IoT-GIS melalui metode rancang bangun dan evaluasi eksperimental pada arsitektur tiga lapis. Node ESP32 mengakuisisi CO2 menggunakan MH-Z19B, CH4 menggunakan DFRobot Gravity MQ4, N2O menggunakan sensor NDIR IRS4-N2O, suhu, dan koordinat GPS NEO-6M. Data dikirim dalam format JSON melalui HTTP ke server MySQL dan divisualisasikan dengan Leaflet. Backend menyimpan data mentah dan menghitung secara dinamis indeks konsentrasi berbobot Global Warming Potential (IK-GWP), sehingga faktor pengali dapat diperbarui tanpa mengubah data historis. Evaluasi meliputi kalibrasi CO2 lima titik terhadap portable gas analyzer, pengiriman 300 paket pada tiga zona, serta pengujian logging dan GIS. Rata-rata galat CO2 sebesar 1,24%, sedangkan delay dan packet loss rata-rata masing-masing 231,6 ms dan 2,33%. Sistem juga berhasil memetakan node dan merekam lebih dari 1.400 data telemetri. Kontribusi penelitian terletak pada integrasi end-to-end multi-gas-GPS-GIS, kalkulasi indeks dinamis pada lapis aplikasi, dan evaluasi lintas lapis. IK-GWP digunakan sebagai indikator operasional dan tidak menggantikan inventaris emisi berbasis massa atau laju emisi. Abstract Greenhouse-gas (GHG) monitoring in agro-industrial areas requires concentration, time, and location data to be accessed simultaneously. This study develops an IoT-GIS using an engineering design and experimental evaluation method with a three-layer architecture. An ESP32 node acquires CO2 using an MH-Z19B, CH4 using a DFRobot Gravity MQ4, N2O using an IRS4-N2O NDIR sensor, temperature, and NEO-6M GPS coordinates. Data are transmitted as JSON over HTTP to a MySQL server and visualized using Leaflet. The backend stores raw data and dynamically calculates a Global Warming Potential-weighted concentration index (GWP-WCI), allowing weighting factors to be updated without modifying historical records. Evaluation covered five-point CO2 calibration against a portable gas analyzer, transmission of 300 packets across three zones, and logging and GIS functionality. The mean CO2 error was 1.24%, while the mean delay and packet loss were 231.6 ms and 2.33%, respectively. The system also mapped monitoring nodes and recorded more than 1,400 telemetry entries. The study contributes to an end-to-end multi-gas-GPS-GIS pipeline, dynamic application-layer index calculation, and cross-layer evaluation. The GWP-WCI is an operational indicator and is not a substitute for a mass- or flux-based emissions inventory.